File: | dev/pci/drm/amd/display/dc/dml/dcn30/display_mode_vba_30.c |
Warning: | line 6482, column 3 Value stored to 'surface_height_ub_l' is never read |
Press '?' to see keyboard shortcuts
Keyboard shortcuts:
1 | /* |
2 | * Copyright 2020 Advanced Micro Devices, Inc. |
3 | * |
4 | * Permission is hereby granted, free of charge, to any person obtaining a |
5 | * copy of this software and associated documentation files (the "Software"), |
6 | * to deal in the Software without restriction, including without limitation |
7 | * the rights to use, copy, modify, merge, publish, distribute, sublicense, |
8 | * and/or sell copies of the Software, and to permit persons to whom the |
9 | * Software is furnished to do so, subject to the following conditions: |
10 | * |
11 | * The above copyright notice and this permission notice shall be included in |
12 | * all copies or substantial portions of the Software. |
13 | * |
14 | * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR |
15 | * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, |
16 | * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL |
17 | * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR |
18 | * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, |
19 | * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR |
20 | * OTHER DEALINGS IN THE SOFTWARE. |
21 | * |
22 | * Authors: AMD |
23 | * |
24 | */ |
25 | |
26 | #ifdef CONFIG_DRM_AMD_DC_DCN3_01 |
27 | #include "dc.h" |
28 | #include "dc_link.h" |
29 | #include "../display_mode_lib.h" |
30 | #include "display_mode_vba_30.h" |
31 | #include "../dml_inline_defs.h" |
32 | |
33 | |
34 | /* |
35 | * NOTE: |
36 | * This file is gcc-parsable HW gospel, coming straight from HW engineers. |
37 | * |
38 | * It doesn't adhere to Linux kernel style and sometimes will do things in odd |
39 | * ways. Unless there is something clearly wrong with it the code should |
40 | * remain as-is as it provides us with a guarantee from HW that it is correct. |
41 | */ |
42 | |
43 | |
44 | typedef struct { |
45 | double DPPCLK; |
46 | double DISPCLK; |
47 | double PixelClock; |
48 | double DCFCLKDeepSleep; |
49 | unsigned int DPPPerPlane; |
50 | bool_Bool ScalerEnabled; |
51 | enum scan_direction_class SourceScan; |
52 | unsigned int BlockWidth256BytesY; |
53 | unsigned int BlockHeight256BytesY; |
54 | unsigned int BlockWidth256BytesC; |
55 | unsigned int BlockHeight256BytesC; |
56 | unsigned int InterlaceEnable; |
57 | unsigned int NumberOfCursors; |
58 | unsigned int VBlank; |
59 | unsigned int HTotal; |
60 | unsigned int DCCEnable; |
61 | bool_Bool ODMCombineEnabled; |
62 | } Pipe; |
63 | |
64 | #define BPP_INVALID0 0 |
65 | #define BPP_BLENDED_PIPE0xffffffff 0xffffffff |
66 | #define DCN30_MAX_DSC_IMAGE_WIDTH5184 5184 |
67 | #define DCN30_MAX_FMT_420_BUFFER_WIDTH4096 4096 |
68 | |
69 | static void DisplayPipeConfiguration(struct display_mode_lib *mode_lib); |
70 | static void DISPCLKDPPCLKDCFCLKDeepSleepPrefetchParametersWatermarksAndPerformanceCalculation( |
71 | struct display_mode_lib *mode_lib); |
72 | static unsigned int dscceComputeDelay( |
73 | unsigned int bpc, |
74 | double BPP, |
75 | unsigned int sliceWidth, |
76 | unsigned int numSlices, |
77 | enum output_format_class pixelFormat, |
78 | enum output_encoder_class Output); |
79 | static unsigned int dscComputeDelay( |
80 | enum output_format_class pixelFormat, |
81 | enum output_encoder_class Output); |
82 | // Super monster function with some 45 argument |
83 | static bool_Bool CalculatePrefetchSchedule( |
84 | struct display_mode_lib *mode_lib, |
85 | double PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData, |
86 | double PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly, |
87 | Pipe *myPipe, |
88 | unsigned int DSCDelay, |
89 | double DPPCLKDelaySubtotalPlusCNVCFormater, |
90 | double DPPCLKDelaySCL, |
91 | double DPPCLKDelaySCLLBOnly, |
92 | double DPPCLKDelayCNVCCursor, |
93 | double DISPCLKDelaySubtotal, |
94 | unsigned int DPP_RECOUT_WIDTH, |
95 | enum output_format_class OutputFormat, |
96 | unsigned int MaxInterDCNTileRepeaters, |
97 | unsigned int VStartup, |
98 | unsigned int MaxVStartup, |
99 | unsigned int GPUVMPageTableLevels, |
100 | bool_Bool GPUVMEnable, |
101 | bool_Bool HostVMEnable, |
102 | unsigned int HostVMMaxNonCachedPageTableLevels, |
103 | double HostVMMinPageSize, |
104 | bool_Bool DynamicMetadataEnable, |
105 | bool_Bool DynamicMetadataVMEnabled, |
106 | int DynamicMetadataLinesBeforeActiveRequired, |
107 | unsigned int DynamicMetadataTransmittedBytes, |
108 | double UrgentLatency, |
109 | double UrgentExtraLatency, |
110 | double TCalc, |
111 | unsigned int PDEAndMetaPTEBytesFrame, |
112 | unsigned int MetaRowByte, |
113 | unsigned int PixelPTEBytesPerRow, |
114 | double PrefetchSourceLinesY, |
115 | unsigned int SwathWidthY, |
116 | int BytePerPixelY, |
117 | double VInitPreFillY, |
118 | unsigned int MaxNumSwathY, |
119 | double PrefetchSourceLinesC, |
120 | unsigned int SwathWidthC, |
121 | int BytePerPixelC, |
122 | double VInitPreFillC, |
123 | unsigned int MaxNumSwathC, |
124 | long swath_width_luma_ub, |
125 | long swath_width_chroma_ub, |
126 | unsigned int SwathHeightY, |
127 | unsigned int SwathHeightC, |
128 | double TWait, |
129 | bool_Bool ProgressiveToInterlaceUnitInOPP, |
130 | double *DSTXAfterScaler, |
131 | double *DSTYAfterScaler, |
132 | double *DestinationLinesForPrefetch, |
133 | double *PrefetchBandwidth, |
134 | double *DestinationLinesToRequestVMInVBlank, |
135 | double *DestinationLinesToRequestRowInVBlank, |
136 | double *VRatioPrefetchY, |
137 | double *VRatioPrefetchC, |
138 | double *RequiredPrefetchPixDataBWLuma, |
139 | double *RequiredPrefetchPixDataBWChroma, |
140 | bool_Bool *NotEnoughTimeForDynamicMetadata, |
141 | double *Tno_bw, |
142 | double *prefetch_vmrow_bw, |
143 | double *Tdmdl_vm, |
144 | double *Tdmdl, |
145 | unsigned int *VUpdateOffsetPix, |
146 | double *VUpdateWidthPix, |
147 | double *VReadyOffsetPix); |
148 | static double RoundToDFSGranularityUp(double Clock, double VCOSpeed); |
149 | static double RoundToDFSGranularityDown(double Clock, double VCOSpeed); |
150 | static void CalculateDCCConfiguration( |
151 | bool_Bool DCCEnabled, |
152 | bool_Bool DCCProgrammingAssumesScanDirectionUnknown, |
153 | enum source_format_class SourcePixelFormat, |
154 | unsigned int ViewportWidthLuma, |
155 | unsigned int ViewportWidthChroma, |
156 | unsigned int ViewportHeightLuma, |
157 | unsigned int ViewportHeightChroma, |
158 | double DETBufferSize, |
159 | unsigned int RequestHeight256ByteLuma, |
160 | unsigned int RequestHeight256ByteChroma, |
161 | enum dm_swizzle_mode TilingFormat, |
162 | unsigned int BytePerPixelY, |
163 | unsigned int BytePerPixelC, |
164 | double BytePerPixelDETY, |
165 | double BytePerPixelDETC, |
166 | enum scan_direction_class ScanOrientation, |
167 | unsigned int *MaxUncompressedBlockLuma, |
168 | unsigned int *MaxUncompressedBlockChroma, |
169 | unsigned int *MaxCompressedBlockLuma, |
170 | unsigned int *MaxCompressedBlockChroma, |
171 | unsigned int *IndependentBlockLuma, |
172 | unsigned int *IndependentBlockChroma); |
173 | static double CalculatePrefetchSourceLines( |
174 | struct display_mode_lib *mode_lib, |
175 | double VRatio, |
176 | double vtaps, |
177 | bool_Bool Interlace, |
178 | bool_Bool ProgressiveToInterlaceUnitInOPP, |
179 | unsigned int SwathHeight, |
180 | unsigned int ViewportYStart, |
181 | double *VInitPreFill, |
182 | unsigned int *MaxNumSwath); |
183 | static unsigned int CalculateVMAndRowBytes( |
184 | struct display_mode_lib *mode_lib, |
185 | bool_Bool DCCEnable, |
186 | unsigned int BlockHeight256Bytes, |
187 | unsigned int BlockWidth256Bytes, |
188 | enum source_format_class SourcePixelFormat, |
189 | unsigned int SurfaceTiling, |
190 | unsigned int BytePerPixel, |
191 | enum scan_direction_class ScanDirection, |
192 | unsigned int SwathWidth, |
193 | unsigned int ViewportHeight, |
194 | bool_Bool GPUVMEnable, |
195 | bool_Bool HostVMEnable, |
196 | unsigned int HostVMMaxNonCachedPageTableLevels, |
197 | unsigned int GPUVMMinPageSize, |
198 | unsigned int HostVMMinPageSize, |
199 | unsigned int PTEBufferSizeInRequests, |
200 | unsigned int Pitch, |
201 | unsigned int DCCMetaPitch, |
202 | unsigned int *MacroTileWidth, |
203 | unsigned int *MetaRowByte, |
204 | unsigned int *PixelPTEBytesPerRow, |
205 | bool_Bool *PTEBufferSizeNotExceeded, |
206 | unsigned int *dpte_row_width_ub, |
207 | unsigned int *dpte_row_height, |
208 | unsigned int *MetaRequestWidth, |
209 | unsigned int *MetaRequestHeight, |
210 | unsigned int *meta_row_width, |
211 | unsigned int *meta_row_height, |
212 | unsigned int *vm_group_bytes, |
213 | unsigned int *dpte_group_bytes, |
214 | unsigned int *PixelPTEReqWidth, |
215 | unsigned int *PixelPTEReqHeight, |
216 | unsigned int *PTERequestSize, |
217 | unsigned int *DPDE0BytesFrame, |
218 | unsigned int *MetaPTEBytesFrame); |
219 | static double CalculateTWait( |
220 | unsigned int PrefetchMode, |
221 | double DRAMClockChangeLatency, |
222 | double UrgentLatency, |
223 | double SREnterPlusExitTime); |
224 | static void CalculateRowBandwidth( |
225 | bool_Bool GPUVMEnable, |
226 | enum source_format_class SourcePixelFormat, |
227 | double VRatio, |
228 | double VRatioChroma, |
229 | bool_Bool DCCEnable, |
230 | double LineTime, |
231 | unsigned int MetaRowByteLuma, |
232 | unsigned int MetaRowByteChroma, |
233 | unsigned int meta_row_height_luma, |
234 | unsigned int meta_row_height_chroma, |
235 | unsigned int PixelPTEBytesPerRowLuma, |
236 | unsigned int PixelPTEBytesPerRowChroma, |
237 | unsigned int dpte_row_height_luma, |
238 | unsigned int dpte_row_height_chroma, |
239 | double *meta_row_bw, |
240 | double *dpte_row_bw); |
241 | static void CalculateFlipSchedule( |
242 | struct display_mode_lib *mode_lib, |
243 | double PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData, |
244 | double PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly, |
245 | double UrgentExtraLatency, |
246 | double UrgentLatency, |
247 | unsigned int GPUVMMaxPageTableLevels, |
248 | bool_Bool HostVMEnable, |
249 | unsigned int HostVMMaxNonCachedPageTableLevels, |
250 | bool_Bool GPUVMEnable, |
251 | double HostVMMinPageSize, |
252 | double PDEAndMetaPTEBytesPerFrame, |
253 | double MetaRowBytes, |
254 | double DPTEBytesPerRow, |
255 | double BandwidthAvailableForImmediateFlip, |
256 | unsigned int TotImmediateFlipBytes, |
257 | enum source_format_class SourcePixelFormat, |
258 | double LineTime, |
259 | double VRatio, |
260 | double VRatioChroma, |
261 | double Tno_bw, |
262 | bool_Bool DCCEnable, |
263 | unsigned int dpte_row_height, |
264 | unsigned int meta_row_height, |
265 | unsigned int dpte_row_height_chroma, |
266 | unsigned int meta_row_height_chroma, |
267 | double *DestinationLinesToRequestVMInImmediateFlip, |
268 | double *DestinationLinesToRequestRowInImmediateFlip, |
269 | double *final_flip_bw, |
270 | bool_Bool *ImmediateFlipSupportedForPipe); |
271 | static double CalculateWriteBackDelay( |
272 | enum source_format_class WritebackPixelFormat, |
273 | double WritebackHRatio, |
274 | double WritebackVRatio, |
275 | unsigned int WritebackVTaps, |
276 | long WritebackDestinationWidth, |
277 | long WritebackDestinationHeight, |
278 | long WritebackSourceHeight, |
279 | unsigned int HTotal); |
280 | static void CalculateDynamicMetadataParameters( |
281 | int MaxInterDCNTileRepeaters, |
282 | double DPPCLK, |
283 | double DISPCLK, |
284 | double DCFClkDeepSleep, |
285 | double PixelClock, |
286 | long HTotal, |
287 | long VBlank, |
288 | long DynamicMetadataTransmittedBytes, |
289 | long DynamicMetadataLinesBeforeActiveRequired, |
290 | int InterlaceEnable, |
291 | bool_Bool ProgressiveToInterlaceUnitInOPP, |
292 | double *Tsetup, |
293 | double *Tdmbf, |
294 | double *Tdmec, |
295 | double *Tdmsks); |
296 | static void CalculateWatermarksAndDRAMSpeedChangeSupport( |
297 | struct display_mode_lib *mode_lib, |
298 | unsigned int PrefetchMode, |
299 | unsigned int NumberOfActivePlanes, |
300 | unsigned int MaxLineBufferLines, |
301 | unsigned int LineBufferSize, |
302 | unsigned int DPPOutputBufferPixels, |
303 | double DETBufferSizeInKByte, |
304 | unsigned int WritebackInterfaceBufferSize, |
305 | double DCFCLK, |
306 | double ReturnBW, |
307 | bool_Bool GPUVMEnable, |
308 | unsigned int dpte_group_bytes[], |
309 | unsigned int MetaChunkSize, |
310 | double UrgentLatency, |
311 | double ExtraLatency, |
312 | double WritebackLatency, |
313 | double WritebackChunkSize, |
314 | double SOCCLK, |
315 | double DRAMClockChangeLatency, |
316 | double SRExitTime, |
317 | double SREnterPlusExitTime, |
318 | double DCFCLKDeepSleep, |
319 | unsigned int DPPPerPlane[], |
320 | bool_Bool DCCEnable[], |
321 | double DPPCLK[], |
322 | double DETBufferSizeY[], |
323 | double DETBufferSizeC[], |
324 | unsigned int SwathHeightY[], |
325 | unsigned int SwathHeightC[], |
326 | unsigned int LBBitPerPixel[], |
327 | double SwathWidthY[], |
328 | double SwathWidthC[], |
329 | double HRatio[], |
330 | double HRatioChroma[], |
331 | unsigned int vtaps[], |
332 | unsigned int VTAPsChroma[], |
333 | double VRatio[], |
334 | double VRatioChroma[], |
335 | unsigned int HTotal[], |
336 | double PixelClock[], |
337 | unsigned int BlendingAndTiming[], |
338 | double BytePerPixelDETY[], |
339 | double BytePerPixelDETC[], |
340 | double DSTXAfterScaler[], |
341 | double DSTYAfterScaler[], |
342 | bool_Bool WritebackEnable[], |
343 | enum source_format_class WritebackPixelFormat[], |
344 | double WritebackDestinationWidth[], |
345 | double WritebackDestinationHeight[], |
346 | double WritebackSourceHeight[], |
347 | enum clock_change_support *DRAMClockChangeSupport, |
348 | double *UrgentWatermark, |
349 | double *WritebackUrgentWatermark, |
350 | double *DRAMClockChangeWatermark, |
351 | double *WritebackDRAMClockChangeWatermark, |
352 | double *StutterExitWatermark, |
353 | double *StutterEnterPlusExitWatermark, |
354 | double *MinActiveDRAMClockChangeLatencySupported); |
355 | static void CalculateDCFCLKDeepSleep( |
356 | struct display_mode_lib *mode_lib, |
357 | unsigned int NumberOfActivePlanes, |
358 | int BytePerPixelY[], |
359 | int BytePerPixelC[], |
360 | double VRatio[], |
361 | double VRatioChroma[], |
362 | double SwathWidthY[], |
363 | double SwathWidthC[], |
364 | unsigned int DPPPerPlane[], |
365 | double HRatio[], |
366 | double HRatioChroma[], |
367 | double PixelClock[], |
368 | double PSCL_THROUGHPUT[], |
369 | double PSCL_THROUGHPUT_CHROMA[], |
370 | double DPPCLK[], |
371 | double ReadBandwidthLuma[], |
372 | double ReadBandwidthChroma[], |
373 | int ReturnBusWidth, |
374 | double *DCFCLKDeepSleep); |
375 | static void CalculateUrgentBurstFactor( |
376 | long swath_width_luma_ub, |
377 | long swath_width_chroma_ub, |
378 | unsigned int DETBufferSizeInKByte, |
379 | unsigned int SwathHeightY, |
380 | unsigned int SwathHeightC, |
381 | double LineTime, |
382 | double UrgentLatency, |
383 | double CursorBufferSize, |
384 | unsigned int CursorWidth, |
385 | unsigned int CursorBPP, |
386 | double VRatio, |
387 | double VRatioC, |
388 | double BytePerPixelInDETY, |
389 | double BytePerPixelInDETC, |
390 | double DETBufferSizeY, |
391 | double DETBufferSizeC, |
392 | double *UrgentBurstFactorCursor, |
393 | double *UrgentBurstFactorLuma, |
394 | double *UrgentBurstFactorChroma, |
395 | bool_Bool *NotEnoughUrgentLatencyHiding); |
396 | |
397 | static void UseMinimumDCFCLK( |
398 | struct display_mode_lib *mode_lib, |
399 | int MaxInterDCNTileRepeaters, |
400 | int MaxPrefetchMode, |
401 | double FinalDRAMClockChangeLatency, |
402 | double SREnterPlusExitTime, |
403 | int ReturnBusWidth, |
404 | int RoundTripPingLatencyCycles, |
405 | int ReorderingBytes, |
406 | int PixelChunkSizeInKByte, |
407 | int MetaChunkSize, |
408 | bool_Bool GPUVMEnable, |
409 | int GPUVMMaxPageTableLevels, |
410 | bool_Bool HostVMEnable, |
411 | int NumberOfActivePlanes, |
412 | double HostVMMinPageSize, |
413 | int HostVMMaxNonCachedPageTableLevels, |
414 | bool_Bool DynamicMetadataVMEnabled, |
415 | enum immediate_flip_requirement ImmediateFlipRequirement, |
416 | bool_Bool ProgressiveToInterlaceUnitInOPP, |
417 | double MaxAveragePercentOfIdealSDPPortBWDisplayCanUseInNormalSystemOperation, |
418 | double PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData, |
419 | double PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly, |
420 | double PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelDataOnly, |
421 | int VTotal[], |
422 | int VActive[], |
423 | int DynamicMetadataTransmittedBytes[], |
424 | int DynamicMetadataLinesBeforeActiveRequired[], |
425 | bool_Bool Interlace[], |
426 | double RequiredDPPCLK[][2][DC__NUM_DPP__MAX8], |
427 | double RequiredDISPCLK[][2], |
428 | double UrgLatency[], |
429 | unsigned int NoOfDPP[][2][DC__NUM_DPP__MAX8], |
430 | double ProjectedDCFCLKDeepSleep[][2], |
431 | double MaximumVStartup[][2][DC__NUM_DPP__MAX8], |
432 | double TotalVActivePixelBandwidth[][2], |
433 | double TotalVActiveCursorBandwidth[][2], |
434 | double TotalMetaRowBandwidth[][2], |
435 | double TotalDPTERowBandwidth[][2], |
436 | unsigned int TotalNumberOfActiveDPP[][2], |
437 | unsigned int TotalNumberOfDCCActiveDPP[][2], |
438 | int dpte_group_bytes[], |
439 | double PrefetchLinesY[][2][DC__NUM_DPP__MAX8], |
440 | double PrefetchLinesC[][2][DC__NUM_DPP__MAX8], |
441 | int swath_width_luma_ub_all_states[][2][DC__NUM_DPP__MAX8], |
442 | int swath_width_chroma_ub_all_states[][2][DC__NUM_DPP__MAX8], |
443 | int BytePerPixelY[], |
444 | int BytePerPixelC[], |
445 | int HTotal[], |
446 | double PixelClock[], |
447 | double PDEAndMetaPTEBytesPerFrame[][2][DC__NUM_DPP__MAX8], |
448 | double DPTEBytesPerRow[][2][DC__NUM_DPP__MAX8], |
449 | double MetaRowBytes[][2][DC__NUM_DPP__MAX8], |
450 | bool_Bool DynamicMetadataEnable[], |
451 | double VActivePixelBandwidth[][2][DC__NUM_DPP__MAX8], |
452 | double VActiveCursorBandwidth[][2][DC__NUM_DPP__MAX8], |
453 | double ReadBandwidthLuma[], |
454 | double ReadBandwidthChroma[], |
455 | double DCFCLKPerState[], |
456 | double DCFCLKState[][2]); |
457 | static void CalculatePixelDeliveryTimes( |
458 | unsigned int NumberOfActivePlanes, |
459 | double VRatio[], |
460 | double VRatioChroma[], |
461 | double VRatioPrefetchY[], |
462 | double VRatioPrefetchC[], |
463 | unsigned int swath_width_luma_ub[], |
464 | unsigned int swath_width_chroma_ub[], |
465 | unsigned int DPPPerPlane[], |
466 | double HRatio[], |
467 | double HRatioChroma[], |
468 | double PixelClock[], |
469 | double PSCL_THROUGHPUT[], |
470 | double PSCL_THROUGHPUT_CHROMA[], |
471 | double DPPCLK[], |
472 | int BytePerPixelC[], |
473 | enum scan_direction_class SourceScan[], |
474 | unsigned int NumberOfCursors[], |
475 | unsigned int CursorWidth[][2], |
476 | unsigned int CursorBPP[][2], |
477 | unsigned int BlockWidth256BytesY[], |
478 | unsigned int BlockHeight256BytesY[], |
479 | unsigned int BlockWidth256BytesC[], |
480 | unsigned int BlockHeight256BytesC[], |
481 | double DisplayPipeLineDeliveryTimeLuma[], |
482 | double DisplayPipeLineDeliveryTimeChroma[], |
483 | double DisplayPipeLineDeliveryTimeLumaPrefetch[], |
484 | double DisplayPipeLineDeliveryTimeChromaPrefetch[], |
485 | double DisplayPipeRequestDeliveryTimeLuma[], |
486 | double DisplayPipeRequestDeliveryTimeChroma[], |
487 | double DisplayPipeRequestDeliveryTimeLumaPrefetch[], |
488 | double DisplayPipeRequestDeliveryTimeChromaPrefetch[], |
489 | double CursorRequestDeliveryTime[], |
490 | double CursorRequestDeliveryTimePrefetch[]); |
491 | |
492 | static void CalculateMetaAndPTETimes( |
493 | int NumberOfActivePlanes, |
494 | bool_Bool GPUVMEnable, |
495 | int MetaChunkSize, |
496 | int MinMetaChunkSizeBytes, |
497 | int HTotal[], |
498 | double VRatio[], |
499 | double VRatioChroma[], |
500 | double DestinationLinesToRequestRowInVBlank[], |
501 | double DestinationLinesToRequestRowInImmediateFlip[], |
502 | bool_Bool DCCEnable[], |
503 | double PixelClock[], |
504 | int BytePerPixelY[], |
505 | int BytePerPixelC[], |
506 | enum scan_direction_class SourceScan[], |
507 | int dpte_row_height[], |
508 | int dpte_row_height_chroma[], |
509 | int meta_row_width[], |
510 | int meta_row_width_chroma[], |
511 | int meta_row_height[], |
512 | int meta_row_height_chroma[], |
513 | int meta_req_width[], |
514 | int meta_req_width_chroma[], |
515 | int meta_req_height[], |
516 | int meta_req_height_chroma[], |
517 | int dpte_group_bytes[], |
518 | int PTERequestSizeY[], |
519 | int PTERequestSizeC[], |
520 | int PixelPTEReqWidthY[], |
521 | int PixelPTEReqHeightY[], |
522 | int PixelPTEReqWidthC[], |
523 | int PixelPTEReqHeightC[], |
524 | int dpte_row_width_luma_ub[], |
525 | int dpte_row_width_chroma_ub[], |
526 | double DST_Y_PER_PTE_ROW_NOM_L[], |
527 | double DST_Y_PER_PTE_ROW_NOM_C[], |
528 | double DST_Y_PER_META_ROW_NOM_L[], |
529 | double DST_Y_PER_META_ROW_NOM_C[], |
530 | double TimePerMetaChunkNominal[], |
531 | double TimePerChromaMetaChunkNominal[], |
532 | double TimePerMetaChunkVBlank[], |
533 | double TimePerChromaMetaChunkVBlank[], |
534 | double TimePerMetaChunkFlip[], |
535 | double TimePerChromaMetaChunkFlip[], |
536 | double time_per_pte_group_nom_luma[], |
537 | double time_per_pte_group_vblank_luma[], |
538 | double time_per_pte_group_flip_luma[], |
539 | double time_per_pte_group_nom_chroma[], |
540 | double time_per_pte_group_vblank_chroma[], |
541 | double time_per_pte_group_flip_chroma[]); |
542 | |
543 | static void CalculateVMGroupAndRequestTimes( |
544 | unsigned int NumberOfActivePlanes, |
545 | bool_Bool GPUVMEnable, |
546 | unsigned int GPUVMMaxPageTableLevels, |
547 | unsigned int HTotal[], |
548 | int BytePerPixelC[], |
549 | double DestinationLinesToRequestVMInVBlank[], |
550 | double DestinationLinesToRequestVMInImmediateFlip[], |
551 | bool_Bool DCCEnable[], |
552 | double PixelClock[], |
553 | int dpte_row_width_luma_ub[], |
554 | int dpte_row_width_chroma_ub[], |
555 | int vm_group_bytes[], |
556 | unsigned int dpde0_bytes_per_frame_ub_l[], |
557 | unsigned int dpde0_bytes_per_frame_ub_c[], |
558 | int meta_pte_bytes_per_frame_ub_l[], |
559 | int meta_pte_bytes_per_frame_ub_c[], |
560 | double TimePerVMGroupVBlank[], |
561 | double TimePerVMGroupFlip[], |
562 | double TimePerVMRequestVBlank[], |
563 | double TimePerVMRequestFlip[]); |
564 | |
565 | static void CalculateStutterEfficiency( |
566 | int NumberOfActivePlanes, |
567 | long ROBBufferSizeInKByte, |
568 | double TotalDataReadBandwidth, |
569 | double DCFCLK, |
570 | double ReturnBW, |
571 | double SRExitTime, |
572 | bool_Bool SynchronizedVBlank, |
573 | int DPPPerPlane[], |
574 | double DETBufferSizeY[], |
575 | int BytePerPixelY[], |
576 | double BytePerPixelDETY[], |
577 | double SwathWidthY[], |
578 | int SwathHeightY[], |
579 | int SwathHeightC[], |
580 | double DCCRateLuma[], |
581 | double DCCRateChroma[], |
582 | int HTotal[], |
583 | int VTotal[], |
584 | double PixelClock[], |
585 | double VRatio[], |
586 | enum scan_direction_class SourceScan[], |
587 | int BlockHeight256BytesY[], |
588 | int BlockWidth256BytesY[], |
589 | int BlockHeight256BytesC[], |
590 | int BlockWidth256BytesC[], |
591 | int DCCYMaxUncompressedBlock[], |
592 | int DCCCMaxUncompressedBlock[], |
593 | int VActive[], |
594 | bool_Bool DCCEnable[], |
595 | bool_Bool WritebackEnable[], |
596 | double ReadBandwidthPlaneLuma[], |
597 | double ReadBandwidthPlaneChroma[], |
598 | double meta_row_bw[], |
599 | double dpte_row_bw[], |
600 | double *StutterEfficiencyNotIncludingVBlank, |
601 | double *StutterEfficiency, |
602 | double *StutterPeriodOut); |
603 | |
604 | static void CalculateSwathAndDETConfiguration( |
605 | bool_Bool ForceSingleDPP, |
606 | int NumberOfActivePlanes, |
607 | long DETBufferSizeInKByte, |
608 | double MaximumSwathWidthLuma[], |
609 | double MaximumSwathWidthChroma[], |
610 | enum scan_direction_class SourceScan[], |
611 | enum source_format_class SourcePixelFormat[], |
612 | enum dm_swizzle_mode SurfaceTiling[], |
613 | int ViewportWidth[], |
614 | int ViewportHeight[], |
615 | int SurfaceWidthY[], |
616 | int SurfaceWidthC[], |
617 | int SurfaceHeightY[], |
618 | int SurfaceHeightC[], |
619 | int Read256BytesBlockHeightY[], |
620 | int Read256BytesBlockHeightC[], |
621 | int Read256BytesBlockWidthY[], |
622 | int Read256BytesBlockWidthC[], |
623 | enum odm_combine_mode ODMCombineEnabled[], |
624 | int BlendingAndTiming[], |
625 | int BytePerPixY[], |
626 | int BytePerPixC[], |
627 | double BytePerPixDETY[], |
628 | double BytePerPixDETC[], |
629 | int HActive[], |
630 | double HRatio[], |
631 | double HRatioChroma[], |
632 | int DPPPerPlane[], |
633 | int swath_width_luma_ub[], |
634 | int swath_width_chroma_ub[], |
635 | double SwathWidth[], |
636 | double SwathWidthChroma[], |
637 | int SwathHeightY[], |
638 | int SwathHeightC[], |
639 | double DETBufferSizeY[], |
640 | double DETBufferSizeC[], |
641 | bool_Bool ViewportSizeSupportPerPlane[], |
642 | bool_Bool *ViewportSizeSupport); |
643 | static void CalculateSwathWidth( |
644 | bool_Bool ForceSingleDPP, |
645 | int NumberOfActivePlanes, |
646 | enum source_format_class SourcePixelFormat[], |
647 | enum scan_direction_class SourceScan[], |
648 | unsigned int ViewportWidth[], |
649 | unsigned int ViewportHeight[], |
650 | unsigned int SurfaceWidthY[], |
651 | unsigned int SurfaceWidthC[], |
652 | unsigned int SurfaceHeightY[], |
653 | unsigned int SurfaceHeightC[], |
654 | enum odm_combine_mode ODMCombineEnabled[], |
655 | int BytePerPixY[], |
656 | int BytePerPixC[], |
657 | int Read256BytesBlockHeightY[], |
658 | int Read256BytesBlockHeightC[], |
659 | int Read256BytesBlockWidthY[], |
660 | int Read256BytesBlockWidthC[], |
661 | int BlendingAndTiming[], |
662 | unsigned int HActive[], |
663 | double HRatio[], |
664 | int DPPPerPlane[], |
665 | double SwathWidthSingleDPPY[], |
666 | double SwathWidthSingleDPPC[], |
667 | double SwathWidthY[], |
668 | double SwathWidthC[], |
669 | int MaximumSwathHeightY[], |
670 | int MaximumSwathHeightC[], |
671 | unsigned int swath_width_luma_ub[], |
672 | unsigned int swath_width_chroma_ub[]); |
673 | static double CalculateExtraLatency( |
674 | long RoundTripPingLatencyCycles, |
675 | long ReorderingBytes, |
676 | double DCFCLK, |
677 | int TotalNumberOfActiveDPP, |
678 | int PixelChunkSizeInKByte, |
679 | int TotalNumberOfDCCActiveDPP, |
680 | int MetaChunkSize, |
681 | double ReturnBW, |
682 | bool_Bool GPUVMEnable, |
683 | bool_Bool HostVMEnable, |
684 | int NumberOfActivePlanes, |
685 | int NumberOfDPP[], |
686 | int dpte_group_bytes[], |
687 | double PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData, |
688 | double PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly, |
689 | double HostVMMinPageSize, |
690 | int HostVMMaxNonCachedPageTableLevels); |
691 | static double CalculateExtraLatencyBytes( |
692 | long ReorderingBytes, |
693 | int TotalNumberOfActiveDPP, |
694 | int PixelChunkSizeInKByte, |
695 | int TotalNumberOfDCCActiveDPP, |
696 | int MetaChunkSize, |
697 | bool_Bool GPUVMEnable, |
698 | bool_Bool HostVMEnable, |
699 | int NumberOfActivePlanes, |
700 | int NumberOfDPP[], |
701 | int dpte_group_bytes[], |
702 | double PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData, |
703 | double PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly, |
704 | double HostVMMinPageSize, |
705 | int HostVMMaxNonCachedPageTableLevels); |
706 | static double CalculateUrgentLatency( |
707 | double UrgentLatencyPixelDataOnly, |
708 | double UrgentLatencyPixelMixedWithVMData, |
709 | double UrgentLatencyVMDataOnly, |
710 | bool_Bool DoUrgentLatencyAdjustment, |
711 | double UrgentLatencyAdjustmentFabricClockComponent, |
712 | double UrgentLatencyAdjustmentFabricClockReference, |
713 | double FabricClockSingle); |
714 | |
715 | static bool_Bool CalculateBytePerPixelAnd256BBlockSizes( |
716 | enum source_format_class SourcePixelFormat, |
717 | enum dm_swizzle_mode SurfaceTiling, |
718 | unsigned int *BytePerPixelY, |
719 | unsigned int *BytePerPixelC, |
720 | double *BytePerPixelDETY, |
721 | double *BytePerPixelDETC, |
722 | unsigned int *BlockHeight256BytesY, |
723 | unsigned int *BlockHeight256BytesC, |
724 | unsigned int *BlockWidth256BytesY, |
725 | unsigned int *BlockWidth256BytesC); |
726 | |
727 | void dml30_recalculate(struct display_mode_lib *mode_lib) |
728 | { |
729 | ModeSupportAndSystemConfiguration(mode_lib); |
730 | PixelClockAdjustmentForProgressiveToInterlaceUnit(mode_lib); |
731 | DisplayPipeConfiguration(mode_lib); |
732 | DISPCLKDPPCLKDCFCLKDeepSleepPrefetchParametersWatermarksAndPerformanceCalculation(mode_lib); |
733 | } |
734 | |
735 | static unsigned int dscceComputeDelay( |
736 | unsigned int bpc, |
737 | double BPP, |
738 | unsigned int sliceWidth, |
739 | unsigned int numSlices, |
740 | enum output_format_class pixelFormat, |
741 | enum output_encoder_class Output) |
742 | { |
743 | // valid bpc = source bits per component in the set of {8, 10, 12} |
744 | // valid bpp = increments of 1/16 of a bit |
745 | // min = 6/7/8 in N420/N422/444, respectively |
746 | // max = such that compression is 1:1 |
747 | //valid sliceWidth = number of pixels per slice line, must be less than or equal to 5184/numSlices (or 4096/numSlices in 420 mode) |
748 | //valid numSlices = number of slices in the horiziontal direction per DSC engine in the set of {1, 2, 3, 4} |
749 | //valid pixelFormat = pixel/color format in the set of {:N444_RGB, :S422, :N422, :N420} |
750 | |
751 | // fixed value |
752 | unsigned int rcModelSize = 8192; |
753 | |
754 | // N422/N420 operate at 2 pixels per clock |
755 | unsigned int pixelsPerClock, lstall, D, initalXmitDelay, w, s, ix, wx, P, l0, a, ax, L, |
756 | Delay, pixels; |
757 | |
758 | if (pixelFormat == dm_420) |
759 | pixelsPerClock = 2; |
760 | // #all other modes operate at 1 pixel per clock |
761 | else if (pixelFormat == dm_444) |
762 | pixelsPerClock = 1; |
763 | else if (pixelFormat == dm_n422) |
764 | pixelsPerClock = 2; |
765 | else |
766 | pixelsPerClock = 1; |
767 | |
768 | //initial transmit delay as per PPS |
769 | initalXmitDelay = dml_round(rcModelSize / 2.0 / BPP / pixelsPerClock); |
770 | |
771 | //compute ssm delay |
772 | if (bpc == 8) |
773 | D = 81; |
774 | else if (bpc == 10) |
775 | D = 89; |
776 | else |
777 | D = 113; |
778 | |
779 | //divide by pixel per cycle to compute slice width as seen by DSC |
780 | w = sliceWidth / pixelsPerClock; |
781 | |
782 | //422 mode has an additional cycle of delay |
783 | if (pixelFormat == dm_420 || pixelFormat == dm_444 || pixelFormat == dm_n422) |
784 | s = 0; |
785 | else |
786 | s = 1; |
787 | |
788 | //main calculation for the dscce |
789 | ix = initalXmitDelay + 45; |
790 | wx = (w + 2) / 3; |
791 | P = 3 * wx - w; |
792 | l0 = ix / w; |
793 | a = ix + P * l0; |
794 | ax = (a + 2) / 3 + D + 6 + 1; |
795 | L = (ax + wx - 1) / wx; |
796 | if ((ix % w) == 0 && P != 0) |
797 | lstall = 1; |
798 | else |
799 | lstall = 0; |
800 | Delay = L * wx * (numSlices - 1) + ax + s + lstall + 22; |
801 | |
802 | //dsc processes 3 pixel containers per cycle and a container can contain 1 or 2 pixels |
803 | pixels = Delay * 3 * pixelsPerClock; |
804 | return pixels; |
805 | } |
806 | |
807 | static unsigned int dscComputeDelay(enum output_format_class pixelFormat, enum output_encoder_class Output) |
808 | { |
809 | unsigned int Delay = 0; |
810 | |
811 | if (pixelFormat == dm_420) { |
812 | // sfr |
813 | Delay = Delay + 2; |
814 | // dsccif |
815 | Delay = Delay + 0; |
816 | // dscc - input deserializer |
817 | Delay = Delay + 3; |
818 | // dscc gets pixels every other cycle |
819 | Delay = Delay + 2; |
820 | // dscc - input cdc fifo |
821 | Delay = Delay + 12; |
822 | // dscc gets pixels every other cycle |
823 | Delay = Delay + 13; |
824 | // dscc - cdc uncertainty |
825 | Delay = Delay + 2; |
826 | // dscc - output cdc fifo |
827 | Delay = Delay + 7; |
828 | // dscc gets pixels every other cycle |
829 | Delay = Delay + 3; |
830 | // dscc - cdc uncertainty |
831 | Delay = Delay + 2; |
832 | // dscc - output serializer |
833 | Delay = Delay + 1; |
834 | // sft |
835 | Delay = Delay + 1; |
836 | } else if (pixelFormat == dm_n422) { |
837 | // sfr |
838 | Delay = Delay + 2; |
839 | // dsccif |
840 | Delay = Delay + 1; |
841 | // dscc - input deserializer |
842 | Delay = Delay + 5; |
843 | // dscc - input cdc fifo |
844 | Delay = Delay + 25; |
845 | // dscc - cdc uncertainty |
846 | Delay = Delay + 2; |
847 | // dscc - output cdc fifo |
848 | Delay = Delay + 10; |
849 | // dscc - cdc uncertainty |
850 | Delay = Delay + 2; |
851 | // dscc - output serializer |
852 | Delay = Delay + 1; |
853 | // sft |
854 | Delay = Delay + 1; |
855 | } |
856 | else { |
857 | // sfr |
858 | Delay = Delay + 2; |
859 | // dsccif |
860 | Delay = Delay + 0; |
861 | // dscc - input deserializer |
862 | Delay = Delay + 3; |
863 | // dscc - input cdc fifo |
864 | Delay = Delay + 12; |
865 | // dscc - cdc uncertainty |
866 | Delay = Delay + 2; |
867 | // dscc - output cdc fifo |
868 | Delay = Delay + 7; |
869 | // dscc - output serializer |
870 | Delay = Delay + 1; |
871 | // dscc - cdc uncertainty |
872 | Delay = Delay + 2; |
873 | // sft |
874 | Delay = Delay + 1; |
875 | } |
876 | |
877 | return Delay; |
878 | } |
879 | |
880 | static bool_Bool CalculatePrefetchSchedule( |
881 | struct display_mode_lib *mode_lib, |
882 | double PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData, |
883 | double PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly, |
884 | Pipe *myPipe, |
885 | unsigned int DSCDelay, |
886 | double DPPCLKDelaySubtotalPlusCNVCFormater, |
887 | double DPPCLKDelaySCL, |
888 | double DPPCLKDelaySCLLBOnly, |
889 | double DPPCLKDelayCNVCCursor, |
890 | double DISPCLKDelaySubtotal, |
891 | unsigned int DPP_RECOUT_WIDTH, |
892 | enum output_format_class OutputFormat, |
893 | unsigned int MaxInterDCNTileRepeaters, |
894 | unsigned int VStartup, |
895 | unsigned int MaxVStartup, |
896 | unsigned int GPUVMPageTableLevels, |
897 | bool_Bool GPUVMEnable, |
898 | bool_Bool HostVMEnable, |
899 | unsigned int HostVMMaxNonCachedPageTableLevels, |
900 | double HostVMMinPageSize, |
901 | bool_Bool DynamicMetadataEnable, |
902 | bool_Bool DynamicMetadataVMEnabled, |
903 | int DynamicMetadataLinesBeforeActiveRequired, |
904 | unsigned int DynamicMetadataTransmittedBytes, |
905 | double UrgentLatency, |
906 | double UrgentExtraLatency, |
907 | double TCalc, |
908 | unsigned int PDEAndMetaPTEBytesFrame, |
909 | unsigned int MetaRowByte, |
910 | unsigned int PixelPTEBytesPerRow, |
911 | double PrefetchSourceLinesY, |
912 | unsigned int SwathWidthY, |
913 | int BytePerPixelY, |
914 | double VInitPreFillY, |
915 | unsigned int MaxNumSwathY, |
916 | double PrefetchSourceLinesC, |
917 | unsigned int SwathWidthC, |
918 | int BytePerPixelC, |
919 | double VInitPreFillC, |
920 | unsigned int MaxNumSwathC, |
921 | long swath_width_luma_ub, |
922 | long swath_width_chroma_ub, |
923 | unsigned int SwathHeightY, |
924 | unsigned int SwathHeightC, |
925 | double TWait, |
926 | bool_Bool ProgressiveToInterlaceUnitInOPP, |
927 | double *DSTXAfterScaler, |
928 | double *DSTYAfterScaler, |
929 | double *DestinationLinesForPrefetch, |
930 | double *PrefetchBandwidth, |
931 | double *DestinationLinesToRequestVMInVBlank, |
932 | double *DestinationLinesToRequestRowInVBlank, |
933 | double *VRatioPrefetchY, |
934 | double *VRatioPrefetchC, |
935 | double *RequiredPrefetchPixDataBWLuma, |
936 | double *RequiredPrefetchPixDataBWChroma, |
937 | bool_Bool *NotEnoughTimeForDynamicMetadata, |
938 | double *Tno_bw, |
939 | double *prefetch_vmrow_bw, |
940 | double *Tdmdl_vm, |
941 | double *Tdmdl, |
942 | unsigned int *VUpdateOffsetPix, |
943 | double *VUpdateWidthPix, |
944 | double *VReadyOffsetPix) |
945 | { |
946 | bool_Bool MyError = false0; |
947 | unsigned int DPPCycles = 0, DISPCLKCycles = 0; |
948 | double DSTTotalPixelsAfterScaler = 0; |
949 | double LineTime = 0, Tsetup = 0; |
950 | double dst_y_prefetch_equ = 0; |
951 | double Tsw_oto = 0; |
952 | double prefetch_bw_oto = 0; |
953 | double Tvm_oto = 0; |
954 | double Tr0_oto = 0; |
955 | double Tvm_oto_lines = 0; |
956 | double Tr0_oto_lines = 0; |
957 | double dst_y_prefetch_oto = 0; |
958 | double TimeForFetchingMetaPTE = 0; |
959 | double TimeForFetchingRowInVBlank = 0; |
960 | double LinesToRequestPrefetchPixelData = 0; |
961 | double HostVMInefficiencyFactor = 0; |
962 | unsigned int HostVMDynamicLevelsTrips = 0; |
963 | double trip_to_mem = 0; |
964 | double Tvm_trips = 0; |
965 | double Tr0_trips = 0; |
966 | double Tvm_trips_rounded = 0; |
967 | double Tr0_trips_rounded = 0; |
968 | double Lsw_oto = 0; |
969 | double Tpre_rounded = 0; |
970 | double prefetch_bw_equ = 0; |
971 | double Tvm_equ = 0; |
972 | double Tr0_equ = 0; |
973 | double Tdmbf = 0; |
974 | double Tdmec = 0; |
975 | double Tdmsks = 0; |
976 | |
977 | if (GPUVMEnable == true1 && HostVMEnable == true1) { |
978 | HostVMInefficiencyFactor = PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData / PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly; |
979 | HostVMDynamicLevelsTrips = HostVMMaxNonCachedPageTableLevels; |
980 | } else { |
981 | HostVMInefficiencyFactor = 1; |
982 | HostVMDynamicLevelsTrips = 0; |
983 | } |
984 | |
985 | CalculateDynamicMetadataParameters( |
986 | MaxInterDCNTileRepeaters, |
987 | myPipe->DPPCLK, |
988 | myPipe->DISPCLK, |
989 | myPipe->DCFCLKDeepSleep, |
990 | myPipe->PixelClock, |
991 | myPipe->HTotal, |
992 | myPipe->VBlank, |
993 | DynamicMetadataTransmittedBytes, |
994 | DynamicMetadataLinesBeforeActiveRequired, |
995 | myPipe->InterlaceEnable, |
996 | ProgressiveToInterlaceUnitInOPP, |
997 | &Tsetup, |
998 | &Tdmbf, |
999 | &Tdmec, |
1000 | &Tdmsks); |
1001 | |
1002 | LineTime = myPipe->HTotal / myPipe->PixelClock; |
1003 | trip_to_mem = UrgentLatency; |
1004 | Tvm_trips = UrgentExtraLatency + trip_to_mem * (GPUVMPageTableLevels * (HostVMDynamicLevelsTrips + 1) - 1); |
1005 | |
1006 | if (DynamicMetadataVMEnabled == true1 && GPUVMEnable == true1) { |
1007 | *Tdmdl = TWait + Tvm_trips + trip_to_mem; |
1008 | } else { |
1009 | *Tdmdl = TWait + UrgentExtraLatency; |
1010 | } |
1011 | |
1012 | if (DynamicMetadataEnable == true1) { |
1013 | if (VStartup * LineTime < Tsetup + *Tdmdl + Tdmbf + Tdmec + Tdmsks) { |
1014 | *NotEnoughTimeForDynamicMetadata = true1; |
1015 | } else { |
1016 | *NotEnoughTimeForDynamicMetadata = false0; |
1017 | dml_print("DML: Not Enough Time for Dynamic Meta!\n"){do { } while(0); }; |
1018 | dml_print("DML: Tdmbf: %fus - time for dmd transfer from dchub to dio output buffer\n", Tdmbf){do { } while(0); }; |
1019 | dml_print("DML: Tdmec: %fus - time dio takes to transfer dmd\n", Tdmec){do { } while(0); }; |
1020 | dml_print("DML: Tdmsks: %fus - time before active dmd must complete transmission at dio\n", Tdmsks){do { } while(0); }; |
1021 | dml_print("DML: Tdmdl: %fus - time for fabric to become ready and fetch dmd \n", *Tdmdl){do { } while(0); }; |
1022 | } |
1023 | } else { |
1024 | *NotEnoughTimeForDynamicMetadata = false0; |
1025 | } |
1026 | |
1027 | *Tdmdl_vm = (DynamicMetadataEnable == true1 && DynamicMetadataVMEnabled == true1 && GPUVMEnable == true1 ? TWait + Tvm_trips : 0); |
1028 | |
1029 | if (myPipe->ScalerEnabled) |
1030 | DPPCycles = DPPCLKDelaySubtotalPlusCNVCFormater + DPPCLKDelaySCL; |
1031 | else |
1032 | DPPCycles = DPPCLKDelaySubtotalPlusCNVCFormater + DPPCLKDelaySCLLBOnly; |
1033 | |
1034 | DPPCycles = DPPCycles + myPipe->NumberOfCursors * DPPCLKDelayCNVCCursor; |
1035 | |
1036 | DISPCLKCycles = DISPCLKDelaySubtotal; |
1037 | |
1038 | if (myPipe->DPPCLK == 0.0 || myPipe->DISPCLK == 0.0) |
1039 | return true1; |
1040 | |
1041 | *DSTXAfterScaler = DPPCycles * myPipe->PixelClock / myPipe->DPPCLK + DISPCLKCycles * myPipe->PixelClock / myPipe->DISPCLK |
1042 | + DSCDelay; |
1043 | |
1044 | *DSTXAfterScaler = *DSTXAfterScaler + ((myPipe->ODMCombineEnabled)?18:0) + (myPipe->DPPPerPlane - 1) * DPP_RECOUT_WIDTH; |
1045 | |
1046 | if (OutputFormat == dm_420 || (myPipe->InterlaceEnable && ProgressiveToInterlaceUnitInOPP)) |
1047 | *DSTYAfterScaler = 1; |
1048 | else |
1049 | *DSTYAfterScaler = 0; |
1050 | |
1051 | DSTTotalPixelsAfterScaler = *DSTYAfterScaler * myPipe->HTotal + *DSTXAfterScaler; |
1052 | *DSTYAfterScaler = dml_floor(DSTTotalPixelsAfterScaler / myPipe->HTotal, 1); |
1053 | *DSTXAfterScaler = DSTTotalPixelsAfterScaler - ((double) (*DSTYAfterScaler * myPipe->HTotal)); |
1054 | |
1055 | MyError = false0; |
1056 | |
1057 | |
1058 | Tr0_trips = trip_to_mem * (HostVMDynamicLevelsTrips + 1); |
1059 | Tvm_trips_rounded = dml_ceil(4.0 * Tvm_trips / LineTime, 1) / 4 * LineTime; |
1060 | Tr0_trips_rounded = dml_ceil(4.0 * Tr0_trips / LineTime, 1) / 4 * LineTime; |
1061 | |
1062 | if (GPUVMEnable) { |
1063 | if (GPUVMPageTableLevels >= 3) { |
1064 | *Tno_bw = UrgentExtraLatency + trip_to_mem * ((GPUVMPageTableLevels - 2) - 1); |
1065 | } else |
1066 | *Tno_bw = 0; |
1067 | } else if (!myPipe->DCCEnable) |
1068 | *Tno_bw = LineTime; |
1069 | else |
1070 | *Tno_bw = LineTime / 4; |
1071 | |
1072 | dst_y_prefetch_equ = VStartup - (Tsetup + dml_max(TWait + TCalc, *Tdmdl)) / LineTime |
1073 | - (*DSTYAfterScaler + *DSTXAfterScaler / myPipe->HTotal); |
1074 | |
1075 | Lsw_oto = dml_max(PrefetchSourceLinesY, PrefetchSourceLinesC); |
1076 | Tsw_oto = Lsw_oto * LineTime; |
1077 | |
1078 | prefetch_bw_oto = (PrefetchSourceLinesY * swath_width_luma_ub * BytePerPixelY + PrefetchSourceLinesC * swath_width_chroma_ub * BytePerPixelC) / Tsw_oto; |
1079 | |
1080 | if (GPUVMEnable == true1) { |
1081 | Tvm_oto = dml_max3(*Tno_bw + PDEAndMetaPTEBytesFrame * HostVMInefficiencyFactor / prefetch_bw_oto, |
1082 | Tvm_trips, |
1083 | LineTime / 4.0); |
1084 | } else |
1085 | Tvm_oto = LineTime / 4.0; |
1086 | |
1087 | if ((GPUVMEnable == true1 || myPipe->DCCEnable == true1)) { |
1088 | Tr0_oto = dml_max3( |
1089 | (MetaRowByte + PixelPTEBytesPerRow * HostVMInefficiencyFactor) / prefetch_bw_oto, |
1090 | LineTime - Tvm_oto, LineTime / 4); |
1091 | } else |
1092 | Tr0_oto = (LineTime - Tvm_oto) / 2.0; |
1093 | |
1094 | Tvm_oto_lines = dml_ceil(4.0 * Tvm_oto / LineTime, 1) / 4.0; |
1095 | Tr0_oto_lines = dml_ceil(4.0 * Tr0_oto / LineTime, 1) / 4.0; |
1096 | dst_y_prefetch_oto = Tvm_oto_lines + 2 * Tr0_oto_lines + Lsw_oto; |
1097 | |
1098 | dst_y_prefetch_equ = dml_floor(4.0 * (dst_y_prefetch_equ + 0.125), 1) / 4.0; |
1099 | Tpre_rounded = dst_y_prefetch_equ * LineTime; |
1100 | |
1101 | dml_print("DML: dst_y_prefetch_oto: %f\n", dst_y_prefetch_oto){do { } while(0); }; |
1102 | dml_print("DML: dst_y_prefetch_equ: %f\n", dst_y_prefetch_equ){do { } while(0); }; |
1103 | |
1104 | dml_print("DML: LineTime: %f\n", LineTime){do { } while(0); }; |
1105 | dml_print("DML: VStartup: %d\n", VStartup){do { } while(0); }; |
1106 | dml_print("DML: Tvstartup: %fus - time between vstartup and first pixel of active\n", VStartup * LineTime){do { } while(0); }; |
1107 | dml_print("DML: Tsetup: %fus - time from vstartup to vready\n", Tsetup){do { } while(0); }; |
1108 | dml_print("DML: TCalc: %fus - time for calculations in dchub starting at vready\n", TCalc){do { } while(0); }; |
1109 | dml_print("DML: TWait: %fus - time for fabric to become ready max(pstate exit,cstate enter/exit, urgent latency) after TCalc\n", TWait){do { } while(0); }; |
1110 | dml_print("DML: Tdmbf: %fus - time for dmd transfer from dchub to dio output buffer\n", Tdmbf){do { } while(0); }; |
1111 | dml_print("DML: Tdmec: %fus - time dio takes to transfer dmd\n", Tdmec){do { } while(0); }; |
1112 | dml_print("DML: Tdmsks: %fus - time before active dmd must complete transmission at dio\n", Tdmsks){do { } while(0); }; |
1113 | dml_print("DML: Tdmdl_vm: %fus - time for vm stages of dmd \n", *Tdmdl_vm){do { } while(0); }; |
1114 | dml_print("DML: Tdmdl: %fus - time for fabric to become ready and fetch dmd \n", *Tdmdl){do { } while(0); }; |
1115 | dml_print("DML: dst_x_after_scl: %f pixels - number of pixel clocks pipeline and buffer delay after scaler \n", *DSTXAfterScaler){do { } while(0); }; |
1116 | dml_print("DML: dst_y_after_scl: %d lines - number of lines of pipeline and buffer delay after scaler \n", (int)*DSTYAfterScaler){do { } while(0); }; |
1117 | |
1118 | *PrefetchBandwidth = 0; |
1119 | *DestinationLinesToRequestVMInVBlank = 0; |
1120 | *DestinationLinesToRequestRowInVBlank = 0; |
1121 | *VRatioPrefetchY = 0; |
1122 | *VRatioPrefetchC = 0; |
1123 | *RequiredPrefetchPixDataBWLuma = 0; |
1124 | if (dst_y_prefetch_equ > 1) { |
1125 | double PrefetchBandwidth1 = 0; |
1126 | double PrefetchBandwidth2 = 0; |
1127 | double PrefetchBandwidth3 = 0; |
1128 | double PrefetchBandwidth4 = 0; |
1129 | |
1130 | if (Tpre_rounded - *Tno_bw > 0) |
1131 | PrefetchBandwidth1 = (PDEAndMetaPTEBytesFrame * HostVMInefficiencyFactor + 2 * MetaRowByte |
1132 | + 2 * PixelPTEBytesPerRow * HostVMInefficiencyFactor |
1133 | + PrefetchSourceLinesY * swath_width_luma_ub * BytePerPixelY |
1134 | + PrefetchSourceLinesC * swath_width_chroma_ub * BytePerPixelC) |
1135 | / (Tpre_rounded - *Tno_bw); |
1136 | else |
1137 | PrefetchBandwidth1 = 0; |
1138 | |
1139 | if (VStartup == MaxVStartup && (PrefetchBandwidth1 > 4 * prefetch_bw_oto) && (Tpre_rounded - Tsw_oto / 4 - 0.75 * LineTime - *Tno_bw) > 0) { |
1140 | PrefetchBandwidth1 = (PDEAndMetaPTEBytesFrame * HostVMInefficiencyFactor + 2 * MetaRowByte + 2 * PixelPTEBytesPerRow * HostVMInefficiencyFactor) / (Tpre_rounded - Tsw_oto / 4 - 0.75 * LineTime - *Tno_bw); |
1141 | } |
1142 | |
1143 | if (Tpre_rounded - *Tno_bw - 2 * Tr0_trips_rounded > 0) |
1144 | PrefetchBandwidth2 = (PDEAndMetaPTEBytesFrame * |
1145 | HostVMInefficiencyFactor + PrefetchSourceLinesY * |
1146 | swath_width_luma_ub * BytePerPixelY + |
1147 | PrefetchSourceLinesC * swath_width_chroma_ub * |
1148 | BytePerPixelC) / |
1149 | (Tpre_rounded - *Tno_bw - 2 * Tr0_trips_rounded); |
1150 | else |
1151 | PrefetchBandwidth2 = 0; |
1152 | |
1153 | if (Tpre_rounded - Tvm_trips_rounded > 0) |
1154 | PrefetchBandwidth3 = (2 * MetaRowByte + 2 * PixelPTEBytesPerRow * |
1155 | HostVMInefficiencyFactor + PrefetchSourceLinesY * |
1156 | swath_width_luma_ub * BytePerPixelY + PrefetchSourceLinesC * |
1157 | swath_width_chroma_ub * BytePerPixelC) / (Tpre_rounded - |
1158 | Tvm_trips_rounded); |
1159 | else |
1160 | PrefetchBandwidth3 = 0; |
1161 | |
1162 | if (VStartup == MaxVStartup && (PrefetchBandwidth3 > 4 * prefetch_bw_oto) && Tpre_rounded - Tsw_oto / 4 - 0.75 * LineTime - Tvm_trips_rounded > 0) { |
1163 | PrefetchBandwidth3 = (2 * MetaRowByte + 2 * PixelPTEBytesPerRow * HostVMInefficiencyFactor) / (Tpre_rounded - Tsw_oto / 4 - 0.75 * LineTime - Tvm_trips_rounded); |
1164 | } |
1165 | |
1166 | if (Tpre_rounded - Tvm_trips_rounded - 2 * Tr0_trips_rounded > 0) |
1167 | PrefetchBandwidth4 = (PrefetchSourceLinesY * swath_width_luma_ub * BytePerPixelY + PrefetchSourceLinesC * swath_width_chroma_ub * BytePerPixelC) |
1168 | / (Tpre_rounded - Tvm_trips_rounded - 2 * Tr0_trips_rounded); |
1169 | else |
1170 | PrefetchBandwidth4 = 0; |
1171 | |
1172 | { |
1173 | bool_Bool Case1OK; |
1174 | bool_Bool Case2OK; |
1175 | bool_Bool Case3OK; |
1176 | |
1177 | if (PrefetchBandwidth1 > 0) { |
1178 | if (*Tno_bw + PDEAndMetaPTEBytesFrame * HostVMInefficiencyFactor / PrefetchBandwidth1 |
1179 | >= Tvm_trips_rounded && (MetaRowByte + PixelPTEBytesPerRow * HostVMInefficiencyFactor) / PrefetchBandwidth1 >= Tr0_trips_rounded) { |
1180 | Case1OK = true1; |
1181 | } else { |
1182 | Case1OK = false0; |
1183 | } |
1184 | } else { |
1185 | Case1OK = false0; |
1186 | } |
1187 | |
1188 | if (PrefetchBandwidth2 > 0) { |
1189 | if (*Tno_bw + PDEAndMetaPTEBytesFrame * HostVMInefficiencyFactor / PrefetchBandwidth2 |
1190 | >= Tvm_trips_rounded && (MetaRowByte + PixelPTEBytesPerRow * HostVMInefficiencyFactor) / PrefetchBandwidth2 < Tr0_trips_rounded) { |
1191 | Case2OK = true1; |
1192 | } else { |
1193 | Case2OK = false0; |
1194 | } |
1195 | } else { |
1196 | Case2OK = false0; |
1197 | } |
1198 | |
1199 | if (PrefetchBandwidth3 > 0) { |
1200 | if (*Tno_bw + PDEAndMetaPTEBytesFrame * HostVMInefficiencyFactor / PrefetchBandwidth3 |
1201 | < Tvm_trips_rounded && (MetaRowByte + PixelPTEBytesPerRow * HostVMInefficiencyFactor) / PrefetchBandwidth3 >= Tr0_trips_rounded) { |
1202 | Case3OK = true1; |
1203 | } else { |
1204 | Case3OK = false0; |
1205 | } |
1206 | } else { |
1207 | Case3OK = false0; |
1208 | } |
1209 | |
1210 | if (Case1OK) { |
1211 | prefetch_bw_equ = PrefetchBandwidth1; |
1212 | } else if (Case2OK) { |
1213 | prefetch_bw_equ = PrefetchBandwidth2; |
1214 | } else if (Case3OK) { |
1215 | prefetch_bw_equ = PrefetchBandwidth3; |
1216 | } else { |
1217 | prefetch_bw_equ = PrefetchBandwidth4; |
1218 | } |
1219 | |
1220 | dml_print("DML: prefetch_bw_equ: %f\n", prefetch_bw_equ){do { } while(0); }; |
1221 | |
1222 | if (prefetch_bw_equ > 0) { |
1223 | if (GPUVMEnable == true1) { |
1224 | Tvm_equ = dml_max3(*Tno_bw + PDEAndMetaPTEBytesFrame * HostVMInefficiencyFactor / prefetch_bw_equ, Tvm_trips, LineTime / 4); |
1225 | } else { |
1226 | Tvm_equ = LineTime / 4; |
1227 | } |
1228 | |
1229 | if ((GPUVMEnable == true1 || myPipe->DCCEnable == true1)) { |
1230 | Tr0_equ = dml_max4( |
1231 | (MetaRowByte + PixelPTEBytesPerRow * HostVMInefficiencyFactor) / prefetch_bw_equ, |
1232 | Tr0_trips, |
1233 | (LineTime - Tvm_equ) / 2, |
1234 | LineTime / 4); |
1235 | } else { |
1236 | Tr0_equ = (LineTime - Tvm_equ) / 2; |
1237 | } |
1238 | } else { |
1239 | Tvm_equ = 0; |
1240 | Tr0_equ = 0; |
1241 | dml_print("DML: prefetch_bw_equ equals 0! %s:%d\n", __FILE__, __LINE__){do { } while(0); }; |
1242 | } |
1243 | } |
1244 | |
1245 | if (dst_y_prefetch_oto < dst_y_prefetch_equ) { |
1246 | *DestinationLinesForPrefetch = dst_y_prefetch_oto; |
1247 | TimeForFetchingMetaPTE = Tvm_oto; |
1248 | TimeForFetchingRowInVBlank = Tr0_oto; |
1249 | *PrefetchBandwidth = prefetch_bw_oto; |
1250 | } else { |
1251 | *DestinationLinesForPrefetch = dst_y_prefetch_equ; |
1252 | TimeForFetchingMetaPTE = Tvm_equ; |
1253 | TimeForFetchingRowInVBlank = Tr0_equ; |
1254 | *PrefetchBandwidth = prefetch_bw_equ; |
1255 | } |
1256 | |
1257 | *DestinationLinesToRequestVMInVBlank = dml_ceil(4.0 * TimeForFetchingMetaPTE / LineTime, 1.0) / 4.0; |
1258 | |
1259 | *DestinationLinesToRequestRowInVBlank = dml_ceil(4.0 * TimeForFetchingRowInVBlank / LineTime, 1.0) / 4.0; |
1260 | |
1261 | |
1262 | LinesToRequestPrefetchPixelData = *DestinationLinesForPrefetch - *DestinationLinesToRequestVMInVBlank |
1263 | - 2 * *DestinationLinesToRequestRowInVBlank; |
1264 | |
1265 | if (LinesToRequestPrefetchPixelData > 0 && prefetch_bw_equ > 0) { |
1266 | |
1267 | *VRatioPrefetchY = (double) PrefetchSourceLinesY |
1268 | / LinesToRequestPrefetchPixelData; |
1269 | *VRatioPrefetchY = dml_max(*VRatioPrefetchY, 1.0); |
1270 | if ((SwathHeightY > 4) && (VInitPreFillY > 3)) { |
1271 | if (LinesToRequestPrefetchPixelData > (VInitPreFillY - 3.0) / 2.0) { |
1272 | *VRatioPrefetchY = dml_max((double) PrefetchSourceLinesY / LinesToRequestPrefetchPixelData, |
1273 | (double) MaxNumSwathY * SwathHeightY / (LinesToRequestPrefetchPixelData - (VInitPreFillY - 3.0) / 2.0)); |
1274 | *VRatioPrefetchY = dml_max(*VRatioPrefetchY, 1.0); |
1275 | } else { |
1276 | MyError = true1; |
1277 | dml_print("DML: MyErr set %s:%d\n", __FILE__, __LINE__){do { } while(0); }; |
1278 | *VRatioPrefetchY = 0; |
1279 | } |
1280 | } |
1281 | |
1282 | *VRatioPrefetchC = (double) PrefetchSourceLinesC / LinesToRequestPrefetchPixelData; |
1283 | *VRatioPrefetchC = dml_max(*VRatioPrefetchC, 1.0); |
1284 | |
1285 | if ((SwathHeightC > 4)) { |
1286 | if (LinesToRequestPrefetchPixelData > (VInitPreFillC - 3.0) / 2.0) { |
1287 | *VRatioPrefetchC = dml_max(*VRatioPrefetchC, |
1288 | (double) MaxNumSwathC * SwathHeightC / (LinesToRequestPrefetchPixelData - (VInitPreFillC - 3.0) / 2.0)); |
1289 | *VRatioPrefetchC = dml_max(*VRatioPrefetchC, 1.0); |
1290 | } else { |
1291 | MyError = true1; |
1292 | dml_print("DML: MyErr set %s:%d\n", __FILE__, __LINE__){do { } while(0); }; |
1293 | *VRatioPrefetchC = 0; |
1294 | } |
1295 | } |
1296 | |
1297 | *RequiredPrefetchPixDataBWLuma = (double) PrefetchSourceLinesY / LinesToRequestPrefetchPixelData * BytePerPixelY * swath_width_luma_ub / LineTime; |
1298 | *RequiredPrefetchPixDataBWChroma = (double) PrefetchSourceLinesC / LinesToRequestPrefetchPixelData * BytePerPixelC * swath_width_chroma_ub / LineTime; |
1299 | } else { |
1300 | MyError = true1; |
1301 | dml_print("DML: MyErr set %s:%d\n", __FILE__, __LINE__){do { } while(0); }; |
1302 | dml_print("DML: LinesToRequestPrefetchPixelData: %f, should be > 0\n", LinesToRequestPrefetchPixelData){do { } while(0); }; |
1303 | *VRatioPrefetchY = 0; |
1304 | *VRatioPrefetchC = 0; |
1305 | *RequiredPrefetchPixDataBWLuma = 0; |
1306 | *RequiredPrefetchPixDataBWChroma = 0; |
1307 | } |
1308 | |
1309 | dml_print("DML: Tpre: %fus - sum of tim to request meta pte, 2 x data pte + meta data, swaths\n", (double)LinesToRequestPrefetchPixelData * LineTime + 2.0*TimeForFetchingRowInVBlank + TimeForFetchingMetaPTE){do { } while(0); }; |
1310 | dml_print("DML: Tvm: %fus - time to fetch page tables for meta surface\n", TimeForFetchingMetaPTE){do { } while(0); }; |
1311 | dml_print("DML: Tr0: %fus - time to fetch first row of data pagetables and first row of meta data (done in parallel)\n", TimeForFetchingRowInVBlank){do { } while(0); }; |
1312 | dml_print("DML: Tr1: %fus - time to fetch second row of data pagetables and second row of meta data (done in parallel)\n", TimeForFetchingRowInVBlank){do { } while(0); }; |
1313 | dml_print("DML: Tsw: %fus = time to fetch enough pixel data and cursor data to feed the scalers init position and detile\n", (double)LinesToRequestPrefetchPixelData * LineTime){do { } while(0); }; |
1314 | dml_print("DML: To: %fus - time for propagation from scaler to optc\n", (*DSTYAfterScaler + ((*DSTXAfterScaler) / (double) myPipe->HTotal)) * LineTime){do { } while(0); }; |
1315 | dml_print("DML: Tvstartup - Tsetup - Tcalc - Twait - Tpre - To > 0\n"){do { } while(0); }; |
1316 | dml_print("DML: Tslack(pre): %fus - time left over in schedule\n", VStartup * LineTime - TimeForFetchingMetaPTE - 2 * TimeForFetchingRowInVBlank - (*DSTYAfterScaler + ((*DSTXAfterScaler) / (double) myPipe->HTotal)) * LineTime - TWait - TCalc - Tsetup){do { } while(0); }; |
1317 | dml_print("DML: row_bytes = dpte_row_bytes (per_pipe) = PixelPTEBytesPerRow = : %d\n", PixelPTEBytesPerRow){do { } while(0); }; |
1318 | |
1319 | } else { |
1320 | MyError = true1; |
1321 | dml_print("DML: MyErr set %s:%d\n", __FILE__, __LINE__){do { } while(0); }; |
1322 | } |
1323 | |
1324 | { |
1325 | double prefetch_vm_bw = 0; |
1326 | double prefetch_row_bw = 0; |
1327 | |
1328 | if (PDEAndMetaPTEBytesFrame == 0) { |
1329 | prefetch_vm_bw = 0; |
1330 | } else if (*DestinationLinesToRequestVMInVBlank > 0) { |
1331 | prefetch_vm_bw = PDEAndMetaPTEBytesFrame * HostVMInefficiencyFactor / (*DestinationLinesToRequestVMInVBlank * LineTime); |
1332 | } else { |
1333 | prefetch_vm_bw = 0; |
1334 | MyError = true1; |
1335 | dml_print("DML: MyErr set %s:%d\n", __FILE__, __LINE__){do { } while(0); }; |
1336 | } |
1337 | if (MetaRowByte + PixelPTEBytesPerRow == 0) { |
1338 | prefetch_row_bw = 0; |
1339 | } else if (*DestinationLinesToRequestRowInVBlank > 0) { |
1340 | prefetch_row_bw = (MetaRowByte + PixelPTEBytesPerRow * HostVMInefficiencyFactor) / (*DestinationLinesToRequestRowInVBlank * LineTime); |
1341 | } else { |
1342 | prefetch_row_bw = 0; |
1343 | MyError = true1; |
1344 | dml_print("DML: MyErr set %s:%d\n", __FILE__, __LINE__){do { } while(0); }; |
1345 | } |
1346 | |
1347 | *prefetch_vmrow_bw = dml_max(prefetch_vm_bw, prefetch_row_bw); |
1348 | } |
1349 | |
1350 | if (MyError) { |
1351 | *PrefetchBandwidth = 0; |
1352 | TimeForFetchingMetaPTE = 0; |
1353 | TimeForFetchingRowInVBlank = 0; |
1354 | *DestinationLinesToRequestVMInVBlank = 0; |
1355 | *DestinationLinesToRequestRowInVBlank = 0; |
1356 | *DestinationLinesForPrefetch = 0; |
1357 | LinesToRequestPrefetchPixelData = 0; |
1358 | *VRatioPrefetchY = 0; |
1359 | *VRatioPrefetchC = 0; |
1360 | *RequiredPrefetchPixDataBWLuma = 0; |
1361 | *RequiredPrefetchPixDataBWChroma = 0; |
1362 | } |
1363 | |
1364 | return MyError; |
1365 | } |
1366 | |
1367 | static double RoundToDFSGranularityUp(double Clock, double VCOSpeed) |
1368 | { |
1369 | return VCOSpeed * 4 / dml_floor(VCOSpeed * 4 / Clock, 1); |
1370 | } |
1371 | |
1372 | static double RoundToDFSGranularityDown(double Clock, double VCOSpeed) |
1373 | { |
1374 | return VCOSpeed * 4 / dml_ceil(VCOSpeed * 4.0 / Clock, 1); |
1375 | } |
1376 | |
1377 | static void CalculateDCCConfiguration( |
1378 | bool_Bool DCCEnabled, |
1379 | bool_Bool DCCProgrammingAssumesScanDirectionUnknown, |
1380 | enum source_format_class SourcePixelFormat, |
1381 | unsigned int SurfaceWidthLuma, |
1382 | unsigned int SurfaceWidthChroma, |
1383 | unsigned int SurfaceHeightLuma, |
1384 | unsigned int SurfaceHeightChroma, |
1385 | double DETBufferSize, |
1386 | unsigned int RequestHeight256ByteLuma, |
1387 | unsigned int RequestHeight256ByteChroma, |
1388 | enum dm_swizzle_mode TilingFormat, |
1389 | unsigned int BytePerPixelY, |
1390 | unsigned int BytePerPixelC, |
1391 | double BytePerPixelDETY, |
1392 | double BytePerPixelDETC, |
1393 | enum scan_direction_class ScanOrientation, |
1394 | unsigned int *MaxUncompressedBlockLuma, |
1395 | unsigned int *MaxUncompressedBlockChroma, |
1396 | unsigned int *MaxCompressedBlockLuma, |
1397 | unsigned int *MaxCompressedBlockChroma, |
1398 | unsigned int *IndependentBlockLuma, |
1399 | unsigned int *IndependentBlockChroma) |
1400 | { |
1401 | int yuv420 = 0; |
1402 | int horz_div_l = 0; |
1403 | int horz_div_c = 0; |
1404 | int vert_div_l = 0; |
1405 | int vert_div_c = 0; |
1406 | |
1407 | int req128_horz_wc_l = 0; |
1408 | int req128_horz_wc_c = 0; |
1409 | int req128_vert_wc_l = 0; |
1410 | int req128_vert_wc_c = 0; |
1411 | int segment_order_horz_contiguous_luma = 0; |
1412 | int segment_order_horz_contiguous_chroma = 0; |
1413 | int segment_order_vert_contiguous_luma = 0; |
1414 | int segment_order_vert_contiguous_chroma = 0; |
1415 | |
1416 | long full_swath_bytes_horz_wc_l = 0; |
1417 | long full_swath_bytes_horz_wc_c = 0; |
1418 | long full_swath_bytes_vert_wc_l = 0; |
1419 | long full_swath_bytes_vert_wc_c = 0; |
1420 | |
1421 | long swath_buf_size = 0; |
1422 | double detile_buf_vp_horz_limit = 0; |
1423 | double detile_buf_vp_vert_limit = 0; |
1424 | |
1425 | long MAS_vp_horz_limit = 0; |
1426 | long MAS_vp_vert_limit = 0; |
1427 | long max_vp_horz_width = 0; |
1428 | long max_vp_vert_height = 0; |
1429 | long eff_surf_width_l = 0; |
1430 | long eff_surf_width_c = 0; |
1431 | long eff_surf_height_l = 0; |
1432 | long eff_surf_height_c = 0; |
1433 | |
1434 | typedef enum { |
1435 | REQ_256Bytes, |
1436 | REQ_128BytesNonContiguous, |
1437 | REQ_128BytesContiguous, |
1438 | REQ_NA |
1439 | } RequestType; |
1440 | |
1441 | RequestType RequestLuma; |
1442 | RequestType RequestChroma; |
1443 | |
1444 | yuv420 = ((SourcePixelFormat == dm_420_8 || SourcePixelFormat == dm_420_10 || SourcePixelFormat == dm_420_12) ? 1 : 0); |
1445 | horz_div_l = 1; |
1446 | horz_div_c = 1; |
1447 | vert_div_l = 1; |
1448 | vert_div_c = 1; |
1449 | |
1450 | if (BytePerPixelY == 1) |
1451 | vert_div_l = 0; |
1452 | if (BytePerPixelC == 1) |
1453 | vert_div_c = 0; |
1454 | if (BytePerPixelY == 8 |
1455 | && (TilingFormat == dm_sw_64kb_s || TilingFormat == dm_sw_64kb_s_t |
1456 | || TilingFormat == dm_sw_64kb_s_x)) |
1457 | horz_div_l = 0; |
1458 | if (BytePerPixelC == 8 |
1459 | && (TilingFormat == dm_sw_64kb_s || TilingFormat == dm_sw_64kb_s_t |
1460 | || TilingFormat == dm_sw_64kb_s_x)) |
1461 | horz_div_c = 0; |
1462 | |
1463 | if (BytePerPixelC == 0) { |
1464 | swath_buf_size = DETBufferSize / 2 - 2 * 256; |
1465 | detile_buf_vp_horz_limit = (double) swath_buf_size |
1466 | / ((double) RequestHeight256ByteLuma * BytePerPixelY |
1467 | / (1 + horz_div_l)); |
1468 | detile_buf_vp_vert_limit = (double) swath_buf_size |
1469 | / (256.0 / RequestHeight256ByteLuma / (1 + vert_div_l)); |
1470 | } else { |
1471 | swath_buf_size = DETBufferSize / 2 - 2 * 2 * 256; |
1472 | detile_buf_vp_horz_limit = (double) swath_buf_size |
1473 | / ((double) RequestHeight256ByteLuma * BytePerPixelY |
1474 | / (1 + horz_div_l) |
1475 | + (double) RequestHeight256ByteChroma |
1476 | * BytePerPixelC / (1 + horz_div_c) |
1477 | / (1 + yuv420)); |
1478 | detile_buf_vp_vert_limit = (double) swath_buf_size |
1479 | / (256.0 / RequestHeight256ByteLuma / (1 + vert_div_l) |
1480 | + 256.0 / RequestHeight256ByteChroma |
1481 | / (1 + vert_div_c) / (1 + yuv420)); |
1482 | } |
1483 | |
1484 | if (SourcePixelFormat == dm_420_10) { |
1485 | detile_buf_vp_horz_limit = 1.5 * detile_buf_vp_horz_limit; |
1486 | detile_buf_vp_vert_limit = 1.5 * detile_buf_vp_vert_limit; |
1487 | } |
1488 | |
1489 | detile_buf_vp_horz_limit = dml_floor(detile_buf_vp_horz_limit - 1, 16); |
1490 | detile_buf_vp_vert_limit = dml_floor(detile_buf_vp_vert_limit - 1, 16); |
1491 | |
1492 | MAS_vp_horz_limit = 5760; |
1493 | MAS_vp_vert_limit = (BytePerPixelC > 0 ? 2880 : 5760); |
1494 | max_vp_horz_width = dml_min((double) MAS_vp_horz_limit, detile_buf_vp_horz_limit); |
1495 | max_vp_vert_height = dml_min((double) MAS_vp_vert_limit, detile_buf_vp_vert_limit); |
1496 | eff_surf_width_l = |
1497 | (SurfaceWidthLuma > max_vp_horz_width ? max_vp_horz_width : SurfaceWidthLuma); |
1498 | eff_surf_width_c = eff_surf_width_l / (1 + yuv420); |
1499 | eff_surf_height_l = ( |
1500 | SurfaceHeightLuma > max_vp_vert_height ? |
1501 | max_vp_vert_height : SurfaceHeightLuma); |
1502 | eff_surf_height_c = eff_surf_height_l / (1 + yuv420); |
1503 | |
1504 | full_swath_bytes_horz_wc_l = eff_surf_width_l * RequestHeight256ByteLuma * BytePerPixelY; |
1505 | full_swath_bytes_vert_wc_l = eff_surf_height_l * 256 / RequestHeight256ByteLuma; |
1506 | if (BytePerPixelC > 0) { |
1507 | full_swath_bytes_horz_wc_c = eff_surf_width_c * RequestHeight256ByteChroma |
1508 | * BytePerPixelC; |
1509 | full_swath_bytes_vert_wc_c = eff_surf_height_c * 256 / RequestHeight256ByteChroma; |
1510 | } else { |
1511 | full_swath_bytes_horz_wc_c = 0; |
1512 | full_swath_bytes_vert_wc_c = 0; |
1513 | } |
1514 | |
1515 | if (SourcePixelFormat == dm_420_10) { |
1516 | full_swath_bytes_horz_wc_l = dml_ceil(full_swath_bytes_horz_wc_l * 2 / 3, 256); |
1517 | full_swath_bytes_horz_wc_c = dml_ceil(full_swath_bytes_horz_wc_c * 2 / 3, 256); |
1518 | full_swath_bytes_vert_wc_l = dml_ceil(full_swath_bytes_vert_wc_l * 2 / 3, 256); |
1519 | full_swath_bytes_vert_wc_c = dml_ceil(full_swath_bytes_vert_wc_c * 2 / 3, 256); |
1520 | } |
1521 | |
1522 | if (2 * full_swath_bytes_horz_wc_l + 2 * full_swath_bytes_horz_wc_c <= DETBufferSize) { |
1523 | req128_horz_wc_l = 0; |
1524 | req128_horz_wc_c = 0; |
1525 | } else if (full_swath_bytes_horz_wc_l < 1.5 * full_swath_bytes_horz_wc_c |
1526 | && 2 * full_swath_bytes_horz_wc_l + full_swath_bytes_horz_wc_c |
1527 | <= DETBufferSize) { |
1528 | req128_horz_wc_l = 0; |
1529 | req128_horz_wc_c = 1; |
1530 | } else if (full_swath_bytes_horz_wc_l >= 1.5 * full_swath_bytes_horz_wc_c |
1531 | && full_swath_bytes_horz_wc_l + 2 * full_swath_bytes_horz_wc_c |
1532 | <= DETBufferSize) { |
1533 | req128_horz_wc_l = 1; |
1534 | req128_horz_wc_c = 0; |
1535 | } else { |
1536 | req128_horz_wc_l = 1; |
1537 | req128_horz_wc_c = 1; |
1538 | } |
1539 | |
1540 | if (2 * full_swath_bytes_vert_wc_l + 2 * full_swath_bytes_vert_wc_c <= DETBufferSize) { |
1541 | req128_vert_wc_l = 0; |
1542 | req128_vert_wc_c = 0; |
1543 | } else if (full_swath_bytes_vert_wc_l < 1.5 * full_swath_bytes_vert_wc_c |
1544 | && 2 * full_swath_bytes_vert_wc_l + full_swath_bytes_vert_wc_c |
1545 | <= DETBufferSize) { |
1546 | req128_vert_wc_l = 0; |
1547 | req128_vert_wc_c = 1; |
1548 | } else if (full_swath_bytes_vert_wc_l >= 1.5 * full_swath_bytes_vert_wc_c |
1549 | && full_swath_bytes_vert_wc_l + 2 * full_swath_bytes_vert_wc_c |
1550 | <= DETBufferSize) { |
1551 | req128_vert_wc_l = 1; |
1552 | req128_vert_wc_c = 0; |
1553 | } else { |
1554 | req128_vert_wc_l = 1; |
1555 | req128_vert_wc_c = 1; |
1556 | } |
1557 | |
1558 | if (BytePerPixelY == 2 || (BytePerPixelY == 4 && TilingFormat != dm_sw_64kb_r_x)) { |
1559 | segment_order_horz_contiguous_luma = 0; |
1560 | } else { |
1561 | segment_order_horz_contiguous_luma = 1; |
1562 | } |
1563 | if ((BytePerPixelY == 8 |
1564 | && (TilingFormat == dm_sw_64kb_d || TilingFormat == dm_sw_64kb_d_x |
1565 | || TilingFormat == dm_sw_64kb_d_t |
1566 | || TilingFormat == dm_sw_64kb_r_x)) |
1567 | || (BytePerPixelY == 4 && TilingFormat == dm_sw_64kb_r_x)) { |
1568 | segment_order_vert_contiguous_luma = 0; |
1569 | } else { |
1570 | segment_order_vert_contiguous_luma = 1; |
1571 | } |
1572 | if (BytePerPixelC == 2 || (BytePerPixelC == 4 && TilingFormat != dm_sw_64kb_r_x)) { |
1573 | segment_order_horz_contiguous_chroma = 0; |
1574 | } else { |
1575 | segment_order_horz_contiguous_chroma = 1; |
1576 | } |
1577 | if ((BytePerPixelC == 8 |
1578 | && (TilingFormat == dm_sw_64kb_d || TilingFormat == dm_sw_64kb_d_x |
1579 | || TilingFormat == dm_sw_64kb_d_t |
1580 | || TilingFormat == dm_sw_64kb_r_x)) |
1581 | || (BytePerPixelC == 4 && TilingFormat == dm_sw_64kb_r_x)) { |
1582 | segment_order_vert_contiguous_chroma = 0; |
1583 | } else { |
1584 | segment_order_vert_contiguous_chroma = 1; |
1585 | } |
1586 | |
1587 | if (DCCProgrammingAssumesScanDirectionUnknown == true1) { |
1588 | if (req128_horz_wc_l == 0 && req128_vert_wc_l == 0) { |
1589 | RequestLuma = REQ_256Bytes; |
1590 | } else if ((req128_horz_wc_l == 1 && segment_order_horz_contiguous_luma == 0) |
1591 | || (req128_vert_wc_l == 1 && segment_order_vert_contiguous_luma == 0)) { |
1592 | RequestLuma = REQ_128BytesNonContiguous; |
1593 | } else { |
1594 | RequestLuma = REQ_128BytesContiguous; |
1595 | } |
1596 | if (req128_horz_wc_c == 0 && req128_vert_wc_c == 0) { |
1597 | RequestChroma = REQ_256Bytes; |
1598 | } else if ((req128_horz_wc_c == 1 && segment_order_horz_contiguous_chroma == 0) |
1599 | || (req128_vert_wc_c == 1 |
1600 | && segment_order_vert_contiguous_chroma == 0)) { |
1601 | RequestChroma = REQ_128BytesNonContiguous; |
1602 | } else { |
1603 | RequestChroma = REQ_128BytesContiguous; |
1604 | } |
1605 | } else if (ScanOrientation != dm_vert) { |
1606 | if (req128_horz_wc_l == 0) { |
1607 | RequestLuma = REQ_256Bytes; |
1608 | } else if (segment_order_horz_contiguous_luma == 0) { |
1609 | RequestLuma = REQ_128BytesNonContiguous; |
1610 | } else { |
1611 | RequestLuma = REQ_128BytesContiguous; |
1612 | } |
1613 | if (req128_horz_wc_c == 0) { |
1614 | RequestChroma = REQ_256Bytes; |
1615 | } else if (segment_order_horz_contiguous_chroma == 0) { |
1616 | RequestChroma = REQ_128BytesNonContiguous; |
1617 | } else { |
1618 | RequestChroma = REQ_128BytesContiguous; |
1619 | } |
1620 | } else { |
1621 | if (req128_vert_wc_l == 0) { |
1622 | RequestLuma = REQ_256Bytes; |
1623 | } else if (segment_order_vert_contiguous_luma == 0) { |
1624 | RequestLuma = REQ_128BytesNonContiguous; |
1625 | } else { |
1626 | RequestLuma = REQ_128BytesContiguous; |
1627 | } |
1628 | if (req128_vert_wc_c == 0) { |
1629 | RequestChroma = REQ_256Bytes; |
1630 | } else if (segment_order_vert_contiguous_chroma == 0) { |
1631 | RequestChroma = REQ_128BytesNonContiguous; |
1632 | } else { |
1633 | RequestChroma = REQ_128BytesContiguous; |
1634 | } |
1635 | } |
1636 | |
1637 | if (RequestLuma == REQ_256Bytes) { |
1638 | *MaxUncompressedBlockLuma = 256; |
1639 | *MaxCompressedBlockLuma = 256; |
1640 | *IndependentBlockLuma = 0; |
1641 | } else if (RequestLuma == REQ_128BytesContiguous) { |
1642 | *MaxUncompressedBlockLuma = 256; |
1643 | *MaxCompressedBlockLuma = 128; |
1644 | *IndependentBlockLuma = 128; |
1645 | } else { |
1646 | *MaxUncompressedBlockLuma = 256; |
1647 | *MaxCompressedBlockLuma = 64; |
1648 | *IndependentBlockLuma = 64; |
1649 | } |
1650 | |
1651 | if (RequestChroma == REQ_256Bytes) { |
1652 | *MaxUncompressedBlockChroma = 256; |
1653 | *MaxCompressedBlockChroma = 256; |
1654 | *IndependentBlockChroma = 0; |
1655 | } else if (RequestChroma == REQ_128BytesContiguous) { |
1656 | *MaxUncompressedBlockChroma = 256; |
1657 | *MaxCompressedBlockChroma = 128; |
1658 | *IndependentBlockChroma = 128; |
1659 | } else { |
1660 | *MaxUncompressedBlockChroma = 256; |
1661 | *MaxCompressedBlockChroma = 64; |
1662 | *IndependentBlockChroma = 64; |
1663 | } |
1664 | |
1665 | if (DCCEnabled != true1 || BytePerPixelC == 0) { |
1666 | *MaxUncompressedBlockChroma = 0; |
1667 | *MaxCompressedBlockChroma = 0; |
1668 | *IndependentBlockChroma = 0; |
1669 | } |
1670 | |
1671 | if (DCCEnabled != true1) { |
1672 | *MaxUncompressedBlockLuma = 0; |
1673 | *MaxCompressedBlockLuma = 0; |
1674 | *IndependentBlockLuma = 0; |
1675 | } |
1676 | } |
1677 | |
1678 | |
1679 | static double CalculatePrefetchSourceLines( |
1680 | struct display_mode_lib *mode_lib, |
1681 | double VRatio, |
1682 | double vtaps, |
1683 | bool_Bool Interlace, |
1684 | bool_Bool ProgressiveToInterlaceUnitInOPP, |
1685 | unsigned int SwathHeight, |
1686 | unsigned int ViewportYStart, |
1687 | double *VInitPreFill, |
1688 | unsigned int *MaxNumSwath) |
1689 | { |
1690 | unsigned int MaxPartialSwath = 0; |
1691 | |
1692 | if (ProgressiveToInterlaceUnitInOPP) |
1693 | *VInitPreFill = dml_floor((VRatio + vtaps + 1) / 2.0, 1); |
1694 | else |
1695 | *VInitPreFill = dml_floor((VRatio + vtaps + 1 + Interlace * 0.5 * VRatio) / 2.0, 1); |
1696 | |
1697 | if (!mode_lib->vba.IgnoreViewportPositioning) { |
1698 | |
1699 | *MaxNumSwath = dml_ceil((*VInitPreFill - 1.0) / SwathHeight, 1) + 1.0; |
1700 | |
1701 | if (*VInitPreFill > 1.0) |
1702 | MaxPartialSwath = (unsigned int) (*VInitPreFill - 2) % SwathHeight; |
1703 | else |
1704 | MaxPartialSwath = (unsigned int) (*VInitPreFill + SwathHeight - 2) |
1705 | % SwathHeight; |
1706 | MaxPartialSwath = dml_max(1U, MaxPartialSwath); |
1707 | |
1708 | } else { |
1709 | |
1710 | if (ViewportYStart != 0) |
1711 | dml_print({do { } while(0); } |
1712 | "WARNING DML: using viewport y position of 0 even though actual viewport y position is non-zero in prefetch source lines calculation\n"){do { } while(0); }; |
1713 | |
1714 | *MaxNumSwath = dml_ceil(*VInitPreFill / SwathHeight, 1); |
1715 | |
1716 | if (*VInitPreFill > 1.0) |
1717 | MaxPartialSwath = (unsigned int) (*VInitPreFill - 1) % SwathHeight; |
1718 | else |
1719 | MaxPartialSwath = (unsigned int) (*VInitPreFill + SwathHeight - 1) |
1720 | % SwathHeight; |
1721 | } |
1722 | |
1723 | return *MaxNumSwath * SwathHeight + MaxPartialSwath; |
1724 | } |
1725 | |
1726 | static unsigned int CalculateVMAndRowBytes( |
1727 | struct display_mode_lib *mode_lib, |
1728 | bool_Bool DCCEnable, |
1729 | unsigned int BlockHeight256Bytes, |
1730 | unsigned int BlockWidth256Bytes, |
1731 | enum source_format_class SourcePixelFormat, |
1732 | unsigned int SurfaceTiling, |
1733 | unsigned int BytePerPixel, |
1734 | enum scan_direction_class ScanDirection, |
1735 | unsigned int SwathWidth, |
1736 | unsigned int ViewportHeight, |
1737 | bool_Bool GPUVMEnable, |
1738 | bool_Bool HostVMEnable, |
1739 | unsigned int HostVMMaxNonCachedPageTableLevels, |
1740 | unsigned int GPUVMMinPageSize, |
1741 | unsigned int HostVMMinPageSize, |
1742 | unsigned int PTEBufferSizeInRequests, |
1743 | unsigned int Pitch, |
1744 | unsigned int DCCMetaPitch, |
1745 | unsigned int *MacroTileWidth, |
1746 | unsigned int *MetaRowByte, |
1747 | unsigned int *PixelPTEBytesPerRow, |
1748 | bool_Bool *PTEBufferSizeNotExceeded, |
1749 | unsigned int *dpte_row_width_ub, |
1750 | unsigned int *dpte_row_height, |
1751 | unsigned int *MetaRequestWidth, |
1752 | unsigned int *MetaRequestHeight, |
1753 | unsigned int *meta_row_width, |
1754 | unsigned int *meta_row_height, |
1755 | unsigned int *vm_group_bytes, |
1756 | unsigned int *dpte_group_bytes, |
1757 | unsigned int *PixelPTEReqWidth, |
1758 | unsigned int *PixelPTEReqHeight, |
1759 | unsigned int *PTERequestSize, |
1760 | unsigned int *DPDE0BytesFrame, |
1761 | unsigned int *MetaPTEBytesFrame) |
1762 | { |
1763 | unsigned int MPDEBytesFrame = 0; |
1764 | unsigned int DCCMetaSurfaceBytes = 0; |
1765 | unsigned int MacroTileSizeBytes = 0; |
1766 | unsigned int MacroTileHeight = 0; |
1767 | unsigned int ExtraDPDEBytesFrame = 0; |
1768 | unsigned int PDEAndMetaPTEBytesFrame = 0; |
1769 | unsigned int PixelPTEReqHeightPTEs = 0; |
1770 | unsigned int HostVMDynamicLevels = 0; |
1771 | |
1772 | double FractionOfPTEReturnDrop; |
1773 | |
1774 | if (GPUVMEnable == true1 && HostVMEnable == true1) { |
1775 | if (HostVMMinPageSize < 2048) { |
1776 | HostVMDynamicLevels = HostVMMaxNonCachedPageTableLevels; |
1777 | } else if (HostVMMinPageSize >= 2048 && HostVMMinPageSize < 1048576) { |
1778 | HostVMDynamicLevels = dml_max(0, (int) HostVMMaxNonCachedPageTableLevels - 1); |
1779 | } else { |
1780 | HostVMDynamicLevels = dml_max(0, (int) HostVMMaxNonCachedPageTableLevels - 2); |
1781 | } |
1782 | } |
1783 | |
1784 | *MetaRequestHeight = 8 * BlockHeight256Bytes; |
1785 | *MetaRequestWidth = 8 * BlockWidth256Bytes; |
1786 | if (ScanDirection != dm_vert) { |
1787 | *meta_row_height = *MetaRequestHeight; |
1788 | *meta_row_width = dml_ceil((double) SwathWidth - 1, *MetaRequestWidth) |
1789 | + *MetaRequestWidth; |
1790 | *MetaRowByte = *meta_row_width * *MetaRequestHeight * BytePerPixel / 256.0; |
1791 | } else { |
1792 | *meta_row_height = *MetaRequestWidth; |
1793 | *meta_row_width = dml_ceil((double) SwathWidth - 1, *MetaRequestHeight) |
1794 | + *MetaRequestHeight; |
1795 | *MetaRowByte = *meta_row_width * *MetaRequestWidth * BytePerPixel / 256.0; |
1796 | } |
1797 | DCCMetaSurfaceBytes = DCCMetaPitch * (dml_ceil(ViewportHeight - 1, 64 * BlockHeight256Bytes) |
1798 | + 64 * BlockHeight256Bytes) * BytePerPixel / 256; |
1799 | if (GPUVMEnable == true1) { |
1800 | *MetaPTEBytesFrame = (dml_ceil((double) (DCCMetaSurfaceBytes - 4.0 * 1024.0) / (8 * 4.0 * 1024), 1) + 1) * 64; |
1801 | MPDEBytesFrame = 128 * (mode_lib->vba.GPUVMMaxPageTableLevels - 1); |
1802 | } else { |
1803 | *MetaPTEBytesFrame = 0; |
1804 | MPDEBytesFrame = 0; |
1805 | } |
1806 | |
1807 | if (DCCEnable != true1) { |
1808 | *MetaPTEBytesFrame = 0; |
1809 | MPDEBytesFrame = 0; |
1810 | *MetaRowByte = 0; |
1811 | } |
1812 | |
1813 | if (SurfaceTiling == dm_sw_linear) { |
1814 | MacroTileSizeBytes = 256; |
1815 | MacroTileHeight = BlockHeight256Bytes; |
1816 | } else { |
1817 | MacroTileSizeBytes = 65536; |
1818 | MacroTileHeight = 16 * BlockHeight256Bytes; |
1819 | } |
1820 | *MacroTileWidth = MacroTileSizeBytes / BytePerPixel / MacroTileHeight; |
1821 | |
1822 | if (GPUVMEnable == true1 && mode_lib->vba.GPUVMMaxPageTableLevels > 1) { |
1823 | if (ScanDirection != dm_vert) { |
1824 | *DPDE0BytesFrame = 64 * (dml_ceil(((Pitch * (dml_ceil(ViewportHeight - 1, MacroTileHeight) + MacroTileHeight) * BytePerPixel) - MacroTileSizeBytes) / (8 * 2097152), 1) + 1); |
1825 | } else { |
1826 | *DPDE0BytesFrame = 64 * (dml_ceil(((Pitch * (dml_ceil((double) SwathWidth - 1, MacroTileHeight) + MacroTileHeight) * BytePerPixel) - MacroTileSizeBytes) / (8 * 2097152), 1) + 1); |
1827 | } |
1828 | ExtraDPDEBytesFrame = 128 * (mode_lib->vba.GPUVMMaxPageTableLevels - 2); |
1829 | } else { |
1830 | *DPDE0BytesFrame = 0; |
1831 | ExtraDPDEBytesFrame = 0; |
1832 | } |
1833 | |
1834 | PDEAndMetaPTEBytesFrame = *MetaPTEBytesFrame + MPDEBytesFrame + *DPDE0BytesFrame |
1835 | + ExtraDPDEBytesFrame; |
1836 | |
1837 | if (HostVMEnable == true1) { |
1838 | PDEAndMetaPTEBytesFrame = PDEAndMetaPTEBytesFrame * (1 + 8 * HostVMDynamicLevels); |
1839 | } |
1840 | |
1841 | if (SurfaceTiling == dm_sw_linear) { |
1842 | PixelPTEReqHeightPTEs = 1; |
1843 | *PixelPTEReqHeight = 1; |
1844 | *PixelPTEReqWidth = 32768.0 / BytePerPixel; |
1845 | *PTERequestSize = 64; |
1846 | FractionOfPTEReturnDrop = 0; |
1847 | } else if (MacroTileSizeBytes == 4096) { |
1848 | PixelPTEReqHeightPTEs = 1; |
1849 | *PixelPTEReqHeight = MacroTileHeight; |
1850 | *PixelPTEReqWidth = 8 * *MacroTileWidth; |
1851 | *PTERequestSize = 64; |
1852 | if (ScanDirection != dm_vert) |
1853 | FractionOfPTEReturnDrop = 0; |
1854 | else |
1855 | FractionOfPTEReturnDrop = 7 / 8; |
1856 | } else if (GPUVMMinPageSize == 4 && MacroTileSizeBytes > 4096) { |
1857 | PixelPTEReqHeightPTEs = 16; |
1858 | *PixelPTEReqHeight = 16 * BlockHeight256Bytes; |
1859 | *PixelPTEReqWidth = 16 * BlockWidth256Bytes; |
1860 | *PTERequestSize = 128; |
1861 | FractionOfPTEReturnDrop = 0; |
1862 | } else { |
1863 | PixelPTEReqHeightPTEs = 1; |
1864 | *PixelPTEReqHeight = MacroTileHeight; |
1865 | *PixelPTEReqWidth = 8 * *MacroTileWidth; |
1866 | *PTERequestSize = 64; |
1867 | FractionOfPTEReturnDrop = 0; |
1868 | } |
1869 | |
1870 | if (SurfaceTiling == dm_sw_linear) { |
1871 | *dpte_row_height = dml_min(128, 1 << (unsigned int) dml_floor(dml_log2(PTEBufferSizeInRequests * *PixelPTEReqWidth / Pitch), 1)); |
1872 | *dpte_row_width_ub = (dml_ceil(((double) SwathWidth - 1) / *PixelPTEReqWidth, 1) + 1) * *PixelPTEReqWidth; |
1873 | *PixelPTEBytesPerRow = *dpte_row_width_ub / *PixelPTEReqWidth * *PTERequestSize; |
1874 | } else if (ScanDirection != dm_vert) { |
1875 | *dpte_row_height = *PixelPTEReqHeight; |
1876 | *dpte_row_width_ub = (dml_ceil((double) (SwathWidth - 1) / *PixelPTEReqWidth, 1) + 1) * *PixelPTEReqWidth; |
1877 | *PixelPTEBytesPerRow = *dpte_row_width_ub / *PixelPTEReqWidth * *PTERequestSize; |
1878 | } else { |
1879 | *dpte_row_height = dml_min(*PixelPTEReqWidth, *MacroTileWidth); |
1880 | *dpte_row_width_ub = (dml_ceil((double) (SwathWidth - 1) / *PixelPTEReqHeight, 1) + 1) * *PixelPTEReqHeight; |
1881 | *PixelPTEBytesPerRow = *dpte_row_width_ub / *PixelPTEReqHeight * *PTERequestSize; |
1882 | } |
1883 | if (*PixelPTEBytesPerRow * (1 - FractionOfPTEReturnDrop) |
1884 | <= 64 * PTEBufferSizeInRequests) { |
1885 | *PTEBufferSizeNotExceeded = true1; |
1886 | } else { |
1887 | *PTEBufferSizeNotExceeded = false0; |
1888 | } |
1889 | |
1890 | if (GPUVMEnable != true1) { |
1891 | *PixelPTEBytesPerRow = 0; |
1892 | *PTEBufferSizeNotExceeded = true1; |
1893 | } |
1894 | dml_print("DML: vm_bytes = meta_pte_bytes_per_frame (per_pipe) = MetaPTEBytesFrame = : %i\n", *MetaPTEBytesFrame){do { } while(0); }; |
1895 | |
1896 | if (HostVMEnable == true1) { |
1897 | *PixelPTEBytesPerRow = *PixelPTEBytesPerRow * (1 + 8 * HostVMDynamicLevels); |
1898 | } |
1899 | |
1900 | if (HostVMEnable == true1) { |
1901 | *vm_group_bytes = 512; |
1902 | *dpte_group_bytes = 512; |
1903 | } else if (GPUVMEnable == true1) { |
1904 | *vm_group_bytes = 2048; |
1905 | if (SurfaceTiling != dm_sw_linear && PixelPTEReqHeightPTEs == 1 && ScanDirection == dm_vert) { |
1906 | *dpte_group_bytes = 512; |
1907 | } else { |
1908 | *dpte_group_bytes = 2048; |
1909 | } |
1910 | } else { |
1911 | *vm_group_bytes = 0; |
1912 | *dpte_group_bytes = 0; |
1913 | } |
1914 | |
1915 | return PDEAndMetaPTEBytesFrame; |
1916 | } |
1917 | |
1918 | static void DISPCLKDPPCLKDCFCLKDeepSleepPrefetchParametersWatermarksAndPerformanceCalculation( |
1919 | struct display_mode_lib *mode_lib) |
1920 | { |
1921 | struct vba_vars_st *v = &mode_lib->vba; |
1922 | unsigned int j, k; |
1923 | long ReorderBytes = 0; |
1924 | unsigned int PrefetchMode = v->PrefetchModePerState[v->VoltageLevel][v->maxMpcComb]; |
1925 | double MaxTotalRDBandwidth = 0; |
1926 | double MaxTotalRDBandwidthNoUrgentBurst = 0; |
1927 | bool_Bool DestinationLineTimesForPrefetchLessThan2 = false0; |
1928 | bool_Bool VRatioPrefetchMoreThan4 = false0; |
1929 | double TWait; |
1930 | |
1931 | v->WritebackDISPCLK = 0.0; |
1932 | v->DISPCLKWithRamping = 0; |
1933 | v->DISPCLKWithoutRamping = 0; |
1934 | v->GlobalDPPCLK = 0.0; |
1935 | /* DAL custom code: need to update ReturnBW in case min dcfclk is overriden */ |
1936 | v->IdealSDPPortBandwidthPerState[v->VoltageLevel][v->maxMpcComb] = dml_min3( |
1937 | v->ReturnBusWidth * v->DCFCLK, |
1938 | v->DRAMSpeedPerState[v->VoltageLevel] * v->NumberOfChannels * v->DRAMChannelWidth, |
1939 | v->FabricClockPerState[v->VoltageLevel] * v->FabricDatapathToDCNDataReturn); |
1940 | if (v->HostVMEnable != true1) { |
1941 | v->ReturnBW = v->IdealSDPPortBandwidthPerState[v->VoltageLevel][v->maxMpcComb] * v->PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelDataOnly / 100; |
1942 | } else { |
1943 | v->ReturnBW = v->IdealSDPPortBandwidthPerState[v->VoltageLevel][v->maxMpcComb] * v->PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData / 100; |
1944 | } |
1945 | /* End DAL custom code */ |
1946 | |
1947 | // DISPCLK and DPPCLK Calculation |
1948 | // |
1949 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
1950 | if (v->WritebackEnable[k]) { |
1951 | v->WritebackDISPCLK = dml_max(v->WritebackDISPCLK, |
1952 | dml30_CalculateWriteBackDISPCLK( |
1953 | v->WritebackPixelFormat[k], |
1954 | v->PixelClock[k], |
1955 | v->WritebackHRatio[k], |
1956 | v->WritebackVRatio[k], |
1957 | v->WritebackHTaps[k], |
1958 | v->WritebackVTaps[k], |
1959 | v->WritebackSourceWidth[k], |
1960 | v->WritebackDestinationWidth[k], |
1961 | v->HTotal[k], |
1962 | v->WritebackLineBufferSize)); |
1963 | } |
1964 | } |
1965 | |
1966 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
1967 | if (v->HRatio[k] > 1) { |
1968 | v->PSCL_THROUGHPUT_LUMA[k] = dml_min(v->MaxDCHUBToPSCLThroughput, |
1969 | v->MaxPSCLToLBThroughput * v->HRatio[k] / dml_ceil(v->htaps[k] / 6.0, 1)); |
1970 | } else { |
1971 | v->PSCL_THROUGHPUT_LUMA[k] = dml_min( |
1972 | v->MaxDCHUBToPSCLThroughput, |
1973 | v->MaxPSCLToLBThroughput); |
1974 | } |
1975 | |
1976 | v->DPPCLKUsingSingleDPPLuma = v->PixelClock[k] |
1977 | * dml_max(v->vtaps[k] / 6.0 * dml_min(1.0, v->HRatio[k]), |
1978 | dml_max(v->HRatio[k] * v->VRatio[k] / v->PSCL_THROUGHPUT_LUMA[k], 1.0)); |
1979 | |
1980 | if ((v->htaps[k] > 6 || v->vtaps[k] > 6) |
1981 | && v->DPPCLKUsingSingleDPPLuma < 2 * v->PixelClock[k]) { |
1982 | v->DPPCLKUsingSingleDPPLuma = 2 * v->PixelClock[k]; |
1983 | } |
1984 | |
1985 | if ((v->SourcePixelFormat[k] != dm_420_8 |
1986 | && v->SourcePixelFormat[k] != dm_420_10 |
1987 | && v->SourcePixelFormat[k] != dm_420_12 |
1988 | && v->SourcePixelFormat[k] != dm_rgbe_alpha)) { |
1989 | v->PSCL_THROUGHPUT_CHROMA[k] = 0.0; |
1990 | v->DPPCLKUsingSingleDPP[k] = v->DPPCLKUsingSingleDPPLuma; |
1991 | } else { |
1992 | if (v->HRatioChroma[k] > 1) { |
1993 | v->PSCL_THROUGHPUT_CHROMA[k] = dml_min(v->MaxDCHUBToPSCLThroughput, |
1994 | v->MaxPSCLToLBThroughput * v->HRatioChroma[k] / dml_ceil(v->HTAPsChroma[k] / 6.0, 1.0)); |
1995 | } else { |
1996 | v->PSCL_THROUGHPUT_CHROMA[k] = dml_min( |
1997 | v->MaxDCHUBToPSCLThroughput, |
1998 | v->MaxPSCLToLBThroughput); |
1999 | } |
2000 | v->DPPCLKUsingSingleDPPChroma = v->PixelClock[k] |
2001 | * dml_max3(v->VTAPsChroma[k] / 6.0 * dml_min(1.0, v->HRatioChroma[k]), |
2002 | v->HRatioChroma[k] * v->VRatioChroma[k] / v->PSCL_THROUGHPUT_CHROMA[k], 1.0); |
2003 | |
2004 | if ((v->HTAPsChroma[k] > 6 || v->VTAPsChroma[k] > 6) |
2005 | && v->DPPCLKUsingSingleDPPChroma |
2006 | < 2 * v->PixelClock[k]) { |
2007 | v->DPPCLKUsingSingleDPPChroma = 2 |
2008 | * v->PixelClock[k]; |
2009 | } |
2010 | |
2011 | v->DPPCLKUsingSingleDPP[k] = dml_max( |
2012 | v->DPPCLKUsingSingleDPPLuma, |
2013 | v->DPPCLKUsingSingleDPPChroma); |
2014 | } |
2015 | } |
2016 | |
2017 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
2018 | if (v->BlendingAndTiming[k] != k) |
2019 | continue; |
2020 | if (v->ODMCombineEnabled[k] == dm_odm_combine_mode_4to1) { |
2021 | v->DISPCLKWithRamping = dml_max(v->DISPCLKWithRamping, |
2022 | v->PixelClock[k] / 4 * (1 + v->DISPCLKDPPCLKDSCCLKDownSpreading / 100) |
2023 | * (1 + v->DISPCLKRampingMargin / 100)); |
2024 | v->DISPCLKWithoutRamping = dml_max(v->DISPCLKWithoutRamping, |
2025 | v->PixelClock[k] / 4 * (1 + v->DISPCLKDPPCLKDSCCLKDownSpreading / 100)); |
2026 | } else if (v->ODMCombineEnabled[k] == dm_odm_combine_mode_2to1) { |
2027 | v->DISPCLKWithRamping = dml_max(v->DISPCLKWithRamping, |
2028 | v->PixelClock[k] / 2 * (1 + v->DISPCLKDPPCLKDSCCLKDownSpreading / 100) |
2029 | * (1 + v->DISPCLKRampingMargin / 100)); |
2030 | v->DISPCLKWithoutRamping = dml_max(v->DISPCLKWithoutRamping, |
2031 | v->PixelClock[k] / 2 * (1 + v->DISPCLKDPPCLKDSCCLKDownSpreading / 100)); |
2032 | } else { |
2033 | v->DISPCLKWithRamping = dml_max(v->DISPCLKWithRamping, |
2034 | v->PixelClock[k] * (1 + v->DISPCLKDPPCLKDSCCLKDownSpreading / 100) |
2035 | * (1 + v->DISPCLKRampingMargin / 100)); |
2036 | v->DISPCLKWithoutRamping = dml_max(v->DISPCLKWithoutRamping, |
2037 | v->PixelClock[k] * (1 + v->DISPCLKDPPCLKDSCCLKDownSpreading / 100)); |
2038 | } |
2039 | } |
2040 | |
2041 | v->DISPCLKWithRamping = dml_max( |
2042 | v->DISPCLKWithRamping, |
2043 | v->WritebackDISPCLK); |
2044 | v->DISPCLKWithoutRamping = dml_max( |
2045 | v->DISPCLKWithoutRamping, |
2046 | v->WritebackDISPCLK); |
2047 | |
2048 | ASSERT(v->DISPCLKDPPCLKVCOSpeed != 0)do { if (({ static int __warned; int __ret = !!(!(v->DISPCLKDPPCLKVCOSpeed != 0)); if (__ret && !__warned) { printf("WARNING %s failed at %s:%d\n" , "!(v->DISPCLKDPPCLKVCOSpeed != 0)", "/usr/src/sys/dev/pci/drm/amd/display/dc/dml/dcn30/display_mode_vba_30.c" , 2048); __warned = 1; } __builtin_expect(!!(__ret), 0); })) do {} while (0); } while (0); |
2049 | v->DISPCLKWithRampingRoundedToDFSGranularity = RoundToDFSGranularityUp( |
2050 | v->DISPCLKWithRamping, |
2051 | v->DISPCLKDPPCLKVCOSpeed); |
2052 | v->DISPCLKWithoutRampingRoundedToDFSGranularity = RoundToDFSGranularityUp( |
2053 | v->DISPCLKWithoutRamping, |
2054 | v->DISPCLKDPPCLKVCOSpeed); |
2055 | v->MaxDispclkRoundedToDFSGranularity = RoundToDFSGranularityDown( |
2056 | v->soc.clock_limits[mode_lib->soc.num_states - 1].dispclk_mhz, |
2057 | v->DISPCLKDPPCLKVCOSpeed); |
2058 | if (v->DISPCLKWithoutRampingRoundedToDFSGranularity |
2059 | > v->MaxDispclkRoundedToDFSGranularity) { |
2060 | v->DISPCLK_calculated = |
2061 | v->DISPCLKWithoutRampingRoundedToDFSGranularity; |
2062 | } else if (v->DISPCLKWithRampingRoundedToDFSGranularity |
2063 | > v->MaxDispclkRoundedToDFSGranularity) { |
2064 | v->DISPCLK_calculated = v->MaxDispclkRoundedToDFSGranularity; |
2065 | } else { |
2066 | v->DISPCLK_calculated = |
2067 | v->DISPCLKWithRampingRoundedToDFSGranularity; |
2068 | } |
2069 | v->DISPCLK = v->DISPCLK_calculated; |
2070 | DTRACE(" dispclk_mhz (calculated) = %f", v->DISPCLK_calculated){do { } while(0); }; |
2071 | |
2072 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
2073 | v->DPPCLK_calculated[k] = v->DPPCLKUsingSingleDPP[k] |
2074 | / v->DPPPerPlane[k] |
2075 | * (1 + v->DISPCLKDPPCLKDSCCLKDownSpreading / 100); |
2076 | v->GlobalDPPCLK = dml_max( |
2077 | v->GlobalDPPCLK, |
2078 | v->DPPCLK_calculated[k]); |
2079 | } |
2080 | v->GlobalDPPCLK = RoundToDFSGranularityUp( |
2081 | v->GlobalDPPCLK, |
2082 | v->DISPCLKDPPCLKVCOSpeed); |
2083 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
2084 | v->DPPCLK_calculated[k] = v->GlobalDPPCLK / 255 |
2085 | * dml_ceil( |
2086 | v->DPPCLK_calculated[k] * 255.0 |
2087 | / v->GlobalDPPCLK, |
2088 | 1); |
2089 | DTRACE(" dppclk_mhz[%i] (calculated) = %f", k, v->DPPCLK_calculated[k]){do { } while(0); }; |
2090 | v->DPPCLK[k] = v->DPPCLK_calculated[k]; |
2091 | } |
2092 | |
2093 | // Urgent and B P-State/DRAM Clock Change Watermark |
2094 | DTRACE(" dcfclk_mhz = %f", v->DCFCLK){do { } while(0); }; |
2095 | DTRACE(" return_bus_bw = %f", v->ReturnBW){do { } while(0); }; |
2096 | |
2097 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
2098 | CalculateBytePerPixelAnd256BBlockSizes( |
2099 | v->SourcePixelFormat[k], |
2100 | v->SurfaceTiling[k], |
2101 | &v->BytePerPixelY[k], |
2102 | &v->BytePerPixelC[k], |
2103 | &v->BytePerPixelDETY[k], |
2104 | &v->BytePerPixelDETC[k], |
2105 | &v->BlockHeight256BytesY[k], |
2106 | &v->BlockHeight256BytesC[k], |
2107 | &v->BlockWidth256BytesY[k], |
2108 | &v->BlockWidth256BytesC[k]); |
2109 | } |
2110 | |
2111 | CalculateSwathWidth( |
2112 | false0, |
2113 | v->NumberOfActivePlanes, |
2114 | v->SourcePixelFormat, |
2115 | v->SourceScan, |
2116 | v->ViewportWidth, |
2117 | v->ViewportHeight, |
2118 | v->SurfaceWidthY, |
2119 | v->SurfaceWidthC, |
2120 | v->SurfaceHeightY, |
2121 | v->SurfaceHeightC, |
2122 | v->ODMCombineEnabled, |
2123 | v->BytePerPixelY, |
2124 | v->BytePerPixelC, |
2125 | v->BlockHeight256BytesY, |
2126 | v->BlockHeight256BytesC, |
2127 | v->BlockWidth256BytesY, |
2128 | v->BlockWidth256BytesC, |
2129 | v->BlendingAndTiming, |
2130 | v->HActive, |
2131 | v->HRatio, |
2132 | v->DPPPerPlane, |
2133 | v->SwathWidthSingleDPPY, |
2134 | v->SwathWidthSingleDPPC, |
2135 | v->SwathWidthY, |
2136 | v->SwathWidthC, |
2137 | v->dummyinteger3, |
2138 | v->dummyinteger4, |
2139 | v->swath_width_luma_ub, |
2140 | v->swath_width_chroma_ub); |
2141 | |
2142 | |
2143 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
2144 | v->ReadBandwidthPlaneLuma[k] = v->SwathWidthSingleDPPY[k] * v->BytePerPixelY[k] / (v->HTotal[k] / v->PixelClock[k]) * v->VRatio[k]; |
2145 | v->ReadBandwidthPlaneChroma[k] = v->SwathWidthSingleDPPC[k] * v->BytePerPixelC[k] / (v->HTotal[k] / v->PixelClock[k]) * v->VRatioChroma[k]; |
2146 | DTRACE("read_bw[%i] = %fBps", k, v->ReadBandwidthPlaneLuma[k] + v->ReadBandwidthPlaneChroma[k]){do { } while(0); }; |
2147 | } |
2148 | |
2149 | |
2150 | // DCFCLK Deep Sleep |
2151 | CalculateDCFCLKDeepSleep( |
2152 | mode_lib, |
2153 | v->NumberOfActivePlanes, |
2154 | v->BytePerPixelY, |
2155 | v->BytePerPixelC, |
2156 | v->VRatio, |
2157 | v->VRatioChroma, |
2158 | v->SwathWidthY, |
2159 | v->SwathWidthC, |
2160 | v->DPPPerPlane, |
2161 | v->HRatio, |
2162 | v->HRatioChroma, |
2163 | v->PixelClock, |
2164 | v->PSCL_THROUGHPUT_LUMA, |
2165 | v->PSCL_THROUGHPUT_CHROMA, |
2166 | v->DPPCLK, |
2167 | v->ReadBandwidthPlaneLuma, |
2168 | v->ReadBandwidthPlaneChroma, |
2169 | v->ReturnBusWidth, |
2170 | &v->DCFCLKDeepSleep); |
2171 | |
2172 | // DSCCLK |
2173 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
2174 | if ((v->BlendingAndTiming[k] != k) || !v->DSCEnabled[k]) { |
2175 | v->DSCCLK_calculated[k] = 0.0; |
2176 | } else { |
2177 | if (v->OutputFormat[k] == dm_420) |
2178 | v->DSCFormatFactor = 2; |
2179 | else if (v->OutputFormat[k] == dm_444) |
2180 | v->DSCFormatFactor = 1; |
2181 | else if (v->OutputFormat[k] == dm_n422) |
2182 | v->DSCFormatFactor = 2; |
2183 | else |
2184 | v->DSCFormatFactor = 1; |
2185 | if (v->ODMCombineEnabled[k] == dm_odm_combine_mode_4to1) |
2186 | v->DSCCLK_calculated[k] = v->PixelClockBackEnd[k] / 12 |
2187 | / v->DSCFormatFactor / (1 - v->DISPCLKDPPCLKDSCCLKDownSpreading / 100); |
2188 | else if (v->ODMCombineEnabled[k] == dm_odm_combine_mode_2to1) |
2189 | v->DSCCLK_calculated[k] = v->PixelClockBackEnd[k] / 6 |
2190 | / v->DSCFormatFactor / (1 - v->DISPCLKDPPCLKDSCCLKDownSpreading / 100); |
2191 | else |
2192 | v->DSCCLK_calculated[k] = v->PixelClockBackEnd[k] / 3 |
2193 | / v->DSCFormatFactor / (1 - v->DISPCLKDPPCLKDSCCLKDownSpreading / 100); |
2194 | } |
2195 | } |
2196 | |
2197 | // DSC Delay |
2198 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
2199 | double BPP = v->OutputBppPerState[k][v->VoltageLevel]; |
2200 | |
2201 | if (v->DSCEnabled[k] && BPP != 0) { |
2202 | if (v->ODMCombineEnabled[k] == dm_odm_combine_mode_disabled) { |
2203 | v->DSCDelay[k] = dscceComputeDelay(v->DSCInputBitPerComponent[k], |
2204 | BPP, |
2205 | dml_ceil((double) v->HActive[k] / v->NumberOfDSCSlices[k], 1), |
2206 | v->NumberOfDSCSlices[k], |
2207 | v->OutputFormat[k], |
2208 | v->Output[k]) |
2209 | + dscComputeDelay(v->OutputFormat[k], v->Output[k]); |
2210 | } else if (v->ODMCombineEnabled[k] == dm_odm_combine_mode_2to1) { |
2211 | v->DSCDelay[k] = 2 * dscceComputeDelay(v->DSCInputBitPerComponent[k], |
2212 | BPP, |
2213 | dml_ceil((double) v->HActive[k] / v->NumberOfDSCSlices[k], 1), |
2214 | v->NumberOfDSCSlices[k] / 2.0, |
2215 | v->OutputFormat[k], |
2216 | v->Output[k]) |
2217 | + dscComputeDelay(v->OutputFormat[k], v->Output[k]); |
2218 | } else { |
2219 | v->DSCDelay[k] = 4 * dscceComputeDelay(v->DSCInputBitPerComponent[k], |
2220 | BPP, |
2221 | dml_ceil((double) v->HActive[k] / v->NumberOfDSCSlices[k], 1), |
2222 | v->NumberOfDSCSlices[k] / 4.0, |
2223 | v->OutputFormat[k], |
2224 | v->Output[k]) |
2225 | + dscComputeDelay(v->OutputFormat[k], v->Output[k]); |
2226 | } |
2227 | v->DSCDelay[k] = v->DSCDelay[k] * v->PixelClock[k] / v->PixelClockBackEnd[k]; |
2228 | } else { |
2229 | v->DSCDelay[k] = 0; |
2230 | } |
2231 | } |
2232 | |
2233 | for (k = 0; k < v->NumberOfActivePlanes; ++k) |
2234 | for (j = 0; j < v->NumberOfActivePlanes; ++j) // NumberOfPlanes |
2235 | if (j != k && v->BlendingAndTiming[k] == j |
2236 | && v->DSCEnabled[j]) |
2237 | v->DSCDelay[k] = v->DSCDelay[j]; |
2238 | |
2239 | // Prefetch |
2240 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
2241 | unsigned int PDEAndMetaPTEBytesFrameY = 0; |
2242 | unsigned int PixelPTEBytesPerRowY = 0; |
2243 | unsigned int MetaRowByteY = 0; |
2244 | unsigned int MetaRowByteC = 0; |
2245 | unsigned int PDEAndMetaPTEBytesFrameC = 0; |
2246 | unsigned int PixelPTEBytesPerRowC = 0; |
2247 | bool_Bool PTEBufferSizeNotExceededY = 0; |
2248 | bool_Bool PTEBufferSizeNotExceededC = 0; |
2249 | |
2250 | |
2251 | if (v->SourcePixelFormat[k] == dm_420_8 || v->SourcePixelFormat[k] == dm_420_10 || v->SourcePixelFormat[k] == dm_420_12 || v->SourcePixelFormat[k] == dm_rgbe_alpha) { |
2252 | if ((v->SourcePixelFormat[k] == dm_420_10 || v->SourcePixelFormat[k] == dm_420_12) && v->SourceScan[k] != dm_vert) { |
2253 | v->PTEBufferSizeInRequestsForLuma = (v->PTEBufferSizeInRequestsLuma + v->PTEBufferSizeInRequestsChroma) / 2; |
2254 | v->PTEBufferSizeInRequestsForChroma = v->PTEBufferSizeInRequestsForLuma; |
2255 | } else { |
2256 | v->PTEBufferSizeInRequestsForLuma = v->PTEBufferSizeInRequestsLuma; |
2257 | v->PTEBufferSizeInRequestsForChroma = v->PTEBufferSizeInRequestsChroma; |
2258 | |
2259 | } |
2260 | PDEAndMetaPTEBytesFrameC = CalculateVMAndRowBytes( |
2261 | mode_lib, |
2262 | v->DCCEnable[k], |
2263 | v->BlockHeight256BytesC[k], |
2264 | v->BlockWidth256BytesC[k], |
2265 | v->SourcePixelFormat[k], |
2266 | v->SurfaceTiling[k], |
2267 | v->BytePerPixelC[k], |
2268 | v->SourceScan[k], |
2269 | v->SwathWidthC[k], |
2270 | v->ViewportHeightChroma[k], |
2271 | v->GPUVMEnable, |
2272 | v->HostVMEnable, |
2273 | v->HostVMMaxNonCachedPageTableLevels, |
2274 | v->GPUVMMinPageSize, |
2275 | v->HostVMMinPageSize, |
2276 | v->PTEBufferSizeInRequestsForChroma, |
2277 | v->PitchC[k], |
2278 | v->DCCMetaPitchC[k], |
2279 | &v->MacroTileWidthC[k], |
2280 | &MetaRowByteC, |
2281 | &PixelPTEBytesPerRowC, |
2282 | &PTEBufferSizeNotExceededC, |
2283 | &v->dpte_row_width_chroma_ub[k], |
2284 | &v->dpte_row_height_chroma[k], |
2285 | &v->meta_req_width_chroma[k], |
2286 | &v->meta_req_height_chroma[k], |
2287 | &v->meta_row_width_chroma[k], |
2288 | &v->meta_row_height_chroma[k], |
2289 | &v->dummyinteger1, |
2290 | &v->dummyinteger2, |
2291 | &v->PixelPTEReqWidthC[k], |
2292 | &v->PixelPTEReqHeightC[k], |
2293 | &v->PTERequestSizeC[k], |
2294 | &v->dpde0_bytes_per_frame_ub_c[k], |
2295 | &v->meta_pte_bytes_per_frame_ub_c[k]); |
2296 | |
2297 | v->PrefetchSourceLinesC[k] = CalculatePrefetchSourceLines( |
2298 | mode_lib, |
2299 | v->VRatioChroma[k], |
2300 | v->VTAPsChroma[k], |
2301 | v->Interlace[k], |
2302 | v->ProgressiveToInterlaceUnitInOPP, |
2303 | v->SwathHeightC[k], |
2304 | v->ViewportYStartC[k], |
2305 | &v->VInitPreFillC[k], |
2306 | &v->MaxNumSwathC[k]); |
2307 | } else { |
2308 | v->PTEBufferSizeInRequestsForLuma = v->PTEBufferSizeInRequestsLuma + v->PTEBufferSizeInRequestsChroma; |
2309 | v->PTEBufferSizeInRequestsForChroma = 0; |
2310 | PixelPTEBytesPerRowC = 0; |
2311 | PDEAndMetaPTEBytesFrameC = 0; |
2312 | MetaRowByteC = 0; |
2313 | v->MaxNumSwathC[k] = 0; |
2314 | v->PrefetchSourceLinesC[k] = 0; |
2315 | } |
2316 | |
2317 | PDEAndMetaPTEBytesFrameY = CalculateVMAndRowBytes( |
2318 | mode_lib, |
2319 | v->DCCEnable[k], |
2320 | v->BlockHeight256BytesY[k], |
2321 | v->BlockWidth256BytesY[k], |
2322 | v->SourcePixelFormat[k], |
2323 | v->SurfaceTiling[k], |
2324 | v->BytePerPixelY[k], |
2325 | v->SourceScan[k], |
2326 | v->SwathWidthY[k], |
2327 | v->ViewportHeight[k], |
2328 | v->GPUVMEnable, |
2329 | v->HostVMEnable, |
2330 | v->HostVMMaxNonCachedPageTableLevels, |
2331 | v->GPUVMMinPageSize, |
2332 | v->HostVMMinPageSize, |
2333 | v->PTEBufferSizeInRequestsForLuma, |
2334 | v->PitchY[k], |
2335 | v->DCCMetaPitchY[k], |
2336 | &v->MacroTileWidthY[k], |
2337 | &MetaRowByteY, |
2338 | &PixelPTEBytesPerRowY, |
2339 | &PTEBufferSizeNotExceededY, |
2340 | &v->dpte_row_width_luma_ub[k], |
2341 | &v->dpte_row_height[k], |
2342 | &v->meta_req_width[k], |
2343 | &v->meta_req_height[k], |
2344 | &v->meta_row_width[k], |
2345 | &v->meta_row_height[k], |
2346 | &v->vm_group_bytes[k], |
2347 | &v->dpte_group_bytes[k], |
2348 | &v->PixelPTEReqWidthY[k], |
2349 | &v->PixelPTEReqHeightY[k], |
2350 | &v->PTERequestSizeY[k], |
2351 | &v->dpde0_bytes_per_frame_ub_l[k], |
2352 | &v->meta_pte_bytes_per_frame_ub_l[k]); |
2353 | |
2354 | v->PrefetchSourceLinesY[k] = CalculatePrefetchSourceLines( |
2355 | mode_lib, |
2356 | v->VRatio[k], |
2357 | v->vtaps[k], |
2358 | v->Interlace[k], |
2359 | v->ProgressiveToInterlaceUnitInOPP, |
2360 | v->SwathHeightY[k], |
2361 | v->ViewportYStartY[k], |
2362 | &v->VInitPreFillY[k], |
2363 | &v->MaxNumSwathY[k]); |
2364 | v->PixelPTEBytesPerRow[k] = PixelPTEBytesPerRowY + PixelPTEBytesPerRowC; |
2365 | v->PDEAndMetaPTEBytesFrame[k] = PDEAndMetaPTEBytesFrameY |
2366 | + PDEAndMetaPTEBytesFrameC; |
2367 | v->MetaRowByte[k] = MetaRowByteY + MetaRowByteC; |
2368 | |
2369 | CalculateRowBandwidth( |
2370 | v->GPUVMEnable, |
2371 | v->SourcePixelFormat[k], |
2372 | v->VRatio[k], |
2373 | v->VRatioChroma[k], |
2374 | v->DCCEnable[k], |
2375 | v->HTotal[k] / v->PixelClock[k], |
2376 | MetaRowByteY, |
2377 | MetaRowByteC, |
2378 | v->meta_row_height[k], |
2379 | v->meta_row_height_chroma[k], |
2380 | PixelPTEBytesPerRowY, |
2381 | PixelPTEBytesPerRowC, |
2382 | v->dpte_row_height[k], |
2383 | v->dpte_row_height_chroma[k], |
2384 | &v->meta_row_bw[k], |
2385 | &v->dpte_row_bw[k]); |
2386 | } |
2387 | |
2388 | v->TotalDCCActiveDPP = 0; |
2389 | v->TotalActiveDPP = 0; |
2390 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
2391 | v->TotalActiveDPP = v->TotalActiveDPP |
2392 | + v->DPPPerPlane[k]; |
2393 | if (v->DCCEnable[k]) |
2394 | v->TotalDCCActiveDPP = v->TotalDCCActiveDPP |
2395 | + v->DPPPerPlane[k]; |
2396 | } |
2397 | |
2398 | |
2399 | ReorderBytes = v->NumberOfChannels * dml_max3( |
2400 | v->UrgentOutOfOrderReturnPerChannelPixelDataOnly, |
2401 | v->UrgentOutOfOrderReturnPerChannelPixelMixedWithVMData, |
2402 | v->UrgentOutOfOrderReturnPerChannelVMDataOnly); |
2403 | |
2404 | v->UrgentExtraLatency = CalculateExtraLatency( |
2405 | v->RoundTripPingLatencyCycles, |
2406 | ReorderBytes, |
2407 | v->DCFCLK, |
2408 | v->TotalActiveDPP, |
2409 | v->PixelChunkSizeInKByte, |
2410 | v->TotalDCCActiveDPP, |
2411 | v->MetaChunkSize, |
2412 | v->ReturnBW, |
2413 | v->GPUVMEnable, |
2414 | v->HostVMEnable, |
2415 | v->NumberOfActivePlanes, |
2416 | v->DPPPerPlane, |
2417 | v->dpte_group_bytes, |
2418 | v->PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData, |
2419 | v->PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly, |
2420 | v->HostVMMinPageSize, |
2421 | v->HostVMMaxNonCachedPageTableLevels); |
2422 | |
2423 | v->TCalc = 24.0 / v->DCFCLKDeepSleep; |
2424 | |
2425 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
2426 | if (v->BlendingAndTiming[k] == k) { |
2427 | if (v->WritebackEnable[k] == true1) { |
2428 | v->WritebackDelay[v->VoltageLevel][k] = v->WritebackLatency + |
2429 | CalculateWriteBackDelay(v->WritebackPixelFormat[k], |
2430 | v->WritebackHRatio[k], |
2431 | v->WritebackVRatio[k], |
2432 | v->WritebackVTaps[k], |
2433 | v->WritebackDestinationWidth[k], |
2434 | v->WritebackDestinationHeight[k], |
2435 | v->WritebackSourceHeight[k], |
2436 | v->HTotal[k]) / v->DISPCLK; |
2437 | } else |
2438 | v->WritebackDelay[v->VoltageLevel][k] = 0; |
2439 | for (j = 0; j < v->NumberOfActivePlanes; ++j) { |
2440 | if (v->BlendingAndTiming[j] == k |
2441 | && v->WritebackEnable[j] == true1) { |
2442 | v->WritebackDelay[v->VoltageLevel][k] = dml_max(v->WritebackDelay[v->VoltageLevel][k], |
2443 | v->WritebackLatency + CalculateWriteBackDelay( |
2444 | v->WritebackPixelFormat[j], |
2445 | v->WritebackHRatio[j], |
2446 | v->WritebackVRatio[j], |
2447 | v->WritebackVTaps[j], |
2448 | v->WritebackDestinationWidth[j], |
2449 | v->WritebackDestinationHeight[j], |
2450 | v->WritebackSourceHeight[j], |
2451 | v->HTotal[k]) / v->DISPCLK); |
2452 | } |
2453 | } |
2454 | } |
2455 | } |
2456 | |
2457 | for (k = 0; k < v->NumberOfActivePlanes; ++k) |
2458 | for (j = 0; j < v->NumberOfActivePlanes; ++j) |
2459 | if (v->BlendingAndTiming[k] == j) |
2460 | v->WritebackDelay[v->VoltageLevel][k] = v->WritebackDelay[v->VoltageLevel][j]; |
2461 | |
2462 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
2463 | v->MaxVStartupLines[k] = v->VTotal[k] - v->VActive[k] - dml_max(1.0, dml_ceil((double) v->WritebackDelay[v->VoltageLevel][k] / (v->HTotal[k] / v->PixelClock[k]), 1)); |
2464 | } |
2465 | |
2466 | v->MaximumMaxVStartupLines = 0; |
2467 | for (k = 0; k < v->NumberOfActivePlanes; ++k) |
2468 | v->MaximumMaxVStartupLines = dml_max(v->MaximumMaxVStartupLines, v->MaxVStartupLines[k]); |
2469 | |
2470 | if (v->DRAMClockChangeLatencyOverride > 0.0) { |
2471 | v->FinalDRAMClockChangeLatency = v->DRAMClockChangeLatencyOverride; |
2472 | } else { |
2473 | v->FinalDRAMClockChangeLatency = v->DRAMClockChangeLatency; |
2474 | } |
2475 | v->UrgentLatency = CalculateUrgentLatency(v->UrgentLatencyPixelDataOnly, v->UrgentLatencyPixelMixedWithVMData, v->UrgentLatencyVMDataOnly, v->DoUrgentLatencyAdjustment, v->UrgentLatencyAdjustmentFabricClockComponent, v->UrgentLatencyAdjustmentFabricClockReference, v->FabricClock); |
2476 | |
2477 | |
2478 | v->FractionOfUrgentBandwidth = 0.0; |
2479 | v->FractionOfUrgentBandwidthImmediateFlip = 0.0; |
2480 | |
2481 | v->VStartupLines = 13; |
2482 | |
2483 | do { |
2484 | MaxTotalRDBandwidth = 0; |
2485 | MaxTotalRDBandwidthNoUrgentBurst = 0; |
2486 | DestinationLineTimesForPrefetchLessThan2 = false0; |
2487 | VRatioPrefetchMoreThan4 = false0; |
2488 | TWait = CalculateTWait( |
2489 | PrefetchMode, |
2490 | v->FinalDRAMClockChangeLatency, |
2491 | v->UrgentLatency, |
2492 | v->SREnterPlusExitTime); |
2493 | |
2494 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
2495 | Pipe myPipe = { 0 }; |
2496 | |
2497 | myPipe.DPPCLK = v->DPPCLK[k]; |
2498 | myPipe.DISPCLK = v->DISPCLK; |
2499 | myPipe.PixelClock = v->PixelClock[k]; |
2500 | myPipe.DCFCLKDeepSleep = v->DCFCLKDeepSleep; |
2501 | myPipe.DPPPerPlane = v->DPPPerPlane[k]; |
2502 | myPipe.ScalerEnabled = v->ScalerEnabled[k]; |
2503 | myPipe.SourceScan = v->SourceScan[k]; |
2504 | myPipe.BlockWidth256BytesY = v->BlockWidth256BytesY[k]; |
2505 | myPipe.BlockHeight256BytesY = v->BlockHeight256BytesY[k]; |
2506 | myPipe.BlockWidth256BytesC = v->BlockWidth256BytesC[k]; |
2507 | myPipe.BlockHeight256BytesC = v->BlockHeight256BytesC[k]; |
2508 | myPipe.InterlaceEnable = v->Interlace[k]; |
2509 | myPipe.NumberOfCursors = v->NumberOfCursors[k]; |
2510 | myPipe.VBlank = v->VTotal[k] - v->VActive[k]; |
2511 | myPipe.HTotal = v->HTotal[k]; |
2512 | myPipe.DCCEnable = v->DCCEnable[k]; |
2513 | myPipe.ODMCombineEnabled = !!v->ODMCombineEnabled[k]; |
2514 | |
2515 | v->ErrorResult[k] = CalculatePrefetchSchedule( |
2516 | mode_lib, |
2517 | v->PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData, |
2518 | v->PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly, |
2519 | &myPipe, |
2520 | v->DSCDelay[k], |
2521 | v->DPPCLKDelaySubtotal |
2522 | + v->DPPCLKDelayCNVCFormater, |
2523 | v->DPPCLKDelaySCL, |
2524 | v->DPPCLKDelaySCLLBOnly, |
2525 | v->DPPCLKDelayCNVCCursor, |
2526 | v->DISPCLKDelaySubtotal, |
2527 | (unsigned int) (v->SwathWidthY[k] / v->HRatio[k]), |
2528 | v->OutputFormat[k], |
2529 | v->MaxInterDCNTileRepeaters, |
2530 | dml_min(v->VStartupLines, v->MaxVStartupLines[k]), |
2531 | v->MaxVStartupLines[k], |
2532 | v->GPUVMMaxPageTableLevels, |
2533 | v->GPUVMEnable, |
2534 | v->HostVMEnable, |
2535 | v->HostVMMaxNonCachedPageTableLevels, |
2536 | v->HostVMMinPageSize, |
2537 | v->DynamicMetadataEnable[k], |
2538 | v->DynamicMetadataVMEnabled, |
2539 | v->DynamicMetadataLinesBeforeActiveRequired[k], |
2540 | v->DynamicMetadataTransmittedBytes[k], |
2541 | v->UrgentLatency, |
2542 | v->UrgentExtraLatency, |
2543 | v->TCalc, |
2544 | v->PDEAndMetaPTEBytesFrame[k], |
2545 | v->MetaRowByte[k], |
2546 | v->PixelPTEBytesPerRow[k], |
2547 | v->PrefetchSourceLinesY[k], |
2548 | v->SwathWidthY[k], |
2549 | v->BytePerPixelY[k], |
2550 | v->VInitPreFillY[k], |
2551 | v->MaxNumSwathY[k], |
2552 | v->PrefetchSourceLinesC[k], |
2553 | v->SwathWidthC[k], |
2554 | v->BytePerPixelC[k], |
2555 | v->VInitPreFillC[k], |
2556 | v->MaxNumSwathC[k], |
2557 | v->swath_width_luma_ub[k], |
2558 | v->swath_width_chroma_ub[k], |
2559 | v->SwathHeightY[k], |
2560 | v->SwathHeightC[k], |
2561 | TWait, |
2562 | v->ProgressiveToInterlaceUnitInOPP, |
2563 | &v->DSTXAfterScaler[k], |
2564 | &v->DSTYAfterScaler[k], |
2565 | &v->DestinationLinesForPrefetch[k], |
2566 | &v->PrefetchBandwidth[k], |
2567 | &v->DestinationLinesToRequestVMInVBlank[k], |
2568 | &v->DestinationLinesToRequestRowInVBlank[k], |
2569 | &v->VRatioPrefetchY[k], |
2570 | &v->VRatioPrefetchC[k], |
2571 | &v->RequiredPrefetchPixDataBWLuma[k], |
2572 | &v->RequiredPrefetchPixDataBWChroma[k], |
2573 | &v->NotEnoughTimeForDynamicMetadata[k], |
2574 | &v->Tno_bw[k], |
2575 | &v->prefetch_vmrow_bw[k], |
2576 | &v->Tdmdl_vm[k], |
2577 | &v->Tdmdl[k], |
2578 | &v->VUpdateOffsetPix[k], |
2579 | &v->VUpdateWidthPix[k], |
2580 | &v->VReadyOffsetPix[k]); |
2581 | if (v->BlendingAndTiming[k] == k) { |
2582 | double TotalRepeaterDelayTime = v->MaxInterDCNTileRepeaters * (2 / v->DPPCLK[k] + 3 / v->DISPCLK); |
2583 | v->VUpdateWidthPix[k] = (14 / v->DCFCLKDeepSleep + 12 / v->DPPCLK[k] + TotalRepeaterDelayTime) * v->PixelClock[k]; |
2584 | v->VReadyOffsetPix[k] = dml_max(150.0 / v->DPPCLK[k], TotalRepeaterDelayTime + 20 / v->DCFCLKDeepSleep + 10 / v->DPPCLK[k]) * v->PixelClock[k]; |
2585 | v->VUpdateOffsetPix[k] = dml_ceil(v->HTotal[k] / 4.0, 1); |
2586 | v->VStartup[k] = dml_min(v->VStartupLines, v->MaxVStartupLines[k]); |
2587 | } else { |
2588 | int x = v->BlendingAndTiming[k]; |
2589 | double TotalRepeaterDelayTime = v->MaxInterDCNTileRepeaters * (2 / v->DPPCLK[k] + 3 / v->DISPCLK); |
2590 | v->VUpdateWidthPix[k] = (14 / v->DCFCLKDeepSleep + 12 / v->DPPCLK[k] + TotalRepeaterDelayTime) * v->PixelClock[x]; |
2591 | v->VReadyOffsetPix[k] = dml_max(150.0 / v->DPPCLK[k], TotalRepeaterDelayTime + 20 / v->DCFCLKDeepSleep + 10 / v->DPPCLK[k]) * v->PixelClock[x]; |
2592 | v->VUpdateOffsetPix[k] = dml_ceil(v->HTotal[x] / 4.0, 1); |
2593 | if (!v->MaxVStartupLines[x]) |
2594 | v->MaxVStartupLines[x] = v->MaxVStartupLines[k]; |
2595 | v->VStartup[k] = dml_min(v->VStartupLines, v->MaxVStartupLines[x]); |
2596 | } |
2597 | } |
2598 | |
2599 | v->NotEnoughUrgentLatencyHiding = false0; |
2600 | v->NotEnoughUrgentLatencyHidingPre = false0; |
2601 | |
2602 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
2603 | v->cursor_bw[k] = v->NumberOfCursors[k] |
2604 | * v->CursorWidth[k][0] * v->CursorBPP[k][0] |
2605 | / 8.0 |
2606 | / (v->HTotal[k] / v->PixelClock[k]) |
2607 | * v->VRatio[k]; |
2608 | v->cursor_bw_pre[k] = v->NumberOfCursors[k] |
2609 | * v->CursorWidth[k][0] * v->CursorBPP[k][0] |
2610 | / 8.0 |
2611 | / (v->HTotal[k] / v->PixelClock[k]) |
2612 | * v->VRatioPrefetchY[k]; |
2613 | |
2614 | CalculateUrgentBurstFactor( |
2615 | v->swath_width_luma_ub[k], |
2616 | v->swath_width_chroma_ub[k], |
2617 | v->DETBufferSizeInKByte, |
2618 | v->SwathHeightY[k], |
2619 | v->SwathHeightC[k], |
2620 | v->HTotal[k] / v->PixelClock[k], |
2621 | v->UrgentLatency, |
2622 | v->CursorBufferSize, |
2623 | v->CursorWidth[k][0], |
2624 | v->CursorBPP[k][0], |
2625 | v->VRatio[k], |
2626 | v->VRatioChroma[k], |
2627 | v->BytePerPixelDETY[k], |
2628 | v->BytePerPixelDETC[k], |
2629 | v->DETBufferSizeY[k], |
2630 | v->DETBufferSizeC[k], |
2631 | &v->UrgentBurstFactorCursor[k], |
2632 | &v->UrgentBurstFactorLuma[k], |
2633 | &v->UrgentBurstFactorChroma[k], |
2634 | &v->NoUrgentLatencyHiding[k]); |
2635 | |
2636 | CalculateUrgentBurstFactor( |
2637 | v->swath_width_luma_ub[k], |
2638 | v->swath_width_chroma_ub[k], |
2639 | v->DETBufferSizeInKByte, |
2640 | v->SwathHeightY[k], |
2641 | v->SwathHeightC[k], |
2642 | v->HTotal[k] / v->PixelClock[k], |
2643 | v->UrgentLatency, |
2644 | v->CursorBufferSize, |
2645 | v->CursorWidth[k][0], |
2646 | v->CursorBPP[k][0], |
2647 | v->VRatioPrefetchY[k], |
2648 | v->VRatioPrefetchC[k], |
2649 | v->BytePerPixelDETY[k], |
2650 | v->BytePerPixelDETC[k], |
2651 | v->DETBufferSizeY[k], |
2652 | v->DETBufferSizeC[k], |
2653 | &v->UrgentBurstFactorCursorPre[k], |
2654 | &v->UrgentBurstFactorLumaPre[k], |
2655 | &v->UrgentBurstFactorChromaPre[k], |
2656 | &v->NoUrgentLatencyHidingPre[k]); |
2657 | |
2658 | MaxTotalRDBandwidth = MaxTotalRDBandwidth + |
2659 | dml_max3(v->DPPPerPlane[k] * v->prefetch_vmrow_bw[k], |
2660 | v->ReadBandwidthPlaneLuma[k] * |
2661 | v->UrgentBurstFactorLuma[k] + |
2662 | v->ReadBandwidthPlaneChroma[k] * |
2663 | v->UrgentBurstFactorChroma[k] + |
2664 | v->cursor_bw[k] * |
2665 | v->UrgentBurstFactorCursor[k] + |
2666 | v->DPPPerPlane[k] * (v->meta_row_bw[k] + v->dpte_row_bw[k]), |
2667 | v->DPPPerPlane[k] * (v->RequiredPrefetchPixDataBWLuma[k] * v->UrgentBurstFactorLumaPre[k] + |
2668 | v->RequiredPrefetchPixDataBWChroma[k] * v->UrgentBurstFactorChromaPre[k]) + v->cursor_bw_pre[k] * |
2669 | v->UrgentBurstFactorCursorPre[k]); |
2670 | |
2671 | MaxTotalRDBandwidthNoUrgentBurst = MaxTotalRDBandwidthNoUrgentBurst + |
2672 | dml_max3(v->DPPPerPlane[k] * v->prefetch_vmrow_bw[k], |
2673 | v->ReadBandwidthPlaneLuma[k] + |
2674 | v->ReadBandwidthPlaneChroma[k] + |
2675 | v->cursor_bw[k] + |
2676 | v->DPPPerPlane[k] * (v->meta_row_bw[k] + v->dpte_row_bw[k]), |
2677 | v->DPPPerPlane[k] * (v->RequiredPrefetchPixDataBWLuma[k] + v->RequiredPrefetchPixDataBWChroma[k]) + v->cursor_bw_pre[k]); |
2678 | |
2679 | if (v->DestinationLinesForPrefetch[k] < 2) |
2680 | DestinationLineTimesForPrefetchLessThan2 = true1; |
2681 | if (v->VRatioPrefetchY[k] > 4 || v->VRatioPrefetchC[k] > 4) |
2682 | VRatioPrefetchMoreThan4 = true1; |
2683 | if (v->NoUrgentLatencyHiding[k] == true1) |
2684 | v->NotEnoughUrgentLatencyHiding = true1; |
2685 | |
2686 | if (v->NoUrgentLatencyHidingPre[k] == true1) |
2687 | v->NotEnoughUrgentLatencyHidingPre = true1; |
2688 | } |
2689 | v->FractionOfUrgentBandwidth = MaxTotalRDBandwidthNoUrgentBurst / v->ReturnBW; |
2690 | |
2691 | |
2692 | if (MaxTotalRDBandwidth <= v->ReturnBW && v->NotEnoughUrgentLatencyHiding == 0 && v->NotEnoughUrgentLatencyHidingPre == 0 && !VRatioPrefetchMoreThan4 |
2693 | && !DestinationLineTimesForPrefetchLessThan2) |
2694 | v->PrefetchModeSupported = true1; |
2695 | else { |
2696 | v->PrefetchModeSupported = false0; |
2697 | dml_print("DML: CalculatePrefetchSchedule ***failed***. Bandwidth violation. Results are NOT valid\n"){do { } while(0); }; |
2698 | dml_print("DML: MaxTotalRDBandwidth:%f AvailReturnBandwidth:%f\n", MaxTotalRDBandwidth, v->ReturnBW){do { } while(0); }; |
2699 | dml_print("DML: VRatioPrefetch %s more than 4\n", (VRatioPrefetchMoreThan4) ? "is" : "is not"){do { } while(0); }; |
2700 | dml_print("DML: DestinationLines for Prefetch %s less than 2\n", (DestinationLineTimesForPrefetchLessThan2) ? "is" : "is not"){do { } while(0); }; |
2701 | } |
2702 | |
2703 | if (v->PrefetchModeSupported == true1 && v->ImmediateFlipSupport == true1) { |
2704 | v->BandwidthAvailableForImmediateFlip = v->ReturnBW; |
2705 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
2706 | v->BandwidthAvailableForImmediateFlip = |
2707 | v->BandwidthAvailableForImmediateFlip |
2708 | - dml_max( |
2709 | v->ReadBandwidthPlaneLuma[k] * v->UrgentBurstFactorLuma[k] |
2710 | + v->ReadBandwidthPlaneChroma[k] * v->UrgentBurstFactorChroma[k] |
2711 | + v->cursor_bw[k] * v->UrgentBurstFactorCursor[k], |
2712 | v->DPPPerPlane[k] * (v->RequiredPrefetchPixDataBWLuma[k] * v->UrgentBurstFactorLumaPre[k] + |
2713 | v->RequiredPrefetchPixDataBWChroma[k] * v->UrgentBurstFactorChromaPre[k]) + |
2714 | v->cursor_bw_pre[k] * v->UrgentBurstFactorCursorPre[k]); |
2715 | } |
2716 | |
2717 | v->TotImmediateFlipBytes = 0; |
2718 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
2719 | v->TotImmediateFlipBytes = v->TotImmediateFlipBytes + v->DPPPerPlane[k] * (v->PDEAndMetaPTEBytesFrame[k] + v->MetaRowByte[k] + v->PixelPTEBytesPerRow[k]); |
2720 | } |
2721 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
2722 | CalculateFlipSchedule( |
2723 | mode_lib, |
2724 | v->PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData, |
2725 | v->PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly, |
2726 | v->UrgentExtraLatency, |
2727 | v->UrgentLatency, |
2728 | v->GPUVMMaxPageTableLevels, |
2729 | v->HostVMEnable, |
2730 | v->HostVMMaxNonCachedPageTableLevels, |
2731 | v->GPUVMEnable, |
2732 | v->HostVMMinPageSize, |
2733 | v->PDEAndMetaPTEBytesFrame[k], |
2734 | v->MetaRowByte[k], |
2735 | v->PixelPTEBytesPerRow[k], |
2736 | v->BandwidthAvailableForImmediateFlip, |
2737 | v->TotImmediateFlipBytes, |
2738 | v->SourcePixelFormat[k], |
2739 | v->HTotal[k] / v->PixelClock[k], |
2740 | v->VRatio[k], |
2741 | v->VRatioChroma[k], |
2742 | v->Tno_bw[k], |
2743 | v->DCCEnable[k], |
2744 | v->dpte_row_height[k], |
2745 | v->meta_row_height[k], |
2746 | v->dpte_row_height_chroma[k], |
2747 | v->meta_row_height_chroma[k], |
2748 | &v->DestinationLinesToRequestVMInImmediateFlip[k], |
2749 | &v->DestinationLinesToRequestRowInImmediateFlip[k], |
2750 | &v->final_flip_bw[k], |
2751 | &v->ImmediateFlipSupportedForPipe[k]); |
2752 | } |
2753 | v->total_dcn_read_bw_with_flip = 0.0; |
2754 | v->total_dcn_read_bw_with_flip_no_urgent_burst = 0.0; |
2755 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
2756 | v->total_dcn_read_bw_with_flip = v->total_dcn_read_bw_with_flip + dml_max3( |
2757 | v->DPPPerPlane[k] * v->prefetch_vmrow_bw[k], |
2758 | v->DPPPerPlane[k] * v->final_flip_bw[k] + |
2759 | v->ReadBandwidthLuma[k] * v->UrgentBurstFactorLuma[k] + |
2760 | v->ReadBandwidthChroma[k] * v->UrgentBurstFactorChroma[k] + |
2761 | v->cursor_bw[k] * v->UrgentBurstFactorCursor[k], |
2762 | v->DPPPerPlane[k] * (v->final_flip_bw[k] + |
2763 | v->RequiredPrefetchPixDataBWLuma[k] * v->UrgentBurstFactorLumaPre[k] + |
2764 | v->RequiredPrefetchPixDataBWChroma[k] * v->UrgentBurstFactorChromaPre[k]) + |
2765 | v->cursor_bw_pre[k] * v->UrgentBurstFactorCursorPre[k]); |
2766 | v->total_dcn_read_bw_with_flip_no_urgent_burst = |
2767 | v->total_dcn_read_bw_with_flip_no_urgent_burst + |
2768 | dml_max3(v->DPPPerPlane[k] * v->prefetch_vmrow_bw[k], |
2769 | v->DPPPerPlane[k] * v->final_flip_bw[k] + v->ReadBandwidthPlaneLuma[k] + v->ReadBandwidthPlaneChroma[k] + v->cursor_bw[k], |
2770 | v->DPPPerPlane[k] * (v->final_flip_bw[k] + v->RequiredPrefetchPixDataBWLuma[k] + v->RequiredPrefetchPixDataBWChroma[k]) + v->cursor_bw_pre[k]); |
2771 | |
2772 | } |
2773 | v->FractionOfUrgentBandwidthImmediateFlip = v->total_dcn_read_bw_with_flip_no_urgent_burst / v->ReturnBW; |
2774 | |
2775 | v->ImmediateFlipSupported = true1; |
2776 | if (v->total_dcn_read_bw_with_flip > v->ReturnBW) { |
2777 | v->ImmediateFlipSupported = false0; |
2778 | v->total_dcn_read_bw_with_flip = MaxTotalRDBandwidth; |
2779 | } |
2780 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
2781 | if (v->ImmediateFlipSupportedForPipe[k] == false0) { |
2782 | v->ImmediateFlipSupported = false0; |
2783 | } |
2784 | } |
2785 | } else { |
2786 | v->ImmediateFlipSupported = false0; |
2787 | } |
2788 | |
2789 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
2790 | if (v->ErrorResult[k] || v->NotEnoughTimeForDynamicMetadata[k]) { |
2791 | v->PrefetchModeSupported = false0; |
2792 | dml_print("DML: CalculatePrefetchSchedule ***failed***. Prefetch schedule violation. Results are NOT valid\n"){do { } while(0); }; |
2793 | } |
2794 | } |
2795 | |
2796 | v->VStartupLines = v->VStartupLines + 1; |
2797 | v->PrefetchAndImmediateFlipSupported = (v->PrefetchModeSupported == true1 && ((!v->ImmediateFlipSupport && !v->HostVMEnable && v->ImmediateFlipRequirement != dm_immediate_flip_required) || v->ImmediateFlipSupported)) ? true1 : false0; |
2798 | |
2799 | } while (!v->PrefetchModeSupported && v->VStartupLines <= v->MaximumMaxVStartupLines); |
2800 | ASSERT(v->PrefetchModeSupported)do { if (({ static int __warned; int __ret = !!(!(v->PrefetchModeSupported )); if (__ret && !__warned) { printf("WARNING %s failed at %s:%d\n" , "!(v->PrefetchModeSupported)", "/usr/src/sys/dev/pci/drm/amd/display/dc/dml/dcn30/display_mode_vba_30.c" , 2800); __warned = 1; } __builtin_expect(!!(__ret), 0); })) do {} while (0); } while (0); |
2801 | |
2802 | //Watermarks and NB P-State/DRAM Clock Change Support |
2803 | { |
2804 | enum clock_change_support DRAMClockChangeSupport = 0; // dummy |
2805 | CalculateWatermarksAndDRAMSpeedChangeSupport( |
2806 | mode_lib, |
2807 | PrefetchMode, |
2808 | v->NumberOfActivePlanes, |
2809 | v->MaxLineBufferLines, |
2810 | v->LineBufferSize, |
2811 | v->DPPOutputBufferPixels, |
2812 | v->DETBufferSizeInKByte, |
2813 | v->WritebackInterfaceBufferSize, |
2814 | v->DCFCLK, |
2815 | v->ReturnBW, |
2816 | v->GPUVMEnable, |
2817 | v->dpte_group_bytes, |
2818 | v->MetaChunkSize, |
2819 | v->UrgentLatency, |
2820 | v->UrgentExtraLatency, |
2821 | v->WritebackLatency, |
2822 | v->WritebackChunkSize, |
2823 | v->SOCCLK, |
2824 | v->FinalDRAMClockChangeLatency, |
2825 | v->SRExitTime, |
2826 | v->SREnterPlusExitTime, |
2827 | v->DCFCLKDeepSleep, |
2828 | v->DPPPerPlane, |
2829 | v->DCCEnable, |
2830 | v->DPPCLK, |
2831 | v->DETBufferSizeY, |
2832 | v->DETBufferSizeC, |
2833 | v->SwathHeightY, |
2834 | v->SwathHeightC, |
2835 | v->LBBitPerPixel, |
2836 | v->SwathWidthY, |
2837 | v->SwathWidthC, |
2838 | v->HRatio, |
2839 | v->HRatioChroma, |
2840 | v->vtaps, |
2841 | v->VTAPsChroma, |
2842 | v->VRatio, |
2843 | v->VRatioChroma, |
2844 | v->HTotal, |
2845 | v->PixelClock, |
2846 | v->BlendingAndTiming, |
2847 | v->BytePerPixelDETY, |
2848 | v->BytePerPixelDETC, |
2849 | v->DSTXAfterScaler, |
2850 | v->DSTYAfterScaler, |
2851 | v->WritebackEnable, |
2852 | v->WritebackPixelFormat, |
2853 | v->WritebackDestinationWidth, |
2854 | v->WritebackDestinationHeight, |
2855 | v->WritebackSourceHeight, |
2856 | &DRAMClockChangeSupport, |
2857 | &v->UrgentWatermark, |
2858 | &v->WritebackUrgentWatermark, |
2859 | &v->DRAMClockChangeWatermark, |
2860 | &v->WritebackDRAMClockChangeWatermark, |
2861 | &v->StutterExitWatermark, |
2862 | &v->StutterEnterPlusExitWatermark, |
2863 | &v->MinActiveDRAMClockChangeLatencySupported); |
2864 | |
2865 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
2866 | if (v->WritebackEnable[k] == true1) { |
2867 | if (v->BlendingAndTiming[k] == k) { |
2868 | v->ThisVStartup = v->VStartup[k]; |
2869 | } else { |
2870 | for (j = 0; j < v->NumberOfActivePlanes; ++j) { |
2871 | if (v->BlendingAndTiming[k] == j) { |
2872 | v->ThisVStartup = v->VStartup[j]; |
2873 | } |
2874 | } |
2875 | } |
2876 | v->WritebackAllowDRAMClockChangeEndPosition[k] = dml_max(0, |
2877 | v->ThisVStartup * v->HTotal[k] / v->PixelClock[k] - v->WritebackDRAMClockChangeWatermark); |
2878 | } else { |
2879 | v->WritebackAllowDRAMClockChangeEndPosition[k] = 0; |
2880 | } |
2881 | } |
2882 | |
2883 | } |
2884 | |
2885 | |
2886 | //Display Pipeline Delivery Time in Prefetch, Groups |
2887 | CalculatePixelDeliveryTimes( |
2888 | v->NumberOfActivePlanes, |
2889 | v->VRatio, |
2890 | v->VRatioChroma, |
2891 | v->VRatioPrefetchY, |
2892 | v->VRatioPrefetchC, |
2893 | v->swath_width_luma_ub, |
2894 | v->swath_width_chroma_ub, |
2895 | v->DPPPerPlane, |
2896 | v->HRatio, |
2897 | v->HRatioChroma, |
2898 | v->PixelClock, |
2899 | v->PSCL_THROUGHPUT_LUMA, |
2900 | v->PSCL_THROUGHPUT_CHROMA, |
2901 | v->DPPCLK, |
2902 | v->BytePerPixelC, |
2903 | v->SourceScan, |
2904 | v->NumberOfCursors, |
2905 | v->CursorWidth, |
2906 | v->CursorBPP, |
2907 | v->BlockWidth256BytesY, |
2908 | v->BlockHeight256BytesY, |
2909 | v->BlockWidth256BytesC, |
2910 | v->BlockHeight256BytesC, |
2911 | v->DisplayPipeLineDeliveryTimeLuma, |
2912 | v->DisplayPipeLineDeliveryTimeChroma, |
2913 | v->DisplayPipeLineDeliveryTimeLumaPrefetch, |
2914 | v->DisplayPipeLineDeliveryTimeChromaPrefetch, |
2915 | v->DisplayPipeRequestDeliveryTimeLuma, |
2916 | v->DisplayPipeRequestDeliveryTimeChroma, |
2917 | v->DisplayPipeRequestDeliveryTimeLumaPrefetch, |
2918 | v->DisplayPipeRequestDeliveryTimeChromaPrefetch, |
2919 | v->CursorRequestDeliveryTime, |
2920 | v->CursorRequestDeliveryTimePrefetch); |
2921 | |
2922 | CalculateMetaAndPTETimes( |
2923 | v->NumberOfActivePlanes, |
2924 | v->GPUVMEnable, |
2925 | v->MetaChunkSize, |
2926 | v->MinMetaChunkSizeBytes, |
2927 | v->HTotal, |
2928 | v->VRatio, |
2929 | v->VRatioChroma, |
2930 | v->DestinationLinesToRequestRowInVBlank, |
2931 | v->DestinationLinesToRequestRowInImmediateFlip, |
2932 | v->DCCEnable, |
2933 | v->PixelClock, |
2934 | v->BytePerPixelY, |
2935 | v->BytePerPixelC, |
2936 | v->SourceScan, |
2937 | v->dpte_row_height, |
2938 | v->dpte_row_height_chroma, |
2939 | v->meta_row_width, |
2940 | v->meta_row_width_chroma, |
2941 | v->meta_row_height, |
2942 | v->meta_row_height_chroma, |
2943 | v->meta_req_width, |
2944 | v->meta_req_width_chroma, |
2945 | v->meta_req_height, |
2946 | v->meta_req_height_chroma, |
2947 | v->dpte_group_bytes, |
2948 | v->PTERequestSizeY, |
2949 | v->PTERequestSizeC, |
2950 | v->PixelPTEReqWidthY, |
2951 | v->PixelPTEReqHeightY, |
2952 | v->PixelPTEReqWidthC, |
2953 | v->PixelPTEReqHeightC, |
2954 | v->dpte_row_width_luma_ub, |
2955 | v->dpte_row_width_chroma_ub, |
2956 | v->DST_Y_PER_PTE_ROW_NOM_L, |
2957 | v->DST_Y_PER_PTE_ROW_NOM_C, |
2958 | v->DST_Y_PER_META_ROW_NOM_L, |
2959 | v->DST_Y_PER_META_ROW_NOM_C, |
2960 | v->TimePerMetaChunkNominal, |
2961 | v->TimePerChromaMetaChunkNominal, |
2962 | v->TimePerMetaChunkVBlank, |
2963 | v->TimePerChromaMetaChunkVBlank, |
2964 | v->TimePerMetaChunkFlip, |
2965 | v->TimePerChromaMetaChunkFlip, |
2966 | v->time_per_pte_group_nom_luma, |
2967 | v->time_per_pte_group_vblank_luma, |
2968 | v->time_per_pte_group_flip_luma, |
2969 | v->time_per_pte_group_nom_chroma, |
2970 | v->time_per_pte_group_vblank_chroma, |
2971 | v->time_per_pte_group_flip_chroma); |
2972 | |
2973 | CalculateVMGroupAndRequestTimes( |
2974 | v->NumberOfActivePlanes, |
2975 | v->GPUVMEnable, |
2976 | v->GPUVMMaxPageTableLevels, |
2977 | v->HTotal, |
2978 | v->BytePerPixelC, |
2979 | v->DestinationLinesToRequestVMInVBlank, |
2980 | v->DestinationLinesToRequestVMInImmediateFlip, |
2981 | v->DCCEnable, |
2982 | v->PixelClock, |
2983 | v->dpte_row_width_luma_ub, |
2984 | v->dpte_row_width_chroma_ub, |
2985 | v->vm_group_bytes, |
2986 | v->dpde0_bytes_per_frame_ub_l, |
2987 | v->dpde0_bytes_per_frame_ub_c, |
2988 | v->meta_pte_bytes_per_frame_ub_l, |
2989 | v->meta_pte_bytes_per_frame_ub_c, |
2990 | v->TimePerVMGroupVBlank, |
2991 | v->TimePerVMGroupFlip, |
2992 | v->TimePerVMRequestVBlank, |
2993 | v->TimePerVMRequestFlip); |
2994 | |
2995 | |
2996 | // Min TTUVBlank |
2997 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
2998 | if (PrefetchMode == 0) { |
2999 | v->AllowDRAMClockChangeDuringVBlank[k] = true1; |
3000 | v->AllowDRAMSelfRefreshDuringVBlank[k] = true1; |
3001 | v->MinTTUVBlank[k] = dml_max( |
3002 | v->DRAMClockChangeWatermark, |
3003 | dml_max( |
3004 | v->StutterEnterPlusExitWatermark, |
3005 | v->UrgentWatermark)); |
3006 | } else if (PrefetchMode == 1) { |
3007 | v->AllowDRAMClockChangeDuringVBlank[k] = false0; |
3008 | v->AllowDRAMSelfRefreshDuringVBlank[k] = true1; |
3009 | v->MinTTUVBlank[k] = dml_max( |
3010 | v->StutterEnterPlusExitWatermark, |
3011 | v->UrgentWatermark); |
3012 | } else { |
3013 | v->AllowDRAMClockChangeDuringVBlank[k] = false0; |
3014 | v->AllowDRAMSelfRefreshDuringVBlank[k] = false0; |
3015 | v->MinTTUVBlank[k] = v->UrgentWatermark; |
3016 | } |
3017 | if (!v->DynamicMetadataEnable[k]) |
3018 | v->MinTTUVBlank[k] = v->TCalc |
3019 | + v->MinTTUVBlank[k]; |
3020 | } |
3021 | |
3022 | // DCC Configuration |
3023 | v->ActiveDPPs = 0; |
3024 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
3025 | CalculateDCCConfiguration(v->DCCEnable[k], false0, // We should always know the direction DCCProgrammingAssumesScanDirectionUnknown, |
3026 | v->SourcePixelFormat[k], |
3027 | v->SurfaceWidthY[k], |
3028 | v->SurfaceWidthC[k], |
3029 | v->SurfaceHeightY[k], |
3030 | v->SurfaceHeightC[k], |
3031 | v->DETBufferSizeInKByte * 1024, |
3032 | v->BlockHeight256BytesY[k], |
3033 | v->BlockHeight256BytesC[k], |
3034 | v->SurfaceTiling[k], |
3035 | v->BytePerPixelY[k], |
3036 | v->BytePerPixelC[k], |
3037 | v->BytePerPixelDETY[k], |
3038 | v->BytePerPixelDETC[k], |
3039 | v->SourceScan[k], |
3040 | &v->DCCYMaxUncompressedBlock[k], |
3041 | &v->DCCCMaxUncompressedBlock[k], |
3042 | &v->DCCYMaxCompressedBlock[k], |
3043 | &v->DCCCMaxCompressedBlock[k], |
3044 | &v->DCCYIndependentBlock[k], |
3045 | &v->DCCCIndependentBlock[k]); |
3046 | } |
3047 | |
3048 | { |
3049 | //Maximum Bandwidth Used |
3050 | double TotalWRBandwidth = 0; |
3051 | double MaxPerPlaneVActiveWRBandwidth = 0; |
3052 | double WRBandwidth = 0; |
3053 | double MaxUsedBW = 0; |
3054 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
3055 | if (v->WritebackEnable[k] == true1 |
3056 | && v->WritebackPixelFormat[k] == dm_444_32) { |
3057 | WRBandwidth = v->WritebackDestinationWidth[k] * v->WritebackDestinationHeight[k] |
3058 | / (v->HTotal[k] * v->WritebackSourceHeight[k] / v->PixelClock[k]) * 4; |
3059 | } else if (v->WritebackEnable[k] == true1) { |
3060 | WRBandwidth = v->WritebackDestinationWidth[k] * v->WritebackDestinationHeight[k] |
3061 | / (v->HTotal[k] * v->WritebackSourceHeight[k] / v->PixelClock[k]) * 8; |
3062 | } |
3063 | TotalWRBandwidth = TotalWRBandwidth + WRBandwidth; |
3064 | MaxPerPlaneVActiveWRBandwidth = dml_max(MaxPerPlaneVActiveWRBandwidth, WRBandwidth); |
3065 | } |
3066 | |
3067 | v->TotalDataReadBandwidth = 0; |
3068 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
3069 | v->TotalDataReadBandwidth = v->TotalDataReadBandwidth |
3070 | + v->ReadBandwidthPlaneLuma[k] |
3071 | + v->ReadBandwidthPlaneChroma[k]; |
3072 | } |
3073 | |
3074 | { |
3075 | double MaxPerPlaneVActiveRDBandwidth = 0; |
3076 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
3077 | MaxPerPlaneVActiveRDBandwidth = dml_max(MaxPerPlaneVActiveRDBandwidth, |
3078 | v->ReadBandwidthPlaneLuma[k] + v->ReadBandwidthPlaneChroma[k]); |
3079 | |
3080 | } |
3081 | } |
3082 | |
3083 | MaxUsedBW = MaxTotalRDBandwidth + TotalWRBandwidth; |
3084 | } |
3085 | |
3086 | // VStartup Margin |
3087 | v->VStartupMargin = 0; |
3088 | v->FirstMainPlane = true1; |
3089 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
3090 | if (v->BlendingAndTiming[k] == k) { |
3091 | double margin = (v->MaxVStartupLines[k] - v->VStartup[k]) * v->HTotal[k] |
3092 | / v->PixelClock[k]; |
3093 | if (v->FirstMainPlane == true1) { |
3094 | v->VStartupMargin = margin; |
3095 | v->FirstMainPlane = false0; |
3096 | } else { |
3097 | v->VStartupMargin = dml_min(v->VStartupMargin, margin); |
3098 | } |
3099 | } |
3100 | } |
3101 | |
3102 | // Stutter Efficiency |
3103 | CalculateStutterEfficiency( |
3104 | v->NumberOfActivePlanes, |
3105 | v->ROBBufferSizeInKByte, |
3106 | v->TotalDataReadBandwidth, |
3107 | v->DCFCLK, |
3108 | v->ReturnBW, |
3109 | v->SRExitTime, |
3110 | v->SynchronizedVBlank, |
3111 | v->DPPPerPlane, |
3112 | v->DETBufferSizeY, |
3113 | v->BytePerPixelY, |
3114 | v->BytePerPixelDETY, |
3115 | v->SwathWidthY, |
3116 | v->SwathHeightY, |
3117 | v->SwathHeightC, |
3118 | v->DCCRateLuma, |
3119 | v->DCCRateChroma, |
3120 | v->HTotal, |
3121 | v->VTotal, |
3122 | v->PixelClock, |
3123 | v->VRatio, |
3124 | v->SourceScan, |
3125 | v->BlockHeight256BytesY, |
3126 | v->BlockWidth256BytesY, |
3127 | v->BlockHeight256BytesC, |
3128 | v->BlockWidth256BytesC, |
3129 | v->DCCYMaxUncompressedBlock, |
3130 | v->DCCCMaxUncompressedBlock, |
3131 | v->VActive, |
3132 | v->DCCEnable, |
3133 | v->WritebackEnable, |
3134 | v->ReadBandwidthPlaneLuma, |
3135 | v->ReadBandwidthPlaneChroma, |
3136 | v->meta_row_bw, |
3137 | v->dpte_row_bw, |
3138 | &v->StutterEfficiencyNotIncludingVBlank, |
3139 | &v->StutterEfficiency, |
3140 | &v->StutterPeriod); |
3141 | } |
3142 | |
3143 | static void DisplayPipeConfiguration(struct display_mode_lib *mode_lib) |
3144 | { |
3145 | // Display Pipe Configuration |
3146 | double BytePerPixDETY[DC__NUM_DPP__MAX8] = { 0 }; |
3147 | double BytePerPixDETC[DC__NUM_DPP__MAX8] = { 0 }; |
3148 | int BytePerPixY[DC__NUM_DPP__MAX8] = { 0 }; |
3149 | int BytePerPixC[DC__NUM_DPP__MAX8] = { 0 }; |
3150 | int Read256BytesBlockHeightY[DC__NUM_DPP__MAX8] = { 0 }; |
3151 | int Read256BytesBlockHeightC[DC__NUM_DPP__MAX8] = { 0 }; |
3152 | int Read256BytesBlockWidthY[DC__NUM_DPP__MAX8] = { 0 }; |
3153 | int Read256BytesBlockWidthC[DC__NUM_DPP__MAX8] = { 0 }; |
3154 | double dummy1[DC__NUM_DPP__MAX8] = { 0 }; |
3155 | double dummy2[DC__NUM_DPP__MAX8] = { 0 }; |
3156 | double dummy3[DC__NUM_DPP__MAX8] = { 0 }; |
3157 | double dummy4[DC__NUM_DPP__MAX8] = { 0 }; |
3158 | int dummy5[DC__NUM_DPP__MAX8] = { 0 }; |
3159 | int dummy6[DC__NUM_DPP__MAX8] = { 0 }; |
3160 | bool_Bool dummy7[DC__NUM_DPP__MAX8] = { 0 }; |
3161 | bool_Bool dummysinglestring = 0; |
3162 | unsigned int k; |
3163 | |
3164 | for (k = 0; k < mode_lib->vba.NumberOfActivePlanes; ++k) { |
3165 | |
3166 | CalculateBytePerPixelAnd256BBlockSizes( |
3167 | mode_lib->vba.SourcePixelFormat[k], |
3168 | mode_lib->vba.SurfaceTiling[k], |
3169 | &BytePerPixY[k], |
3170 | &BytePerPixC[k], |
3171 | &BytePerPixDETY[k], |
3172 | &BytePerPixDETC[k], |
3173 | &Read256BytesBlockHeightY[k], |
3174 | &Read256BytesBlockHeightC[k], |
3175 | &Read256BytesBlockWidthY[k], |
3176 | &Read256BytesBlockWidthC[k]); |
3177 | } |
3178 | CalculateSwathAndDETConfiguration( |
3179 | false0, |
3180 | mode_lib->vba.NumberOfActivePlanes, |
3181 | mode_lib->vba.DETBufferSizeInKByte, |
3182 | dummy1, |
3183 | dummy2, |
3184 | mode_lib->vba.SourceScan, |
3185 | mode_lib->vba.SourcePixelFormat, |
3186 | mode_lib->vba.SurfaceTiling, |
3187 | mode_lib->vba.ViewportWidth, |
3188 | mode_lib->vba.ViewportHeight, |
3189 | mode_lib->vba.SurfaceWidthY, |
3190 | mode_lib->vba.SurfaceWidthC, |
3191 | mode_lib->vba.SurfaceHeightY, |
3192 | mode_lib->vba.SurfaceHeightC, |
3193 | Read256BytesBlockHeightY, |
3194 | Read256BytesBlockHeightC, |
3195 | Read256BytesBlockWidthY, |
3196 | Read256BytesBlockWidthC, |
3197 | mode_lib->vba.ODMCombineEnabled, |
3198 | mode_lib->vba.BlendingAndTiming, |
3199 | BytePerPixY, |
3200 | BytePerPixC, |
3201 | BytePerPixDETY, |
3202 | BytePerPixDETC, |
3203 | mode_lib->vba.HActive, |
3204 | mode_lib->vba.HRatio, |
3205 | mode_lib->vba.HRatioChroma, |
3206 | mode_lib->vba.DPPPerPlane, |
3207 | dummy5, |
3208 | dummy6, |
3209 | dummy3, |
3210 | dummy4, |
3211 | mode_lib->vba.SwathHeightY, |
3212 | mode_lib->vba.SwathHeightC, |
3213 | mode_lib->vba.DETBufferSizeY, |
3214 | mode_lib->vba.DETBufferSizeC, |
3215 | dummy7, |
3216 | &dummysinglestring); |
3217 | } |
3218 | |
3219 | static bool_Bool CalculateBytePerPixelAnd256BBlockSizes( |
3220 | enum source_format_class SourcePixelFormat, |
3221 | enum dm_swizzle_mode SurfaceTiling, |
3222 | unsigned int *BytePerPixelY, |
3223 | unsigned int *BytePerPixelC, |
3224 | double *BytePerPixelDETY, |
3225 | double *BytePerPixelDETC, |
3226 | unsigned int *BlockHeight256BytesY, |
3227 | unsigned int *BlockHeight256BytesC, |
3228 | unsigned int *BlockWidth256BytesY, |
3229 | unsigned int *BlockWidth256BytesC) |
3230 | { |
3231 | if (SourcePixelFormat == dm_444_64) { |
3232 | *BytePerPixelDETY = 8; |
3233 | *BytePerPixelDETC = 0; |
3234 | *BytePerPixelY = 8; |
3235 | *BytePerPixelC = 0; |
3236 | } else if (SourcePixelFormat == dm_444_32 || SourcePixelFormat == dm_rgbe) { |
3237 | *BytePerPixelDETY = 4; |
3238 | *BytePerPixelDETC = 0; |
3239 | *BytePerPixelY = 4; |
3240 | *BytePerPixelC = 0; |
3241 | } else if (SourcePixelFormat == dm_444_16) { |
3242 | *BytePerPixelDETY = 2; |
3243 | *BytePerPixelDETC = 0; |
3244 | *BytePerPixelY = 2; |
3245 | *BytePerPixelC = 0; |
3246 | } else if (SourcePixelFormat == dm_444_8) { |
3247 | *BytePerPixelDETY = 1; |
3248 | *BytePerPixelDETC = 0; |
3249 | *BytePerPixelY = 1; |
3250 | *BytePerPixelC = 0; |
3251 | } else if (SourcePixelFormat == dm_rgbe_alpha) { |
3252 | *BytePerPixelDETY = 4; |
3253 | *BytePerPixelDETC = 1; |
3254 | *BytePerPixelY = 4; |
3255 | *BytePerPixelC = 1; |
3256 | } else if (SourcePixelFormat == dm_420_8) { |
3257 | *BytePerPixelDETY = 1; |
3258 | *BytePerPixelDETC = 2; |
3259 | *BytePerPixelY = 1; |
3260 | *BytePerPixelC = 2; |
3261 | } else if (SourcePixelFormat == dm_420_12) { |
3262 | *BytePerPixelDETY = 2; |
3263 | *BytePerPixelDETC = 4; |
3264 | *BytePerPixelY = 2; |
3265 | *BytePerPixelC = 4; |
3266 | } else { |
3267 | *BytePerPixelDETY = 4.0 / 3; |
3268 | *BytePerPixelDETC = 8.0 / 3; |
3269 | *BytePerPixelY = 2; |
3270 | *BytePerPixelC = 4; |
3271 | } |
3272 | |
3273 | if ((SourcePixelFormat == dm_444_64 || SourcePixelFormat == dm_444_32 |
3274 | || SourcePixelFormat == dm_444_16 || SourcePixelFormat == dm_444_8 |
3275 | || SourcePixelFormat == dm_mono_16 || SourcePixelFormat == dm_mono_8 |
3276 | || SourcePixelFormat == dm_rgbe)) { |
3277 | if (SurfaceTiling == dm_sw_linear) { |
3278 | *BlockHeight256BytesY = 1; |
3279 | } else if (SourcePixelFormat == dm_444_64) { |
3280 | *BlockHeight256BytesY = 4; |
3281 | } else if (SourcePixelFormat == dm_444_8) { |
3282 | *BlockHeight256BytesY = 16; |
3283 | } else { |
3284 | *BlockHeight256BytesY = 8; |
3285 | } |
3286 | *BlockWidth256BytesY = 256U / *BytePerPixelY / *BlockHeight256BytesY; |
3287 | *BlockHeight256BytesC = 0; |
3288 | *BlockWidth256BytesC = 0; |
3289 | } else { |
3290 | if (SurfaceTiling == dm_sw_linear) { |
3291 | *BlockHeight256BytesY = 1; |
3292 | *BlockHeight256BytesC = 1; |
3293 | } else if (SourcePixelFormat == dm_rgbe_alpha) { |
3294 | *BlockHeight256BytesY = 8; |
3295 | *BlockHeight256BytesC = 16; |
3296 | } else if (SourcePixelFormat == dm_420_8) { |
3297 | *BlockHeight256BytesY = 16; |
3298 | *BlockHeight256BytesC = 8; |
3299 | } else { |
3300 | *BlockHeight256BytesY = 8; |
3301 | *BlockHeight256BytesC = 8; |
3302 | } |
3303 | *BlockWidth256BytesY = 256U / *BytePerPixelY / *BlockHeight256BytesY; |
3304 | *BlockWidth256BytesC = 256U / *BytePerPixelC / *BlockHeight256BytesC; |
3305 | } |
3306 | return true1; |
3307 | } |
3308 | |
3309 | static double CalculateTWait( |
3310 | unsigned int PrefetchMode, |
3311 | double DRAMClockChangeLatency, |
3312 | double UrgentLatency, |
3313 | double SREnterPlusExitTime) |
3314 | { |
3315 | if (PrefetchMode == 0) { |
3316 | return dml_max(DRAMClockChangeLatency + UrgentLatency, |
3317 | dml_max(SREnterPlusExitTime, UrgentLatency)); |
3318 | } else if (PrefetchMode == 1) { |
3319 | return dml_max(SREnterPlusExitTime, UrgentLatency); |
3320 | } else { |
3321 | return UrgentLatency; |
3322 | } |
3323 | } |
3324 | |
3325 | double dml30_CalculateWriteBackDISPCLK( |
3326 | enum source_format_class WritebackPixelFormat, |
3327 | double PixelClock, |
3328 | double WritebackHRatio, |
3329 | double WritebackVRatio, |
3330 | unsigned int WritebackHTaps, |
3331 | unsigned int WritebackVTaps, |
3332 | long WritebackSourceWidth, |
3333 | long WritebackDestinationWidth, |
3334 | unsigned int HTotal, |
3335 | unsigned int WritebackLineBufferSize) |
3336 | { |
3337 | double DISPCLK_H = 0, DISPCLK_V = 0, DISPCLK_HB = 0; |
3338 | |
3339 | DISPCLK_H = PixelClock * dml_ceil(WritebackHTaps / 8.0, 1) / WritebackHRatio; |
3340 | DISPCLK_V = PixelClock * (WritebackVTaps * dml_ceil(WritebackDestinationWidth / 6.0, 1) + 8.0) / HTotal; |
3341 | DISPCLK_HB = PixelClock * WritebackVTaps * (WritebackDestinationWidth * WritebackVTaps - WritebackLineBufferSize / 57.0) / 6.0 / WritebackSourceWidth; |
3342 | return dml_max3(DISPCLK_H, DISPCLK_V, DISPCLK_HB); |
3343 | } |
3344 | |
3345 | static double CalculateWriteBackDelay( |
3346 | enum source_format_class WritebackPixelFormat, |
3347 | double WritebackHRatio, |
3348 | double WritebackVRatio, |
3349 | unsigned int WritebackVTaps, |
3350 | long WritebackDestinationWidth, |
3351 | long WritebackDestinationHeight, |
3352 | long WritebackSourceHeight, |
3353 | unsigned int HTotal) |
3354 | { |
3355 | double CalculateWriteBackDelay = 0; |
3356 | double Line_length = 0; |
3357 | double Output_lines_last_notclamped = 0; |
3358 | double WritebackVInit = 0; |
3359 | |
3360 | WritebackVInit = (WritebackVRatio + WritebackVTaps + 1) / 2; |
3361 | Line_length = dml_max((double) WritebackDestinationWidth, dml_ceil(WritebackDestinationWidth / 6.0, 1) * WritebackVTaps); |
3362 | Output_lines_last_notclamped = WritebackDestinationHeight - 1 - dml_ceil((WritebackSourceHeight - WritebackVInit) / WritebackVRatio, 1); |
3363 | if (Output_lines_last_notclamped < 0) { |
3364 | CalculateWriteBackDelay = 0; |
3365 | } else { |
3366 | CalculateWriteBackDelay = Output_lines_last_notclamped * Line_length + (HTotal - WritebackDestinationWidth) + 80; |
3367 | } |
3368 | return CalculateWriteBackDelay; |
3369 | } |
3370 | |
3371 | |
3372 | static void CalculateDynamicMetadataParameters(int MaxInterDCNTileRepeaters, double DPPCLK, double DISPCLK, |
3373 | double DCFClkDeepSleep, double PixelClock, long HTotal, long VBlank, long DynamicMetadataTransmittedBytes, |
3374 | long DynamicMetadataLinesBeforeActiveRequired, int InterlaceEnable, bool_Bool ProgressiveToInterlaceUnitInOPP, |
3375 | double *Tsetup, double *Tdmbf, double *Tdmec, double *Tdmsks) |
3376 | { |
3377 | double TotalRepeaterDelayTime = 0; |
3378 | double VUpdateWidthPix = 0; |
3379 | double VReadyOffsetPix = 0; |
3380 | double VUpdateOffsetPix = 0; |
3381 | TotalRepeaterDelayTime = MaxInterDCNTileRepeaters * (2 / DPPCLK + 3 / DISPCLK); |
3382 | VUpdateWidthPix = (14 / DCFClkDeepSleep + 12 / DPPCLK + TotalRepeaterDelayTime) * PixelClock; |
3383 | VReadyOffsetPix = dml_max(150.0 / DPPCLK, TotalRepeaterDelayTime + 20 / DCFClkDeepSleep + 10 / DPPCLK) * PixelClock; |
3384 | VUpdateOffsetPix = dml_ceil(HTotal / 4.0, 1); |
3385 | *Tsetup = (VUpdateOffsetPix + VUpdateWidthPix + VReadyOffsetPix) / PixelClock; |
3386 | *Tdmbf = DynamicMetadataTransmittedBytes / 4.0 / DISPCLK; |
3387 | *Tdmec = HTotal / PixelClock; |
3388 | if (DynamicMetadataLinesBeforeActiveRequired == 0) { |
3389 | *Tdmsks = VBlank * HTotal / PixelClock / 2.0; |
3390 | } else { |
3391 | *Tdmsks = DynamicMetadataLinesBeforeActiveRequired * HTotal / PixelClock; |
3392 | } |
3393 | if (InterlaceEnable == 1 && ProgressiveToInterlaceUnitInOPP == false0) { |
3394 | *Tdmsks = *Tdmsks / 2; |
3395 | } |
3396 | } |
3397 | |
3398 | static void CalculateRowBandwidth( |
3399 | bool_Bool GPUVMEnable, |
3400 | enum source_format_class SourcePixelFormat, |
3401 | double VRatio, |
3402 | double VRatioChroma, |
3403 | bool_Bool DCCEnable, |
3404 | double LineTime, |
3405 | unsigned int MetaRowByteLuma, |
3406 | unsigned int MetaRowByteChroma, |
3407 | unsigned int meta_row_height_luma, |
3408 | unsigned int meta_row_height_chroma, |
3409 | unsigned int PixelPTEBytesPerRowLuma, |
3410 | unsigned int PixelPTEBytesPerRowChroma, |
3411 | unsigned int dpte_row_height_luma, |
3412 | unsigned int dpte_row_height_chroma, |
3413 | double *meta_row_bw, |
3414 | double *dpte_row_bw) |
3415 | { |
3416 | if (DCCEnable != true1) { |
3417 | *meta_row_bw = 0; |
3418 | } else if (SourcePixelFormat == dm_420_8 || SourcePixelFormat == dm_420_10 || SourcePixelFormat == dm_420_12 || SourcePixelFormat == dm_rgbe_alpha) { |
3419 | *meta_row_bw = VRatio * MetaRowByteLuma / (meta_row_height_luma * LineTime) |
3420 | + VRatioChroma * MetaRowByteChroma |
3421 | / (meta_row_height_chroma * LineTime); |
3422 | } else { |
3423 | *meta_row_bw = VRatio * MetaRowByteLuma / (meta_row_height_luma * LineTime); |
3424 | } |
3425 | |
3426 | if (GPUVMEnable != true1) { |
3427 | *dpte_row_bw = 0; |
3428 | } else if (SourcePixelFormat == dm_420_8 || SourcePixelFormat == dm_420_10 || SourcePixelFormat == dm_420_12 || SourcePixelFormat == dm_rgbe_alpha) { |
3429 | *dpte_row_bw = VRatio * PixelPTEBytesPerRowLuma / (dpte_row_height_luma * LineTime) |
3430 | + VRatioChroma * PixelPTEBytesPerRowChroma |
3431 | / (dpte_row_height_chroma * LineTime); |
3432 | } else { |
3433 | *dpte_row_bw = VRatio * PixelPTEBytesPerRowLuma / (dpte_row_height_luma * LineTime); |
3434 | } |
3435 | } |
3436 | |
3437 | static void CalculateFlipSchedule( |
3438 | struct display_mode_lib *mode_lib, |
3439 | double PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData, |
3440 | double PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly, |
3441 | double UrgentExtraLatency, |
3442 | double UrgentLatency, |
3443 | unsigned int GPUVMMaxPageTableLevels, |
3444 | bool_Bool HostVMEnable, |
3445 | unsigned int HostVMMaxNonCachedPageTableLevels, |
3446 | bool_Bool GPUVMEnable, |
3447 | double HostVMMinPageSize, |
3448 | double PDEAndMetaPTEBytesPerFrame, |
3449 | double MetaRowBytes, |
3450 | double DPTEBytesPerRow, |
3451 | double BandwidthAvailableForImmediateFlip, |
3452 | unsigned int TotImmediateFlipBytes, |
3453 | enum source_format_class SourcePixelFormat, |
3454 | double LineTime, |
3455 | double VRatio, |
3456 | double VRatioChroma, |
3457 | double Tno_bw, |
3458 | bool_Bool DCCEnable, |
3459 | unsigned int dpte_row_height, |
3460 | unsigned int meta_row_height, |
3461 | unsigned int dpte_row_height_chroma, |
3462 | unsigned int meta_row_height_chroma, |
3463 | double *DestinationLinesToRequestVMInImmediateFlip, |
3464 | double *DestinationLinesToRequestRowInImmediateFlip, |
3465 | double *final_flip_bw, |
3466 | bool_Bool *ImmediateFlipSupportedForPipe) |
3467 | { |
3468 | double min_row_time = 0.0; |
3469 | unsigned int HostVMDynamicLevelsTrips = 0; |
3470 | double TimeForFetchingMetaPTEImmediateFlip = 0; |
3471 | double TimeForFetchingRowInVBlankImmediateFlip = 0; |
3472 | double ImmediateFlipBW = 0; |
3473 | double HostVMInefficiencyFactor = 0; |
3474 | |
3475 | if (GPUVMEnable == true1 && HostVMEnable == true1) { |
3476 | HostVMInefficiencyFactor = PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData / PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly; |
3477 | HostVMDynamicLevelsTrips = HostVMMaxNonCachedPageTableLevels; |
3478 | } else { |
3479 | HostVMInefficiencyFactor = 1; |
3480 | HostVMDynamicLevelsTrips = 0; |
3481 | } |
3482 | |
3483 | if (GPUVMEnable == true1 || DCCEnable == true1) { |
3484 | ImmediateFlipBW = (PDEAndMetaPTEBytesPerFrame + MetaRowBytes + DPTEBytesPerRow) * BandwidthAvailableForImmediateFlip / TotImmediateFlipBytes; |
3485 | } |
3486 | |
3487 | if (GPUVMEnable == true1) { |
3488 | TimeForFetchingMetaPTEImmediateFlip = dml_max3(Tno_bw + PDEAndMetaPTEBytesPerFrame * HostVMInefficiencyFactor / ImmediateFlipBW, |
3489 | UrgentExtraLatency + UrgentLatency * (GPUVMMaxPageTableLevels * (HostVMDynamicLevelsTrips + 1) - 1), LineTime / 4.0); |
3490 | } else { |
3491 | TimeForFetchingMetaPTEImmediateFlip = 0; |
3492 | } |
3493 | |
3494 | *DestinationLinesToRequestVMInImmediateFlip = dml_ceil(4.0 * (TimeForFetchingMetaPTEImmediateFlip / LineTime), 1) / 4.0; |
3495 | if ((GPUVMEnable == true1 || DCCEnable == true1)) { |
3496 | TimeForFetchingRowInVBlankImmediateFlip = dml_max3((MetaRowBytes + DPTEBytesPerRow * HostVMInefficiencyFactor) / ImmediateFlipBW, |
3497 | UrgentLatency * (HostVMDynamicLevelsTrips + 1), LineTime / 4); |
3498 | } else { |
3499 | TimeForFetchingRowInVBlankImmediateFlip = 0; |
3500 | } |
3501 | |
3502 | *DestinationLinesToRequestRowInImmediateFlip = dml_ceil(4.0 * (TimeForFetchingRowInVBlankImmediateFlip / LineTime), 1) / 4.0; |
3503 | |
3504 | if (GPUVMEnable == true1) { |
3505 | *final_flip_bw = dml_max(PDEAndMetaPTEBytesPerFrame * HostVMInefficiencyFactor / (*DestinationLinesToRequestVMInImmediateFlip * LineTime), |
3506 | (MetaRowBytes + DPTEBytesPerRow * HostVMInefficiencyFactor) / (*DestinationLinesToRequestRowInImmediateFlip * LineTime)); |
3507 | } else if ((GPUVMEnable == true1 || DCCEnable == true1)) { |
3508 | *final_flip_bw = (MetaRowBytes + DPTEBytesPerRow * HostVMInefficiencyFactor) / (*DestinationLinesToRequestRowInImmediateFlip * LineTime); |
3509 | } else { |
3510 | *final_flip_bw = 0; |
3511 | } |
3512 | |
3513 | |
3514 | if (SourcePixelFormat == dm_420_8 || SourcePixelFormat == dm_420_10 || SourcePixelFormat == dm_rgbe_alpha) { |
3515 | if (GPUVMEnable == true1 && DCCEnable != true1) { |
3516 | min_row_time = dml_min(dpte_row_height * LineTime / VRatio, dpte_row_height_chroma * LineTime / VRatioChroma); |
3517 | } else if (GPUVMEnable != true1 && DCCEnable == true1) { |
3518 | min_row_time = dml_min(meta_row_height * LineTime / VRatio, meta_row_height_chroma * LineTime / VRatioChroma); |
3519 | } else { |
3520 | min_row_time = dml_min4(dpte_row_height * LineTime / VRatio, meta_row_height * LineTime / VRatio, |
3521 | dpte_row_height_chroma * LineTime / VRatioChroma, meta_row_height_chroma * LineTime / VRatioChroma); |
3522 | } |
3523 | } else { |
3524 | if (GPUVMEnable == true1 && DCCEnable != true1) { |
3525 | min_row_time = dpte_row_height * LineTime / VRatio; |
3526 | } else if (GPUVMEnable != true1 && DCCEnable == true1) { |
3527 | min_row_time = meta_row_height * LineTime / VRatio; |
3528 | } else { |
3529 | min_row_time = dml_min(dpte_row_height * LineTime / VRatio, meta_row_height * LineTime / VRatio); |
3530 | } |
3531 | } |
3532 | |
3533 | if (*DestinationLinesToRequestVMInImmediateFlip >= 32 || *DestinationLinesToRequestRowInImmediateFlip >= 16 |
3534 | || TimeForFetchingMetaPTEImmediateFlip + 2 * TimeForFetchingRowInVBlankImmediateFlip > min_row_time) { |
3535 | *ImmediateFlipSupportedForPipe = false0; |
3536 | } else { |
3537 | *ImmediateFlipSupportedForPipe = true1; |
3538 | } |
3539 | } |
3540 | |
3541 | static double TruncToValidBPP( |
3542 | double LinkBitRate, |
3543 | int Lanes, |
3544 | long HTotal, |
3545 | long HActive, |
3546 | double PixelClock, |
3547 | double DesiredBPP, |
3548 | bool_Bool DSCEnable, |
3549 | enum output_encoder_class Output, |
3550 | enum output_format_class Format, |
3551 | unsigned int DSCInputBitPerComponent, |
3552 | int DSCSlices, |
3553 | int AudioRate, |
3554 | int AudioLayout, |
3555 | enum odm_combine_mode ODMCombine) |
3556 | { |
3557 | double MaxLinkBPP = 0; |
3558 | int MinDSCBPP = 0; |
3559 | double MaxDSCBPP = 0; |
3560 | int NonDSCBPP0 = 0; |
3561 | int NonDSCBPP1 = 0; |
3562 | int NonDSCBPP2 = 0; |
3563 | |
3564 | if (Format == dm_420) { |
3565 | NonDSCBPP0 = 12; |
3566 | NonDSCBPP1 = 15; |
3567 | NonDSCBPP2 = 18; |
3568 | MinDSCBPP = 6; |
3569 | MaxDSCBPP = 1.5 * DSCInputBitPerComponent - 1.0 / 16; |
3570 | } else if (Format == dm_444) { |
3571 | NonDSCBPP0 = 24; |
3572 | NonDSCBPP1 = 30; |
3573 | NonDSCBPP2 = 36; |
3574 | MinDSCBPP = 8; |
3575 | MaxDSCBPP = 3 * DSCInputBitPerComponent - 1.0 / 16; |
3576 | } else { |
3577 | if (Output == dm_hdmi) { |
3578 | NonDSCBPP0 = 24; |
3579 | NonDSCBPP1 = 24; |
3580 | NonDSCBPP2 = 24; |
3581 | } |
3582 | else { |
3583 | NonDSCBPP0 = 16; |
3584 | NonDSCBPP1 = 20; |
3585 | NonDSCBPP2 = 24; |
3586 | } |
3587 | |
3588 | if (Format == dm_n422) { |
3589 | MinDSCBPP = 7; |
3590 | MaxDSCBPP = 2 * DSCInputBitPerComponent - 1.0 / 16.0; |
3591 | } |
3592 | else { |
3593 | MinDSCBPP = 8; |
3594 | MaxDSCBPP = 3 * DSCInputBitPerComponent - 1.0 / 16.0; |
3595 | } |
3596 | } |
3597 | |
3598 | if (DSCEnable && Output == dm_dp) { |
3599 | MaxLinkBPP = LinkBitRate / 10 * 8 * Lanes / PixelClock * (1 - 2.4 / 100); |
3600 | } else { |
3601 | MaxLinkBPP = LinkBitRate / 10 * 8 * Lanes / PixelClock; |
3602 | } |
3603 | |
3604 | if (ODMCombine == dm_odm_combine_mode_4to1 && MaxLinkBPP > 16) { |
3605 | MaxLinkBPP = 16; |
3606 | } else if (ODMCombine == dm_odm_combine_mode_2to1 && MaxLinkBPP > 32) { |
3607 | MaxLinkBPP = 32; |
3608 | } |
3609 | |
3610 | |
3611 | if (DesiredBPP == 0) { |
3612 | if (DSCEnable) { |
3613 | if (MaxLinkBPP < MinDSCBPP) { |
3614 | return BPP_INVALID0; |
3615 | } else if (MaxLinkBPP >= MaxDSCBPP) { |
3616 | return MaxDSCBPP; |
3617 | } else { |
3618 | return dml_floor(16.0 * MaxLinkBPP, 1.0) / 16.0; |
3619 | } |
3620 | } else { |
3621 | if (MaxLinkBPP >= NonDSCBPP2) { |
3622 | return NonDSCBPP2; |
3623 | } else if (MaxLinkBPP >= NonDSCBPP1) { |
3624 | return NonDSCBPP1; |
3625 | } else if (MaxLinkBPP >= NonDSCBPP0) { |
3626 | return NonDSCBPP0; |
3627 | } else { |
3628 | return BPP_INVALID0; |
3629 | } |
3630 | } |
3631 | } else { |
3632 | if (!((DSCEnable == false0 && (DesiredBPP == NonDSCBPP2 || DesiredBPP == NonDSCBPP1 || DesiredBPP == NonDSCBPP0)) || |
3633 | (DSCEnable && DesiredBPP >= MinDSCBPP && DesiredBPP <= MaxDSCBPP))) { |
3634 | return BPP_INVALID0; |
3635 | } else { |
3636 | return DesiredBPP; |
3637 | } |
3638 | } |
3639 | return BPP_INVALID0; |
3640 | } |
3641 | |
3642 | void dml30_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_lib) |
3643 | { |
3644 | struct vba_vars_st *v = &mode_lib->vba; |
3645 | int MinPrefetchMode = 0; |
3646 | int MaxPrefetchMode = 2; |
3647 | int i; |
3648 | unsigned int j, k, m; |
3649 | bool_Bool EnoughWritebackUnits = true1; |
3650 | bool_Bool WritebackModeSupport = true1; |
3651 | bool_Bool ViewportExceedsSurface = false0; |
3652 | double MaxTotalVActiveRDBandwidth = 0; |
3653 | long ReorderingBytes = 0; |
3654 | bool_Bool NotUrgentLatencyHiding[DC__NUM_DPP__MAX8] = { 0 }; |
3655 | |
3656 | /*MODE SUPPORT, VOLTAGE STATE AND SOC CONFIGURATION*/ |
3657 | |
3658 | /*Scale Ratio, taps Support Check*/ |
3659 | |
3660 | v->ScaleRatioAndTapsSupport = true1; |
3661 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
3662 | if (v->ScalerEnabled[k] == false0 |
3663 | && ((v->SourcePixelFormat[k] != dm_444_64 |
3664 | && v->SourcePixelFormat[k] != dm_444_32 |
3665 | && v->SourcePixelFormat[k] != dm_444_16 |
3666 | && v->SourcePixelFormat[k] != dm_mono_16 |
3667 | && v->SourcePixelFormat[k] != dm_mono_8 |
3668 | && v->SourcePixelFormat[k] != dm_rgbe |
3669 | && v->SourcePixelFormat[k] != dm_rgbe_alpha) |
3670 | || v->HRatio[k] != 1.0 |
3671 | || v->htaps[k] != 1.0 |
3672 | || v->VRatio[k] != 1.0 |
3673 | || v->vtaps[k] != 1.0)) { |
3674 | v->ScaleRatioAndTapsSupport = false0; |
3675 | } else if (v->vtaps[k] < 1.0 || v->vtaps[k] > 8.0 |
3676 | || v->htaps[k] < 1.0 || v->htaps[k] > 8.0 |
3677 | || (v->htaps[k] > 1.0 |
3678 | && (v->htaps[k] % 2) == 1) |
3679 | || v->HRatio[k] > v->MaxHSCLRatio |
3680 | || v->VRatio[k] > v->MaxVSCLRatio |
3681 | || v->HRatio[k] > v->htaps[k] |
3682 | || v->VRatio[k] > v->vtaps[k] |
3683 | || (v->SourcePixelFormat[k] != dm_444_64 |
3684 | && v->SourcePixelFormat[k] != dm_444_32 |
3685 | && v->SourcePixelFormat[k] != dm_444_16 |
3686 | && v->SourcePixelFormat[k] != dm_mono_16 |
3687 | && v->SourcePixelFormat[k] != dm_mono_8 |
3688 | && v->SourcePixelFormat[k] != dm_rgbe |
3689 | && (v->VTAPsChroma[k] < 1 |
3690 | || v->VTAPsChroma[k] > 8 |
3691 | || v->HTAPsChroma[k] < 1 |
3692 | || v->HTAPsChroma[k] > 8 |
3693 | || (v->HTAPsChroma[k] > 1 && v->HTAPsChroma[k] % 2 == 1) |
3694 | || v->HRatioChroma[k] > v->MaxHSCLRatio |
3695 | || v->VRatioChroma[k] > v->MaxVSCLRatio |
3696 | || v->HRatioChroma[k] > v->HTAPsChroma[k] |
3697 | || v->VRatioChroma[k] > v->VTAPsChroma[k]))) { |
3698 | v->ScaleRatioAndTapsSupport = false0; |
3699 | } |
3700 | } |
3701 | /*Source Format, Pixel Format and Scan Support Check*/ |
3702 | |
3703 | v->SourceFormatPixelAndScanSupport = true1; |
3704 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
3705 | if ((v->SurfaceTiling[k] == dm_sw_linear && (!(v->SourceScan[k] != dm_vert) || v->DCCEnable[k] == true1)) |
3706 | || ((v->SurfaceTiling[k] == dm_sw_64kb_d || v->SurfaceTiling[k] == dm_sw_64kb_d_t || v->SurfaceTiling[k] == dm_sw_64kb_d_x) |
3707 | && !(v->SourcePixelFormat[k] == dm_444_64))) { |
3708 | v->SourceFormatPixelAndScanSupport = false0; |
3709 | } |
3710 | } |
3711 | /*Bandwidth Support Check*/ |
3712 | |
3713 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
3714 | CalculateBytePerPixelAnd256BBlockSizes( |
3715 | v->SourcePixelFormat[k], |
3716 | v->SurfaceTiling[k], |
3717 | &v->BytePerPixelY[k], |
3718 | &v->BytePerPixelC[k], |
3719 | &v->BytePerPixelInDETY[k], |
3720 | &v->BytePerPixelInDETC[k], |
3721 | &v->Read256BlockHeightY[k], |
3722 | &v->Read256BlockHeightC[k], |
3723 | &v->Read256BlockWidthY[k], |
3724 | &v->Read256BlockWidthC[k]); |
3725 | } |
3726 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
3727 | if (v->SourceScan[k] != dm_vert) { |
3728 | v->SwathWidthYSingleDPP[k] = v->ViewportWidth[k]; |
3729 | v->SwathWidthCSingleDPP[k] = v->ViewportWidthChroma[k]; |
3730 | } else { |
3731 | v->SwathWidthYSingleDPP[k] = v->ViewportHeight[k]; |
3732 | v->SwathWidthCSingleDPP[k] = v->ViewportHeightChroma[k]; |
3733 | } |
3734 | } |
3735 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
3736 | v->ReadBandwidthLuma[k] = v->SwathWidthYSingleDPP[k] * dml_ceil(v->BytePerPixelInDETY[k], 1.0) / (v->HTotal[k] / v->PixelClock[k]) * v->VRatio[k]; |
3737 | v->ReadBandwidthChroma[k] = v->SwathWidthYSingleDPP[k] / 2 * dml_ceil(v->BytePerPixelInDETC[k], 2.0) / (v->HTotal[k] / v->PixelClock[k]) * v->VRatio[k] / 2.0; |
3738 | } |
3739 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
3740 | if (v->WritebackEnable[k] == true1 |
3741 | && v->WritebackPixelFormat[k] == dm_444_64) { |
3742 | v->WriteBandwidth[k] = v->WritebackDestinationWidth[k] |
3743 | * v->WritebackDestinationHeight[k] |
3744 | / (v->WritebackSourceHeight[k] |
3745 | * v->HTotal[k] |
3746 | / v->PixelClock[k]) * 8.0; |
3747 | } else if (v->WritebackEnable[k] == true1) { |
3748 | v->WriteBandwidth[k] = v->WritebackDestinationWidth[k] |
3749 | * v->WritebackDestinationHeight[k] |
3750 | / (v->WritebackSourceHeight[k] |
3751 | * v->HTotal[k] |
3752 | / v->PixelClock[k]) * 4.0; |
3753 | } else { |
3754 | v->WriteBandwidth[k] = 0.0; |
3755 | } |
3756 | } |
3757 | |
3758 | /*Writeback Latency support check*/ |
3759 | |
3760 | v->WritebackLatencySupport = true1; |
3761 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
3762 | if (v->WritebackEnable[k] == true1) { |
3763 | if (v->WritebackConfiguration == dm_whole_buffer_for_single_stream_no_interleave || |
3764 | v->WritebackConfiguration == dm_whole_buffer_for_single_stream_interleave) { |
3765 | if (v->WriteBandwidth[k] |
3766 | > 2.0 * v->WritebackInterfaceBufferSize * 1024 |
3767 | / v->WritebackLatency) { |
3768 | v->WritebackLatencySupport = false0; |
3769 | } |
3770 | } else { |
3771 | if (v->WriteBandwidth[k] |
3772 | > v->WritebackInterfaceBufferSize * 1024 |
3773 | / v->WritebackLatency) { |
3774 | v->WritebackLatencySupport = false0; |
3775 | } |
3776 | } |
3777 | } |
3778 | } |
3779 | |
3780 | /*Writeback Mode Support Check*/ |
3781 | |
3782 | v->TotalNumberOfActiveWriteback = 0; |
3783 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
3784 | if (v->WritebackEnable[k] == true1) { |
3785 | v->TotalNumberOfActiveWriteback = |
3786 | v->TotalNumberOfActiveWriteback + 1; |
3787 | } |
3788 | } |
3789 | |
3790 | if (v->TotalNumberOfActiveWriteback > v->MaxNumWriteback) { |
3791 | EnoughWritebackUnits = false0; |
3792 | } |
3793 | if (!v->WritebackSupportInterleaveAndUsingWholeBufferForASingleStream |
3794 | && (v->WritebackConfiguration == dm_whole_buffer_for_single_stream_no_interleave |
3795 | || v->WritebackConfiguration == dm_whole_buffer_for_single_stream_interleave)) { |
3796 | |
3797 | WritebackModeSupport = false0; |
3798 | } |
3799 | if (v->WritebackConfiguration == dm_whole_buffer_for_single_stream_no_interleave && v->TotalNumberOfActiveWriteback > 1) { |
3800 | WritebackModeSupport = false0; |
3801 | } |
3802 | |
3803 | /*Writeback Scale Ratio and Taps Support Check*/ |
3804 | |
3805 | v->WritebackScaleRatioAndTapsSupport = true1; |
3806 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
3807 | if (v->WritebackEnable[k] == true1) { |
3808 | if (v->WritebackHRatio[k] > v->WritebackMaxHSCLRatio |
3809 | || v->WritebackVRatio[k] |
3810 | > v->WritebackMaxVSCLRatio |
3811 | || v->WritebackHRatio[k] |
3812 | < v->WritebackMinHSCLRatio |
3813 | || v->WritebackVRatio[k] |
3814 | < v->WritebackMinVSCLRatio |
3815 | || v->WritebackHTaps[k] |
3816 | > v->WritebackMaxHSCLTaps |
3817 | || v->WritebackVTaps[k] |
3818 | > v->WritebackMaxVSCLTaps |
3819 | || v->WritebackHRatio[k] |
3820 | > v->WritebackHTaps[k] |
3821 | || v->WritebackVRatio[k] |
3822 | > v->WritebackVTaps[k] |
3823 | || (v->WritebackHTaps[k] > 2.0 |
3824 | && ((v->WritebackHTaps[k] % 2) |
3825 | == 1))) { |
3826 | v->WritebackScaleRatioAndTapsSupport = false0; |
3827 | } |
3828 | if (2.0 * v->WritebackDestinationWidth[k] * (v->WritebackVTaps[k] - 1) * 57 > v->WritebackLineBufferSize) { |
3829 | v->WritebackScaleRatioAndTapsSupport = false0; |
3830 | } |
3831 | } |
3832 | } |
3833 | /*Maximum DISPCLK/DPPCLK Support check*/ |
3834 | |
3835 | v->WritebackRequiredDISPCLK = 0.0; |
3836 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
3837 | if (v->WritebackEnable[k] == true1) { |
3838 | v->WritebackRequiredDISPCLK = dml_max(v->WritebackRequiredDISPCLK, |
3839 | dml30_CalculateWriteBackDISPCLK( |
3840 | v->WritebackPixelFormat[k], |
3841 | v->PixelClock[k], |
3842 | v->WritebackHRatio[k], |
3843 | v->WritebackVRatio[k], |
3844 | v->WritebackHTaps[k], |
3845 | v->WritebackVTaps[k], |
3846 | v->WritebackSourceWidth[k], |
3847 | v->WritebackDestinationWidth[k], |
3848 | v->HTotal[k], |
3849 | v->WritebackLineBufferSize)); |
3850 | } |
3851 | } |
3852 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
3853 | if (v->HRatio[k] > 1.0) { |
3854 | v->PSCL_FACTOR[k] = dml_min(v->MaxDCHUBToPSCLThroughput, v->MaxPSCLToLBThroughput * v->HRatio[k] / dml_ceil(v->htaps[k] / 6.0, 1.0)); |
3855 | } else { |
3856 | v->PSCL_FACTOR[k] = dml_min(v->MaxDCHUBToPSCLThroughput, v->MaxPSCLToLBThroughput); |
3857 | } |
3858 | if (v->BytePerPixelC[k] == 0.0) { |
3859 | v->PSCL_FACTOR_CHROMA[k] = 0.0; |
3860 | v->MinDPPCLKUsingSingleDPP[k] = v->PixelClock[k] |
3861 | * dml_max3(v->vtaps[k] / 6.0 * dml_min(1.0, v->HRatio[k]), v->HRatio[k] * v->VRatio[k] / v->PSCL_FACTOR[k], 1.0); |
3862 | if ((v->htaps[k] > 6.0 || v->vtaps[k] > 6.0) && v->MinDPPCLKUsingSingleDPP[k] < 2.0 * v->PixelClock[k]) { |
3863 | v->MinDPPCLKUsingSingleDPP[k] = 2.0 * v->PixelClock[k]; |
3864 | } |
3865 | } else { |
3866 | if (v->HRatioChroma[k] > 1.0) { |
3867 | v->PSCL_FACTOR_CHROMA[k] = dml_min(v->MaxDCHUBToPSCLThroughput, |
3868 | v->MaxPSCLToLBThroughput * v->HRatioChroma[k] / dml_ceil(v->HTAPsChroma[k] / 6.0, 1.0)); |
3869 | } else { |
3870 | v->PSCL_FACTOR_CHROMA[k] = dml_min(v->MaxDCHUBToPSCLThroughput, v->MaxPSCLToLBThroughput); |
3871 | } |
3872 | v->MinDPPCLKUsingSingleDPP[k] = v->PixelClock[k] * dml_max5(v->vtaps[k] / 6.0 * dml_min(1.0, v->HRatio[k]), |
3873 | v->HRatio[k] * v->VRatio[k] / v->PSCL_FACTOR[k], |
3874 | v->VTAPsChroma[k] / 6.0 * dml_min(1.0, v->HRatioChroma[k]), |
3875 | v->HRatioChroma[k] * v->VRatioChroma[k] / v->PSCL_FACTOR_CHROMA[k], |
3876 | 1.0); |
3877 | if ((v->htaps[k] > 6.0 || v->vtaps[k] > 6.0 || v->HTAPsChroma[k] > 6.0 || v->VTAPsChroma[k] > 6.0) |
3878 | && v->MinDPPCLKUsingSingleDPP[k] < 2.0 * v->PixelClock[k]) { |
3879 | v->MinDPPCLKUsingSingleDPP[k] = 2.0 * v->PixelClock[k]; |
3880 | } |
3881 | } |
3882 | } |
3883 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
3884 | int MaximumSwathWidthSupportLuma = 0; |
3885 | int MaximumSwathWidthSupportChroma = 0; |
3886 | |
3887 | if (v->SurfaceTiling[k] == dm_sw_linear) { |
3888 | MaximumSwathWidthSupportLuma = 8192.0; |
3889 | } else if (v->SourceScan[k] == dm_vert && v->BytePerPixelC[k] > 0) { |
3890 | MaximumSwathWidthSupportLuma = 2880.0; |
3891 | } else { |
3892 | MaximumSwathWidthSupportLuma = 5760.0; |
3893 | } |
3894 | |
3895 | if (v->SourcePixelFormat[k] == dm_420_8 || v->SourcePixelFormat[k] == dm_420_10 || v->SourcePixelFormat[k] == dm_420_12) { |
3896 | MaximumSwathWidthSupportChroma = MaximumSwathWidthSupportLuma / 2.0; |
3897 | } else { |
3898 | MaximumSwathWidthSupportChroma = MaximumSwathWidthSupportLuma; |
3899 | } |
3900 | v->MaximumSwathWidthInLineBufferLuma = v->LineBufferSize * dml_max(v->HRatio[k], 1.0) / v->LBBitPerPixel[k] |
3901 | / (v->vtaps[k] + dml_max(dml_ceil(v->VRatio[k], 1.0) - 2, 0.0)); |
3902 | if (v->BytePerPixelC[k] == 0.0) { |
3903 | v->MaximumSwathWidthInLineBufferChroma = 0; |
3904 | } else { |
3905 | v->MaximumSwathWidthInLineBufferChroma = v->LineBufferSize * dml_max(v->HRatioChroma[k], 1.0) / v->LBBitPerPixel[k] |
3906 | / (v->VTAPsChroma[k] + dml_max(dml_ceil(v->VRatioChroma[k], 1.0) - 2, 0.0)); |
3907 | } |
3908 | v->MaximumSwathWidthLuma[k] = dml_min(MaximumSwathWidthSupportLuma, v->MaximumSwathWidthInLineBufferLuma); |
3909 | v->MaximumSwathWidthChroma[k] = dml_min(MaximumSwathWidthSupportChroma, v->MaximumSwathWidthInLineBufferChroma); |
3910 | } |
3911 | |
3912 | CalculateSwathAndDETConfiguration( |
3913 | true1, |
3914 | v->NumberOfActivePlanes, |
3915 | v->DETBufferSizeInKByte, |
3916 | v->MaximumSwathWidthLuma, |
3917 | v->MaximumSwathWidthChroma, |
3918 | v->SourceScan, |
3919 | v->SourcePixelFormat, |
3920 | v->SurfaceTiling, |
3921 | v->ViewportWidth, |
3922 | v->ViewportHeight, |
3923 | v->SurfaceWidthY, |
3924 | v->SurfaceWidthC, |
3925 | v->SurfaceHeightY, |
3926 | v->SurfaceHeightC, |
3927 | v->Read256BlockHeightY, |
3928 | v->Read256BlockHeightC, |
3929 | v->Read256BlockWidthY, |
3930 | v->Read256BlockWidthC, |
3931 | v->odm_combine_dummy, |
3932 | v->BlendingAndTiming, |
3933 | v->BytePerPixelY, |
3934 | v->BytePerPixelC, |
3935 | v->BytePerPixelInDETY, |
3936 | v->BytePerPixelInDETC, |
3937 | v->HActive, |
3938 | v->HRatio, |
3939 | v->HRatioChroma, |
3940 | v->DPPPerPlane, |
3941 | v->swath_width_luma_ub, |
3942 | v->swath_width_chroma_ub, |
3943 | v->SwathWidthY, |
3944 | v->SwathWidthC, |
3945 | v->SwathHeightY, |
3946 | v->SwathHeightC, |
3947 | v->DETBufferSizeY, |
3948 | v->DETBufferSizeC, |
3949 | v->SingleDPPViewportSizeSupportPerPlane, |
3950 | &v->ViewportSizeSupport[0][0]); |
3951 | |
3952 | for (i = 0; i < v->soc.num_states; i++) { |
3953 | for (j = 0; j < 2; j++) { |
3954 | v->MaxDispclkRoundedDownToDFSGranularity = RoundToDFSGranularityDown(v->MaxDispclk[i], v->DISPCLKDPPCLKVCOSpeed); |
3955 | v->MaxDppclkRoundedDownToDFSGranularity = RoundToDFSGranularityDown(v->MaxDppclk[i], v->DISPCLKDPPCLKVCOSpeed); |
3956 | v->RequiredDISPCLK[i][j] = 0.0; |
3957 | v->DISPCLK_DPPCLK_Support[i][j] = true1; |
3958 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
3959 | v->PlaneRequiredDISPCLKWithoutODMCombine = v->PixelClock[k] * (1.0 + v->DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) |
3960 | * (1.0 + v->DISPCLKRampingMargin / 100.0); |
3961 | if ((v->PlaneRequiredDISPCLKWithoutODMCombine >= v->MaxDispclk[i] && v->MaxDispclk[i] == v->MaxDispclk[mode_lib->soc.num_states - 1] |
3962 | && v->MaxDppclk[i] == v->MaxDppclk[mode_lib->soc.num_states - 1])) { |
3963 | v->PlaneRequiredDISPCLKWithoutODMCombine = v->PixelClock[k] * (1 + v->DISPCLKDPPCLKDSCCLKDownSpreading / 100.0); |
3964 | } |
3965 | v->PlaneRequiredDISPCLKWithODMCombine2To1 = v->PixelClock[k] / 2 * (1 + v->DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) |
3966 | * (1 + v->DISPCLKRampingMargin / 100.0); |
3967 | if ((v->PlaneRequiredDISPCLKWithODMCombine2To1 >= v->MaxDispclk[i] && v->MaxDispclk[i] == v->MaxDispclk[mode_lib->soc.num_states - 1] |
3968 | && v->MaxDppclk[i] == v->MaxDppclk[mode_lib->soc.num_states - 1])) { |
3969 | v->PlaneRequiredDISPCLKWithODMCombine2To1 = v->PixelClock[k] / 2 * (1 + v->DISPCLKDPPCLKDSCCLKDownSpreading / 100.0); |
3970 | } |
3971 | v->PlaneRequiredDISPCLKWithODMCombine4To1 = v->PixelClock[k] / 4 * (1 + v->DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) |
3972 | * (1 + v->DISPCLKRampingMargin / 100.0); |
3973 | if ((v->PlaneRequiredDISPCLKWithODMCombine4To1 >= v->MaxDispclk[i] && v->MaxDispclk[i] == v->MaxDispclk[mode_lib->soc.num_states - 1] |
3974 | && v->MaxDppclk[i] == v->MaxDppclk[mode_lib->soc.num_states - 1])) { |
3975 | v->PlaneRequiredDISPCLKWithODMCombine4To1 = v->PixelClock[k] / 4 * (1 + v->DISPCLKDPPCLKDSCCLKDownSpreading / 100.0); |
3976 | } |
3977 | |
3978 | if (v->ODMCombinePolicy == dm_odm_combine_policy_none) { |
3979 | v->ODMCombineEnablePerState[i][k] = dm_odm_combine_mode_disabled; |
3980 | v->PlaneRequiredDISPCLK = v->PlaneRequiredDISPCLKWithoutODMCombine; |
3981 | } else if (v->ODMCombinePolicy == dm_odm_combine_policy_2to1) { |
3982 | v->ODMCombineEnablePerState[i][k] = dm_odm_combine_mode_2to1; |
3983 | v->PlaneRequiredDISPCLK = v->PlaneRequiredDISPCLKWithODMCombine2To1; |
3984 | } else if (v->ODMCombinePolicy == dm_odm_combine_policy_4to1 |
3985 | || v->PlaneRequiredDISPCLKWithODMCombine2To1 > v->MaxDispclkRoundedDownToDFSGranularity) { |
3986 | v->ODMCombineEnablePerState[i][k] = dm_odm_combine_mode_4to1; |
3987 | v->PlaneRequiredDISPCLK = v->PlaneRequiredDISPCLKWithODMCombine4To1; |
3988 | } else if (v->PlaneRequiredDISPCLKWithoutODMCombine > v->MaxDispclkRoundedDownToDFSGranularity) { |
3989 | v->ODMCombineEnablePerState[i][k] = dm_odm_combine_mode_2to1; |
3990 | v->PlaneRequiredDISPCLK = v->PlaneRequiredDISPCLKWithODMCombine2To1; |
3991 | } else { |
3992 | v->ODMCombineEnablePerState[i][k] = dm_odm_combine_mode_disabled; |
3993 | v->PlaneRequiredDISPCLK = v->PlaneRequiredDISPCLKWithoutODMCombine; |
3994 | } |
3995 | if (v->DSCEnabled[k] && v->HActive[k] > DCN30_MAX_DSC_IMAGE_WIDTH5184 |
3996 | && v->ODMCombineEnablePerState[i][k] != dm_odm_combine_mode_4to1) { |
3997 | if (v->HActive[k] / 2 > DCN30_MAX_DSC_IMAGE_WIDTH5184) { |
3998 | v->ODMCombineEnablePerState[i][k] = dm_odm_combine_mode_4to1; |
3999 | v->PlaneRequiredDISPCLK = v->PlaneRequiredDISPCLKWithODMCombine4To1; |
4000 | } else { |
4001 | v->ODMCombineEnablePerState[i][k] = dm_odm_combine_mode_2to1; |
4002 | v->PlaneRequiredDISPCLK = v->PlaneRequiredDISPCLKWithODMCombine2To1; |
4003 | } |
4004 | } |
4005 | if (v->OutputFormat[k] == dm_420 && v->HActive[k] > DCN30_MAX_FMT_420_BUFFER_WIDTH4096 |
4006 | && v->ODMCombineEnablePerState[i][k] != dm_odm_combine_mode_4to1) { |
4007 | if (v->HActive[k] / 2 > DCN30_MAX_FMT_420_BUFFER_WIDTH4096) { |
4008 | v->ODMCombineEnablePerState[i][k] = dm_odm_combine_mode_4to1; |
4009 | v->PlaneRequiredDISPCLK = v->PlaneRequiredDISPCLKWithODMCombine4To1; |
4010 | } else { |
4011 | v->ODMCombineEnablePerState[i][k] = dm_odm_combine_mode_2to1; |
4012 | v->PlaneRequiredDISPCLK = v->PlaneRequiredDISPCLKWithODMCombine2To1; |
4013 | } |
4014 | } |
4015 | if (v->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_4to1) { |
4016 | v->MPCCombine[i][j][k] = false0; |
4017 | v->NoOfDPP[i][j][k] = 4; |
4018 | v->RequiredDPPCLK[i][j][k] = v->MinDPPCLKUsingSingleDPP[k] * (1 + v->DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) / 4; |
4019 | } else if (v->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_2to1) { |
4020 | v->MPCCombine[i][j][k] = false0; |
4021 | v->NoOfDPP[i][j][k] = 2; |
4022 | v->RequiredDPPCLK[i][j][k] = v->MinDPPCLKUsingSingleDPP[k] * (1 + v->DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) / 2; |
4023 | } else if ((v->WhenToDoMPCCombine == dm_mpc_never |
4024 | || (v->MinDPPCLKUsingSingleDPP[k] * (1 + v->DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) <= v->MaxDppclkRoundedDownToDFSGranularity |
4025 | && v->SingleDPPViewportSizeSupportPerPlane[k] == true1))) { |
4026 | v->MPCCombine[i][j][k] = false0; |
4027 | v->NoOfDPP[i][j][k] = 1; |
4028 | v->RequiredDPPCLK[i][j][k] = v->MinDPPCLKUsingSingleDPP[k] * (1.0 + v->DISPCLKDPPCLKDSCCLKDownSpreading / 100.0); |
4029 | } else { |
4030 | v->MPCCombine[i][j][k] = true1; |
4031 | v->NoOfDPP[i][j][k] = 2; |
4032 | v->RequiredDPPCLK[i][j][k] = v->MinDPPCLKUsingSingleDPP[k] * (1.0 + v->DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) / 2.0; |
4033 | } |
4034 | v->RequiredDISPCLK[i][j] = dml_max(v->RequiredDISPCLK[i][j], v->PlaneRequiredDISPCLK); |
4035 | if ((v->MinDPPCLKUsingSingleDPP[k] / v->NoOfDPP[i][j][k] * (1.0 + v->DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) |
4036 | > v->MaxDppclkRoundedDownToDFSGranularity) || (v->PlaneRequiredDISPCLK > v->MaxDispclkRoundedDownToDFSGranularity)) { |
4037 | v->DISPCLK_DPPCLK_Support[i][j] = false0; |
4038 | } |
4039 | } |
4040 | v->TotalNumberOfActiveDPP[i][j] = 0; |
4041 | v->TotalNumberOfSingleDPPPlanes[i][j] = 0; |
4042 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
4043 | v->TotalNumberOfActiveDPP[i][j] = v->TotalNumberOfActiveDPP[i][j] + v->NoOfDPP[i][j][k]; |
4044 | if (v->NoOfDPP[i][j][k] == 1) |
4045 | v->TotalNumberOfSingleDPPPlanes[i][j] = v->TotalNumberOfSingleDPPPlanes[i][j] + 1; |
4046 | } |
4047 | if (j == 1 && v->WhenToDoMPCCombine != dm_mpc_never) { |
4048 | while (!(v->TotalNumberOfActiveDPP[i][j] >= v->MaxNumDPP || v->TotalNumberOfSingleDPPPlanes[i][j] == 0)) { |
4049 | double BWOfNonSplitPlaneOfMaximumBandwidth = 0; |
4050 | unsigned int NumberOfNonSplitPlaneOfMaximumBandwidth = 0; |
4051 | BWOfNonSplitPlaneOfMaximumBandwidth = 0; |
4052 | NumberOfNonSplitPlaneOfMaximumBandwidth = 0; |
4053 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
4054 | if (v->ReadBandwidthLuma[k] + v->ReadBandwidthChroma[k] > BWOfNonSplitPlaneOfMaximumBandwidth |
4055 | && v->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_disabled && v->MPCCombine[i][j][k] == false0) { |
4056 | BWOfNonSplitPlaneOfMaximumBandwidth = v->ReadBandwidthLuma[k] + v->ReadBandwidthChroma[k]; |
4057 | NumberOfNonSplitPlaneOfMaximumBandwidth = k; |
4058 | } |
4059 | } |
4060 | v->MPCCombine[i][j][NumberOfNonSplitPlaneOfMaximumBandwidth] = true1; |
4061 | v->NoOfDPP[i][j][NumberOfNonSplitPlaneOfMaximumBandwidth] = 2; |
4062 | v->RequiredDPPCLK[i][j][NumberOfNonSplitPlaneOfMaximumBandwidth] = v->MinDPPCLKUsingSingleDPP[NumberOfNonSplitPlaneOfMaximumBandwidth] |
4063 | * (1 + v->DISPCLKDPPCLKDSCCLKDownSpreading / 100) / 2; |
4064 | v->TotalNumberOfActiveDPP[i][j] = v->TotalNumberOfActiveDPP[i][j] + 1; |
4065 | v->TotalNumberOfSingleDPPPlanes[i][j] = v->TotalNumberOfSingleDPPPlanes[i][j] + 1; |
4066 | } |
4067 | } |
4068 | if (v->TotalNumberOfActiveDPP[i][j] > v->MaxNumDPP) { |
4069 | v->RequiredDISPCLK[i][j] = 0.0; |
4070 | v->DISPCLK_DPPCLK_Support[i][j] = true1; |
4071 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
4072 | v->ODMCombineEnablePerState[i][k] = dm_odm_combine_mode_disabled; |
4073 | if (v->SingleDPPViewportSizeSupportPerPlane[k] == false0 && v->WhenToDoMPCCombine != dm_mpc_never) { |
4074 | v->MPCCombine[i][j][k] = true1; |
4075 | v->NoOfDPP[i][j][k] = 2; |
4076 | v->RequiredDPPCLK[i][j][k] = v->MinDPPCLKUsingSingleDPP[k] * (1.0 + v->DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) / 2.0; |
4077 | } else { |
4078 | v->MPCCombine[i][j][k] = false0; |
4079 | v->NoOfDPP[i][j][k] = 1; |
4080 | v->RequiredDPPCLK[i][j][k] = v->MinDPPCLKUsingSingleDPP[k] * (1.0 + v->DISPCLKDPPCLKDSCCLKDownSpreading / 100.0); |
4081 | } |
4082 | if (!(v->MaxDispclk[i] == v->MaxDispclk[v->soc.num_states - 1] && v->MaxDppclk[i] == v->MaxDppclk[v->soc.num_states - 1])) { |
4083 | v->PlaneRequiredDISPCLK = v->PixelClock[k] * (1.0 + v->DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) |
4084 | * (1.0 + v->DISPCLKRampingMargin / 100.0); |
4085 | } else { |
4086 | v->PlaneRequiredDISPCLK = v->PixelClock[k] * (1.0 + v->DISPCLKDPPCLKDSCCLKDownSpreading / 100.0); |
4087 | } |
4088 | v->RequiredDISPCLK[i][j] = dml_max(v->RequiredDISPCLK[i][j], v->PlaneRequiredDISPCLK); |
4089 | if ((v->MinDPPCLKUsingSingleDPP[k] / v->NoOfDPP[i][j][k] * (1.0 + v->DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) |
4090 | > v->MaxDppclkRoundedDownToDFSGranularity) || (v->PlaneRequiredDISPCLK > v->MaxDispclkRoundedDownToDFSGranularity)) { |
4091 | v->DISPCLK_DPPCLK_Support[i][j] = false0; |
4092 | } |
4093 | } |
4094 | v->TotalNumberOfActiveDPP[i][j] = 0.0; |
4095 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
4096 | v->TotalNumberOfActiveDPP[i][j] = v->TotalNumberOfActiveDPP[i][j] + v->NoOfDPP[i][j][k]; |
4097 | } |
4098 | } |
4099 | v->RequiredDISPCLK[i][j] = dml_max(v->RequiredDISPCLK[i][j], v->WritebackRequiredDISPCLK); |
4100 | if (v->MaxDispclkRoundedDownToDFSGranularity < v->WritebackRequiredDISPCLK) { |
4101 | v->DISPCLK_DPPCLK_Support[i][j] = false0; |
4102 | } |
4103 | } |
4104 | } |
4105 | |
4106 | /*Total Available Pipes Support Check*/ |
4107 | |
4108 | for (i = 0; i < v->soc.num_states; i++) { |
4109 | for (j = 0; j < 2; j++) { |
4110 | if (v->TotalNumberOfActiveDPP[i][j] <= v->MaxNumDPP) { |
4111 | v->TotalAvailablePipesSupport[i][j] = true1; |
4112 | } else { |
4113 | v->TotalAvailablePipesSupport[i][j] = false0; |
4114 | } |
4115 | } |
4116 | } |
4117 | /*Display IO and DSC Support Check*/ |
4118 | |
4119 | v->NonsupportedDSCInputBPC = false0; |
4120 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
4121 | if (!(v->DSCInputBitPerComponent[k] == 12.0 |
4122 | || v->DSCInputBitPerComponent[k] == 10.0 |
4123 | || v->DSCInputBitPerComponent[k] == 8.0)) { |
4124 | v->NonsupportedDSCInputBPC = true1; |
4125 | } |
4126 | } |
4127 | |
4128 | /*Number Of DSC Slices*/ |
4129 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
4130 | if (v->BlendingAndTiming[k] == k) { |
4131 | if (v->PixelClockBackEnd[k] > 3200) { |
4132 | v->NumberOfDSCSlices[k] = dml_ceil(v->PixelClockBackEnd[k] / 400.0, 4.0); |
4133 | } else if (v->PixelClockBackEnd[k] > 1360) { |
4134 | v->NumberOfDSCSlices[k] = 8; |
4135 | } else if (v->PixelClockBackEnd[k] > 680) { |
4136 | v->NumberOfDSCSlices[k] = 4; |
4137 | } else if (v->PixelClockBackEnd[k] > 340) { |
4138 | v->NumberOfDSCSlices[k] = 2; |
4139 | } else { |
4140 | v->NumberOfDSCSlices[k] = 1; |
4141 | } |
4142 | } else { |
4143 | v->NumberOfDSCSlices[k] = 0; |
4144 | } |
4145 | } |
4146 | |
4147 | for (i = 0; i < v->soc.num_states; i++) { |
4148 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
4149 | v->RequiresDSC[i][k] = false0; |
4150 | v->RequiresFEC[i][k] = false0; |
4151 | if (v->BlendingAndTiming[k] == k) { |
4152 | if (v->Output[k] == dm_hdmi) { |
4153 | v->RequiresDSC[i][k] = false0; |
4154 | v->RequiresFEC[i][k] = false0; |
4155 | v->OutputBppPerState[i][k] = TruncToValidBPP( |
4156 | dml_min(600.0, v->PHYCLKPerState[i]) * 10, |
4157 | 3, |
4158 | v->HTotal[k], |
4159 | v->HActive[k], |
4160 | v->PixelClockBackEnd[k], |
4161 | v->ForcedOutputLinkBPP[k], |
4162 | false0, |
4163 | v->Output[k], |
4164 | v->OutputFormat[k], |
4165 | v->DSCInputBitPerComponent[k], |
4166 | v->NumberOfDSCSlices[k], |
4167 | v->AudioSampleRate[k], |
4168 | v->AudioSampleLayout[k], |
4169 | v->ODMCombineEnablePerState[i][k]); |
4170 | } else if (v->Output[k] == dm_dp || v->Output[k] == dm_edp) { |
4171 | if (v->DSCEnable[k] == true1) { |
4172 | v->RequiresDSC[i][k] = true1; |
4173 | v->LinkDSCEnable = true1; |
4174 | if (v->Output[k] == dm_dp) { |
4175 | v->RequiresFEC[i][k] = true1; |
4176 | } else { |
4177 | v->RequiresFEC[i][k] = false0; |
4178 | } |
4179 | } else { |
4180 | v->RequiresDSC[i][k] = false0; |
4181 | v->LinkDSCEnable = false0; |
4182 | v->RequiresFEC[i][k] = false0; |
4183 | } |
4184 | |
4185 | v->Outbpp = BPP_INVALID0; |
4186 | if (v->PHYCLKPerState[i] >= 270.0) { |
4187 | v->Outbpp = TruncToValidBPP( |
4188 | (1.0 - v->Downspreading / 100.0) * 2700, |
4189 | v->OutputLinkDPLanes[k], |
4190 | v->HTotal[k], |
4191 | v->HActive[k], |
4192 | v->PixelClockBackEnd[k], |
4193 | v->ForcedOutputLinkBPP[k], |
4194 | v->LinkDSCEnable, |
4195 | v->Output[k], |
4196 | v->OutputFormat[k], |
4197 | v->DSCInputBitPerComponent[k], |
4198 | v->NumberOfDSCSlices[k], |
4199 | v->AudioSampleRate[k], |
4200 | v->AudioSampleLayout[k], |
4201 | v->ODMCombineEnablePerState[i][k]); |
4202 | v->OutputBppPerState[i][k] = v->Outbpp; |
4203 | // TODO: Need some other way to handle this nonsense |
4204 | // v->OutputTypeAndRatePerState[i][k] = v->Output[k] & " HBR" |
4205 | } |
4206 | if (v->Outbpp == BPP_INVALID0 && v->PHYCLKPerState[i] >= 540.0) { |
4207 | v->Outbpp = TruncToValidBPP( |
4208 | (1.0 - v->Downspreading / 100.0) * 5400, |
4209 | v->OutputLinkDPLanes[k], |
4210 | v->HTotal[k], |
4211 | v->HActive[k], |
4212 | v->PixelClockBackEnd[k], |
4213 | v->ForcedOutputLinkBPP[k], |
4214 | v->LinkDSCEnable, |
4215 | v->Output[k], |
4216 | v->OutputFormat[k], |
4217 | v->DSCInputBitPerComponent[k], |
4218 | v->NumberOfDSCSlices[k], |
4219 | v->AudioSampleRate[k], |
4220 | v->AudioSampleLayout[k], |
4221 | v->ODMCombineEnablePerState[i][k]); |
4222 | v->OutputBppPerState[i][k] = v->Outbpp; |
4223 | // TODO: Need some other way to handle this nonsense |
4224 | // v->OutputTypeAndRatePerState[i][k] = v->Output[k] & " HBR2" |
4225 | } |
4226 | if (v->Outbpp == BPP_INVALID0 && v->PHYCLKPerState[i] >= 810.0) { |
4227 | v->Outbpp = TruncToValidBPP( |
4228 | (1.0 - v->Downspreading / 100.0) * 8100, |
4229 | v->OutputLinkDPLanes[k], |
4230 | v->HTotal[k], |
4231 | v->HActive[k], |
4232 | v->PixelClockBackEnd[k], |
4233 | v->ForcedOutputLinkBPP[k], |
4234 | v->LinkDSCEnable, |
4235 | v->Output[k], |
4236 | v->OutputFormat[k], |
4237 | v->DSCInputBitPerComponent[k], |
4238 | v->NumberOfDSCSlices[k], |
4239 | v->AudioSampleRate[k], |
4240 | v->AudioSampleLayout[k], |
4241 | v->ODMCombineEnablePerState[i][k]); |
4242 | if (v->Outbpp == BPP_INVALID0 && v->ForcedOutputLinkBPP[k] == 0) { |
4243 | //if (v->Outbpp == BPP_INVALID && v->DSCEnabled[k] == dm_dsc_enable_only_if_necessary && v->ForcedOutputLinkBPP[k] == 0) { |
4244 | v->RequiresDSC[i][k] = true1; |
4245 | v->LinkDSCEnable = true1; |
4246 | if (v->Output[k] == dm_dp) { |
4247 | v->RequiresFEC[i][k] = true1; |
4248 | } |
4249 | v->Outbpp = TruncToValidBPP( |
4250 | (1.0 - v->Downspreading / 100.0) * 8100, |
4251 | v->OutputLinkDPLanes[k], |
4252 | v->HTotal[k], |
4253 | v->HActive[k], |
4254 | v->PixelClockBackEnd[k], |
4255 | v->ForcedOutputLinkBPP[k], |
4256 | v->LinkDSCEnable, |
4257 | v->Output[k], |
4258 | v->OutputFormat[k], |
4259 | v->DSCInputBitPerComponent[k], |
4260 | v->NumberOfDSCSlices[k], |
4261 | v->AudioSampleRate[k], |
4262 | v->AudioSampleLayout[k], |
4263 | v->ODMCombineEnablePerState[i][k]); |
4264 | } |
4265 | v->OutputBppPerState[i][k] = v->Outbpp; |
4266 | // TODO: Need some other way to handle this nonsense |
4267 | // v->OutputTypeAndRatePerState[i][k] = v->Output[k] & " HBR3" |
4268 | } |
4269 | } |
4270 | } else { |
4271 | v->OutputBppPerState[i][k] = 0; |
4272 | } |
4273 | } |
4274 | } |
4275 | for (i = 0; i < v->soc.num_states; i++) { |
4276 | v->DIOSupport[i] = true1; |
4277 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
4278 | if (v->BlendingAndTiming[k] == k && (v->Output[k] == dm_dp || v->Output[k] == dm_edp || v->Output[k] == dm_hdmi) |
4279 | && (v->OutputBppPerState[i][k] == 0 |
4280 | || (v->OutputFormat[k] == dm_420 && v->Interlace[k] == true1 && v->ProgressiveToInterlaceUnitInOPP == true1))) { |
4281 | v->DIOSupport[i] = false0; |
4282 | } |
4283 | } |
4284 | } |
4285 | |
4286 | for (i = 0; i < v->soc.num_states; ++i) { |
4287 | v->ODMCombine4To1SupportCheckOK[i] = true1; |
4288 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
4289 | if (v->BlendingAndTiming[k] == k && v->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_4to1 |
4290 | && (v->ODMCombine4To1Supported == false0 || v->Output[k] == dm_dp || v->Output[k] == dm_edp || v->Output[k] == dm_hdmi)) { |
4291 | v->ODMCombine4To1SupportCheckOK[i] = false0; |
4292 | } |
4293 | } |
4294 | } |
4295 | |
4296 | /* Skip dscclk validation: as long as dispclk is supported, dscclk is also implicitly supported */ |
4297 | |
4298 | for (i = 0; i < v->soc.num_states; i++) { |
4299 | v->NotEnoughDSCUnits[i] = false0; |
4300 | v->TotalDSCUnitsRequired = 0.0; |
4301 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
4302 | if (v->RequiresDSC[i][k] == true1) { |
4303 | if (v->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_4to1) { |
4304 | v->TotalDSCUnitsRequired = v->TotalDSCUnitsRequired + 4.0; |
4305 | } else if (v->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_2to1) { |
4306 | v->TotalDSCUnitsRequired = v->TotalDSCUnitsRequired + 2.0; |
4307 | } else { |
4308 | v->TotalDSCUnitsRequired = v->TotalDSCUnitsRequired + 1.0; |
4309 | } |
4310 | } |
4311 | } |
4312 | if (v->TotalDSCUnitsRequired > v->NumberOfDSC) { |
4313 | v->NotEnoughDSCUnits[i] = true1; |
4314 | } |
4315 | } |
4316 | /*DSC Delay per state*/ |
4317 | |
4318 | for (i = 0; i < v->soc.num_states; i++) { |
4319 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
4320 | if (v->OutputBppPerState[i][k] == BPP_INVALID0) { |
4321 | v->BPP = 0.0; |
4322 | } else { |
4323 | v->BPP = v->OutputBppPerState[i][k]; |
4324 | } |
4325 | if (v->RequiresDSC[i][k] == true1 && v->BPP != 0.0) { |
4326 | if (v->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_disabled) { |
4327 | v->DSCDelayPerState[i][k] = dscceComputeDelay( |
4328 | v->DSCInputBitPerComponent[k], |
4329 | v->BPP, |
4330 | dml_ceil(1.0 * v->HActive[k] / v->NumberOfDSCSlices[k], 1.0), |
4331 | v->NumberOfDSCSlices[k], |
4332 | v->OutputFormat[k], |
4333 | v->Output[k]) + dscComputeDelay(v->OutputFormat[k], v->Output[k]); |
4334 | } else if (v->ODMCombineEnablePerState[i][k] == dm_odm_combine_mode_2to1) { |
4335 | v->DSCDelayPerState[i][k] = 2.0 |
4336 | * dscceComputeDelay( |
4337 | v->DSCInputBitPerComponent[k], |
4338 | v->BPP, |
4339 | dml_ceil(1.0 * v->HActive[k] / v->NumberOfDSCSlices[k], 1.0), |
4340 | v->NumberOfDSCSlices[k] / 2, |
4341 | v->OutputFormat[k], |
4342 | v->Output[k]) + dscComputeDelay(v->OutputFormat[k], v->Output[k]); |
4343 | } else { |
4344 | v->DSCDelayPerState[i][k] = 4.0 |
4345 | * (dscceComputeDelay( |
4346 | v->DSCInputBitPerComponent[k], |
4347 | v->BPP, |
4348 | dml_ceil(1.0 * v->HActive[k] / v->NumberOfDSCSlices[k], 1.0), |
4349 | v->NumberOfDSCSlices[k] / 4, |
4350 | v->OutputFormat[k], |
4351 | v->Output[k]) + dscComputeDelay(v->OutputFormat[k], v->Output[k])); |
4352 | } |
4353 | v->DSCDelayPerState[i][k] = v->DSCDelayPerState[i][k] * v->PixelClock[k] / v->PixelClockBackEnd[k]; |
4354 | } else { |
4355 | v->DSCDelayPerState[i][k] = 0.0; |
4356 | } |
4357 | } |
4358 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
4359 | for (m = 0; m <= v->NumberOfActivePlanes - 1; m++) { |
4360 | if (v->BlendingAndTiming[k] == m && v->RequiresDSC[i][m] == true1) { |
4361 | v->DSCDelayPerState[i][k] = v->DSCDelayPerState[i][m]; |
4362 | } |
4363 | } |
4364 | } |
4365 | } |
4366 | |
4367 | //Calculate Swath, DET Configuration, DCFCLKDeepSleep |
4368 | // |
4369 | for (i = 0; i < mode_lib->soc.num_states; ++i) { |
4370 | for (j = 0; j <= 1; ++j) { |
4371 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
4372 | v->RequiredDPPCLKThisState[k] = v->RequiredDPPCLK[i][j][k]; |
4373 | v->NoOfDPPThisState[k] = v->NoOfDPP[i][j][k]; |
4374 | v->ODMCombineEnableThisState[k] = v->ODMCombineEnablePerState[i][k]; |
4375 | } |
4376 | |
4377 | CalculateSwathAndDETConfiguration( |
4378 | false0, |
4379 | v->NumberOfActivePlanes, |
4380 | v->DETBufferSizeInKByte, |
4381 | v->MaximumSwathWidthLuma, |
4382 | v->MaximumSwathWidthChroma, |
4383 | v->SourceScan, |
4384 | v->SourcePixelFormat, |
4385 | v->SurfaceTiling, |
4386 | v->ViewportWidth, |
4387 | v->ViewportHeight, |
4388 | v->SurfaceWidthY, |
4389 | v->SurfaceWidthC, |
4390 | v->SurfaceHeightY, |
4391 | v->SurfaceHeightC, |
4392 | v->Read256BlockHeightY, |
4393 | v->Read256BlockHeightC, |
4394 | v->Read256BlockWidthY, |
4395 | v->Read256BlockWidthC, |
4396 | v->ODMCombineEnableThisState, |
4397 | v->BlendingAndTiming, |
4398 | v->BytePerPixelY, |
4399 | v->BytePerPixelC, |
4400 | v->BytePerPixelInDETY, |
4401 | v->BytePerPixelInDETC, |
4402 | v->HActive, |
4403 | v->HRatio, |
4404 | v->HRatioChroma, |
4405 | v->NoOfDPPThisState, |
4406 | v->swath_width_luma_ub_this_state, |
4407 | v->swath_width_chroma_ub_this_state, |
4408 | v->SwathWidthYThisState, |
4409 | v->SwathWidthCThisState, |
4410 | v->SwathHeightYThisState, |
4411 | v->SwathHeightCThisState, |
4412 | v->DETBufferSizeYThisState, |
4413 | v->DETBufferSizeCThisState, |
4414 | v->dummystring, |
4415 | &v->ViewportSizeSupport[i][j]); |
4416 | |
4417 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
4418 | v->swath_width_luma_ub_all_states[i][j][k] = v->swath_width_luma_ub_this_state[k]; |
4419 | v->swath_width_chroma_ub_all_states[i][j][k] = v->swath_width_chroma_ub_this_state[k]; |
4420 | v->SwathWidthYAllStates[i][j][k] = v->SwathWidthYThisState[k]; |
4421 | v->SwathWidthCAllStates[i][j][k] = v->SwathWidthCThisState[k]; |
4422 | v->SwathHeightYAllStates[i][j][k] = v->SwathHeightYThisState[k]; |
4423 | v->SwathHeightCAllStates[i][j][k] = v->SwathHeightCThisState[k]; |
4424 | v->DETBufferSizeYAllStates[i][j][k] = v->DETBufferSizeYThisState[k]; |
4425 | v->DETBufferSizeCAllStates[i][j][k] = v->DETBufferSizeCThisState[k]; |
4426 | } |
4427 | |
4428 | } |
4429 | } |
4430 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
4431 | v->cursor_bw[k] = v->NumberOfCursors[k] * v->CursorWidth[k][0] * v->CursorBPP[k][0] / 8.0 / (v->HTotal[k] / v->PixelClock[k]) * v->VRatio[k]; |
4432 | } |
4433 | |
4434 | for (i = 0; i < v->soc.num_states; i++) { |
4435 | for (j = 0; j < 2; j++) { |
4436 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
4437 | v->swath_width_luma_ub_this_state[k] = v->swath_width_luma_ub_all_states[i][j][k]; |
4438 | v->swath_width_chroma_ub_this_state[k] = v->swath_width_chroma_ub_all_states[i][j][k]; |
4439 | v->SwathWidthYThisState[k] = v->SwathWidthYAllStates[i][j][k]; |
4440 | v->SwathWidthCThisState[k] = v->SwathWidthCAllStates[i][j][k]; |
4441 | v->SwathHeightYThisState[k] = v->SwathHeightYAllStates[i][j][k]; |
4442 | v->SwathHeightCThisState[k] = v->SwathHeightCAllStates[i][j][k]; |
4443 | v->DETBufferSizeYThisState[k] = v->DETBufferSizeYAllStates[i][j][k]; |
4444 | v->DETBufferSizeCThisState[k] = v->DETBufferSizeCAllStates[i][j][k]; |
4445 | } |
4446 | |
4447 | v->TotalNumberOfDCCActiveDPP[i][j] = 0; |
4448 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
4449 | if (v->DCCEnable[k] == true1) { |
4450 | v->TotalNumberOfDCCActiveDPP[i][j] = v->TotalNumberOfDCCActiveDPP[i][j] + v->NoOfDPP[i][j][k]; |
4451 | } |
4452 | } |
4453 | |
4454 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
4455 | if (v->SourcePixelFormat[k] == dm_420_8 || v->SourcePixelFormat[k] == dm_420_10 || v->SourcePixelFormat[k] == dm_420_12 |
4456 | || v->SourcePixelFormat[k] == dm_rgbe_alpha) { |
4457 | |
4458 | if ((v->SourcePixelFormat[k] == dm_420_10 || v->SourcePixelFormat[k] == dm_420_12) && v->SourceScan[k] != dm_vert) { |
4459 | v->PTEBufferSizeInRequestsForLuma = (v->PTEBufferSizeInRequestsLuma + v->PTEBufferSizeInRequestsChroma) / 2; |
4460 | v->PTEBufferSizeInRequestsForChroma = v->PTEBufferSizeInRequestsForLuma; |
4461 | } else { |
4462 | v->PTEBufferSizeInRequestsForLuma = v->PTEBufferSizeInRequestsLuma; |
4463 | v->PTEBufferSizeInRequestsForChroma = v->PTEBufferSizeInRequestsChroma; |
4464 | } |
4465 | |
4466 | v->PDEAndMetaPTEBytesPerFrameC = CalculateVMAndRowBytes( |
4467 | mode_lib, |
4468 | v->DCCEnable[k], |
4469 | v->Read256BlockHeightC[k], |
4470 | v->Read256BlockWidthY[k], |
4471 | v->SourcePixelFormat[k], |
4472 | v->SurfaceTiling[k], |
4473 | v->BytePerPixelC[k], |
4474 | v->SourceScan[k], |
4475 | v->SwathWidthCThisState[k], |
4476 | v->ViewportHeightChroma[k], |
4477 | v->GPUVMEnable, |
4478 | v->HostVMEnable, |
4479 | v->HostVMMaxNonCachedPageTableLevels, |
4480 | v->GPUVMMinPageSize, |
4481 | v->HostVMMinPageSize, |
4482 | v->PTEBufferSizeInRequestsForChroma, |
4483 | v->PitchC[k], |
4484 | 0.0, |
4485 | &v->MacroTileWidthC[k], |
4486 | &v->MetaRowBytesC, |
4487 | &v->DPTEBytesPerRowC, |
4488 | &v->PTEBufferSizeNotExceededC[i][j][k], |
4489 | &v->dummyinteger7, |
4490 | &v->dpte_row_height_chroma[k], |
4491 | &v->dummyinteger28, |
4492 | &v->dummyinteger26, |
4493 | &v->dummyinteger23, |
4494 | &v->meta_row_height_chroma[k], |
4495 | &v->dummyinteger8, |
4496 | &v->dummyinteger9, |
4497 | &v->dummyinteger19, |
4498 | &v->dummyinteger20, |
4499 | &v->dummyinteger17, |
4500 | &v->dummyinteger10, |
4501 | &v->dummyinteger11); |
4502 | |
4503 | v->PrefetchLinesC[i][j][k] = CalculatePrefetchSourceLines( |
4504 | mode_lib, |
4505 | v->VRatioChroma[k], |
4506 | v->VTAPsChroma[k], |
4507 | v->Interlace[k], |
4508 | v->ProgressiveToInterlaceUnitInOPP, |
4509 | v->SwathHeightCThisState[k], |
4510 | v->ViewportYStartC[k], |
4511 | &v->PrefillC[k], |
4512 | &v->MaxNumSwC[k]); |
4513 | } else { |
4514 | v->PTEBufferSizeInRequestsForLuma = v->PTEBufferSizeInRequestsLuma + v->PTEBufferSizeInRequestsChroma; |
4515 | v->PTEBufferSizeInRequestsForChroma = 0; |
4516 | v->PDEAndMetaPTEBytesPerFrameC = 0.0; |
4517 | v->MetaRowBytesC = 0.0; |
4518 | v->DPTEBytesPerRowC = 0.0; |
4519 | v->PrefetchLinesC[i][j][k] = 0.0; |
4520 | v->PTEBufferSizeNotExceededC[i][j][k] = true1; |
4521 | } |
4522 | v->PDEAndMetaPTEBytesPerFrameY = CalculateVMAndRowBytes( |
4523 | mode_lib, |
4524 | v->DCCEnable[k], |
4525 | v->Read256BlockHeightY[k], |
4526 | v->Read256BlockWidthY[k], |
4527 | v->SourcePixelFormat[k], |
4528 | v->SurfaceTiling[k], |
4529 | v->BytePerPixelY[k], |
4530 | v->SourceScan[k], |
4531 | v->SwathWidthYThisState[k], |
4532 | v->ViewportHeight[k], |
4533 | v->GPUVMEnable, |
4534 | v->HostVMEnable, |
4535 | v->HostVMMaxNonCachedPageTableLevels, |
4536 | v->GPUVMMinPageSize, |
4537 | v->HostVMMinPageSize, |
4538 | v->PTEBufferSizeInRequestsForLuma, |
4539 | v->PitchY[k], |
4540 | v->DCCMetaPitchY[k], |
4541 | &v->MacroTileWidthY[k], |
4542 | &v->MetaRowBytesY, |
4543 | &v->DPTEBytesPerRowY, |
4544 | &v->PTEBufferSizeNotExceededY[i][j][k], |
4545 | v->dummyinteger4, |
4546 | &v->dpte_row_height[k], |
4547 | &v->dummyinteger29, |
4548 | &v->dummyinteger27, |
4549 | &v->dummyinteger24, |
4550 | &v->meta_row_height[k], |
4551 | &v->dummyinteger25, |
4552 | &v->dpte_group_bytes[k], |
4553 | &v->dummyinteger21, |
4554 | &v->dummyinteger22, |
4555 | &v->dummyinteger18, |
4556 | &v->dummyinteger5, |
4557 | &v->dummyinteger6); |
4558 | v->PrefetchLinesY[i][j][k] = CalculatePrefetchSourceLines( |
4559 | mode_lib, |
4560 | v->VRatio[k], |
4561 | v->vtaps[k], |
4562 | v->Interlace[k], |
4563 | v->ProgressiveToInterlaceUnitInOPP, |
4564 | v->SwathHeightYThisState[k], |
4565 | v->ViewportYStartY[k], |
4566 | &v->PrefillY[k], |
4567 | &v->MaxNumSwY[k]); |
4568 | v->PDEAndMetaPTEBytesPerFrame[i][j][k] = v->PDEAndMetaPTEBytesPerFrameY + v->PDEAndMetaPTEBytesPerFrameC; |
4569 | v->MetaRowBytes[i][j][k] = v->MetaRowBytesY + v->MetaRowBytesC; |
4570 | v->DPTEBytesPerRow[i][j][k] = v->DPTEBytesPerRowY + v->DPTEBytesPerRowC; |
4571 | |
4572 | CalculateRowBandwidth( |
4573 | v->GPUVMEnable, |
4574 | v->SourcePixelFormat[k], |
4575 | v->VRatio[k], |
4576 | v->VRatioChroma[k], |
4577 | v->DCCEnable[k], |
4578 | v->HTotal[k] / v->PixelClock[k], |
4579 | v->MetaRowBytesY, |
4580 | v->MetaRowBytesC, |
4581 | v->meta_row_height[k], |
4582 | v->meta_row_height_chroma[k], |
4583 | v->DPTEBytesPerRowY, |
4584 | v->DPTEBytesPerRowC, |
4585 | v->dpte_row_height[k], |
4586 | v->dpte_row_height_chroma[k], |
4587 | &v->meta_row_bandwidth[i][j][k], |
4588 | &v->dpte_row_bandwidth[i][j][k]); |
4589 | } |
4590 | v->UrgLatency[i] = CalculateUrgentLatency( |
4591 | v->UrgentLatencyPixelDataOnly, |
4592 | v->UrgentLatencyPixelMixedWithVMData, |
4593 | v->UrgentLatencyVMDataOnly, |
4594 | v->DoUrgentLatencyAdjustment, |
4595 | v->UrgentLatencyAdjustmentFabricClockComponent, |
4596 | v->UrgentLatencyAdjustmentFabricClockReference, |
4597 | v->FabricClockPerState[i]); |
4598 | |
4599 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
4600 | CalculateUrgentBurstFactor( |
4601 | v->swath_width_luma_ub_this_state[k], |
4602 | v->swath_width_chroma_ub_this_state[k], |
4603 | v->DETBufferSizeInKByte, |
4604 | v->SwathHeightYThisState[k], |
4605 | v->SwathHeightCThisState[k], |
4606 | v->HTotal[k] / v->PixelClock[k], |
4607 | v->UrgLatency[i], |
4608 | v->CursorBufferSize, |
4609 | v->CursorWidth[k][0], |
4610 | v->CursorBPP[k][0], |
4611 | v->VRatio[k], |
4612 | v->VRatioChroma[k], |
4613 | v->BytePerPixelInDETY[k], |
4614 | v->BytePerPixelInDETC[k], |
4615 | v->DETBufferSizeYThisState[k], |
4616 | v->DETBufferSizeCThisState[k], |
4617 | &v->UrgentBurstFactorCursor[k], |
4618 | &v->UrgentBurstFactorLuma[k], |
4619 | &v->UrgentBurstFactorChroma[k], |
4620 | &NotUrgentLatencyHiding[k]); |
4621 | } |
4622 | |
4623 | v->NotUrgentLatencyHiding[i][j] = false0; |
4624 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
4625 | if (NotUrgentLatencyHiding[k]) { |
4626 | v->NotUrgentLatencyHiding[i][j] = true1; |
4627 | } |
4628 | } |
4629 | |
4630 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
4631 | v->VActivePixelBandwidth[i][j][k] = v->ReadBandwidthLuma[k] * v->UrgentBurstFactorLuma[k] |
4632 | + v->ReadBandwidthChroma[k] * v->UrgentBurstFactorChroma[k]; |
4633 | v->VActiveCursorBandwidth[i][j][k] = v->cursor_bw[k] * v->UrgentBurstFactorCursor[k]; |
4634 | } |
4635 | |
4636 | v->TotalVActivePixelBandwidth[i][j] = 0; |
4637 | v->TotalVActiveCursorBandwidth[i][j] = 0; |
4638 | v->TotalMetaRowBandwidth[i][j] = 0; |
4639 | v->TotalDPTERowBandwidth[i][j] = 0; |
4640 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
4641 | v->TotalVActivePixelBandwidth[i][j] = v->TotalVActivePixelBandwidth[i][j] + v->VActivePixelBandwidth[i][j][k]; |
4642 | v->TotalVActiveCursorBandwidth[i][j] = v->TotalVActiveCursorBandwidth[i][j] + v->VActiveCursorBandwidth[i][j][k]; |
4643 | v->TotalMetaRowBandwidth[i][j] = v->TotalMetaRowBandwidth[i][j] + v->NoOfDPP[i][j][k] * v->meta_row_bandwidth[i][j][k]; |
4644 | v->TotalDPTERowBandwidth[i][j] = v->TotalDPTERowBandwidth[i][j] + v->NoOfDPP[i][j][k] * v->dpte_row_bandwidth[i][j][k]; |
4645 | } |
4646 | |
4647 | CalculateDCFCLKDeepSleep( |
4648 | mode_lib, |
4649 | v->NumberOfActivePlanes, |
4650 | v->BytePerPixelY, |
4651 | v->BytePerPixelC, |
4652 | v->VRatio, |
4653 | v->VRatioChroma, |
4654 | v->SwathWidthYThisState, |
4655 | v->SwathWidthCThisState, |
4656 | v->NoOfDPPThisState, |
4657 | v->HRatio, |
4658 | v->HRatioChroma, |
4659 | v->PixelClock, |
4660 | v->PSCL_FACTOR, |
4661 | v->PSCL_FACTOR_CHROMA, |
4662 | v->RequiredDPPCLKThisState, |
4663 | v->ReadBandwidthLuma, |
4664 | v->ReadBandwidthChroma, |
4665 | v->ReturnBusWidth, |
4666 | &v->ProjectedDCFCLKDeepSleep[i][j]); |
4667 | } |
4668 | } |
4669 | |
4670 | //Calculate Return BW |
4671 | |
4672 | for (i = 0; i < mode_lib->soc.num_states; ++i) { |
4673 | for (j = 0; j <= 1; ++j) { |
4674 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
4675 | if (v->BlendingAndTiming[k] == k) { |
4676 | if (v->WritebackEnable[k] == true1) { |
4677 | v->WritebackDelayTime[k] = v->WritebackLatency |
4678 | + CalculateWriteBackDelay( |
4679 | v->WritebackPixelFormat[k], |
4680 | v->WritebackHRatio[k], |
4681 | v->WritebackVRatio[k], |
4682 | v->WritebackVTaps[k], |
4683 | v->WritebackDestinationWidth[k], |
4684 | v->WritebackDestinationHeight[k], |
4685 | v->WritebackSourceHeight[k], |
4686 | v->HTotal[k]) / v->RequiredDISPCLK[i][j]; |
4687 | } else { |
4688 | v->WritebackDelayTime[k] = 0.0; |
4689 | } |
4690 | for (m = 0; m <= v->NumberOfActivePlanes - 1; m++) { |
4691 | if (v->BlendingAndTiming[m] == k && v->WritebackEnable[m] == true1) { |
4692 | v->WritebackDelayTime[k] = dml_max( |
4693 | v->WritebackDelayTime[k], |
4694 | v->WritebackLatency |
4695 | + CalculateWriteBackDelay( |
4696 | v->WritebackPixelFormat[m], |
4697 | v->WritebackHRatio[m], |
4698 | v->WritebackVRatio[m], |
4699 | v->WritebackVTaps[m], |
4700 | v->WritebackDestinationWidth[m], |
4701 | v->WritebackDestinationHeight[m], |
4702 | v->WritebackSourceHeight[m], |
4703 | v->HTotal[m]) / v->RequiredDISPCLK[i][j]); |
4704 | } |
4705 | } |
4706 | } |
4707 | } |
4708 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
4709 | for (m = 0; m <= v->NumberOfActivePlanes - 1; m++) { |
4710 | if (v->BlendingAndTiming[k] == m) { |
4711 | v->WritebackDelayTime[k] = v->WritebackDelayTime[m]; |
4712 | } |
4713 | } |
4714 | } |
4715 | v->MaxMaxVStartup[i][j] = 0; |
4716 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
4717 | v->MaximumVStartup[i][j][k] = v->VTotal[k] - v->VActive[k] |
4718 | - dml_max(1.0, dml_ceil(1.0 * v->WritebackDelayTime[k] / (v->HTotal[k] / v->PixelClock[k]), 1.0)); |
4719 | v->MaxMaxVStartup[i][j] = dml_max(v->MaxMaxVStartup[i][j], v->MaximumVStartup[i][j][k]); |
4720 | } |
4721 | } |
4722 | } |
4723 | |
4724 | ReorderingBytes = v->NumberOfChannels |
4725 | * dml_max3( |
4726 | v->UrgentOutOfOrderReturnPerChannelPixelDataOnly, |
4727 | v->UrgentOutOfOrderReturnPerChannelPixelMixedWithVMData, |
4728 | v->UrgentOutOfOrderReturnPerChannelVMDataOnly); |
4729 | v->FinalDRAMClockChangeLatency = (v->DRAMClockChangeLatencyOverride > 0 ? v->DRAMClockChangeLatencyOverride : v->DRAMClockChangeLatency); |
4730 | |
4731 | for (i = 0; i < mode_lib->soc.num_states; ++i) { |
4732 | for (j = 0; j <= 1; ++j) { |
4733 | v->DCFCLKState[i][j] = v->DCFCLKPerState[i]; |
4734 | } |
4735 | } |
4736 | |
4737 | if (v->UseMinimumRequiredDCFCLK == true1) { |
4738 | UseMinimumDCFCLK( |
4739 | mode_lib, |
4740 | v->MaxInterDCNTileRepeaters, |
4741 | MaxPrefetchMode, |
4742 | v->FinalDRAMClockChangeLatency, |
4743 | v->SREnterPlusExitTime, |
4744 | v->ReturnBusWidth, |
4745 | v->RoundTripPingLatencyCycles, |
4746 | ReorderingBytes, |
4747 | v->PixelChunkSizeInKByte, |
4748 | v->MetaChunkSize, |
4749 | v->GPUVMEnable, |
4750 | v->GPUVMMaxPageTableLevels, |
4751 | v->HostVMEnable, |
4752 | v->NumberOfActivePlanes, |
4753 | v->HostVMMinPageSize, |
4754 | v->HostVMMaxNonCachedPageTableLevels, |
4755 | v->DynamicMetadataVMEnabled, |
4756 | v->ImmediateFlipRequirement, |
4757 | v->ProgressiveToInterlaceUnitInOPP, |
4758 | v->MaxAveragePercentOfIdealSDPPortBWDisplayCanUseInNormalSystemOperation, |
4759 | v->PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData, |
4760 | v->PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly, |
4761 | v->PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelDataOnly, |
4762 | v->VTotal, |
4763 | v->VActive, |
4764 | v->DynamicMetadataTransmittedBytes, |
4765 | v->DynamicMetadataLinesBeforeActiveRequired, |
4766 | v->Interlace, |
4767 | v->RequiredDPPCLK, |
4768 | v->RequiredDISPCLK, |
4769 | v->UrgLatency, |
4770 | v->NoOfDPP, |
4771 | v->ProjectedDCFCLKDeepSleep, |
4772 | v->MaximumVStartup, |
4773 | v->TotalVActivePixelBandwidth, |
4774 | v->TotalVActiveCursorBandwidth, |
4775 | v->TotalMetaRowBandwidth, |
4776 | v->TotalDPTERowBandwidth, |
4777 | v->TotalNumberOfActiveDPP, |
4778 | v->TotalNumberOfDCCActiveDPP, |
4779 | v->dpte_group_bytes, |
4780 | v->PrefetchLinesY, |
4781 | v->PrefetchLinesC, |
4782 | v->swath_width_luma_ub_all_states, |
4783 | v->swath_width_chroma_ub_all_states, |
4784 | v->BytePerPixelY, |
4785 | v->BytePerPixelC, |
4786 | v->HTotal, |
4787 | v->PixelClock, |
4788 | v->PDEAndMetaPTEBytesPerFrame, |
4789 | v->DPTEBytesPerRow, |
4790 | v->MetaRowBytes, |
4791 | v->DynamicMetadataEnable, |
4792 | v->VActivePixelBandwidth, |
4793 | v->VActiveCursorBandwidth, |
4794 | v->ReadBandwidthLuma, |
4795 | v->ReadBandwidthChroma, |
4796 | v->DCFCLKPerState, |
4797 | v->DCFCLKState); |
4798 | |
4799 | if (v->ClampMinDCFCLK) { |
4800 | /* Clamp calculated values to actual minimum */ |
4801 | for (i = 0; i < mode_lib->soc.num_states; ++i) { |
4802 | for (j = 0; j <= 1; ++j) { |
4803 | if (v->DCFCLKState[i][j] < mode_lib->soc.min_dcfclk) { |
4804 | v->DCFCLKState[i][j] = mode_lib->soc.min_dcfclk; |
4805 | } |
4806 | } |
4807 | } |
4808 | } |
4809 | } |
4810 | |
4811 | for (i = 0; i < mode_lib->soc.num_states; ++i) { |
4812 | for (j = 0; j <= 1; ++j) { |
4813 | v->IdealSDPPortBandwidthPerState[i][j] = dml_min3( |
4814 | v->ReturnBusWidth * v->DCFCLKState[i][j], |
4815 | v->DRAMSpeedPerState[i] * v->NumberOfChannels * v->DRAMChannelWidth, |
4816 | v->FabricClockPerState[i] * v->FabricDatapathToDCNDataReturn); |
4817 | if (v->HostVMEnable != true1) { |
4818 | v->ReturnBWPerState[i][j] = v->IdealSDPPortBandwidthPerState[i][j] * v->PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelDataOnly |
4819 | / 100; |
4820 | } else { |
4821 | v->ReturnBWPerState[i][j] = v->IdealSDPPortBandwidthPerState[i][j] |
4822 | * v->PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData / 100; |
4823 | } |
4824 | } |
4825 | } |
4826 | |
4827 | //Re-ordering Buffer Support Check |
4828 | |
4829 | for (i = 0; i < mode_lib->soc.num_states; ++i) { |
4830 | for (j = 0; j <= 1; ++j) { |
4831 | if ((v->ROBBufferSizeInKByte - v->PixelChunkSizeInKByte) * 1024 / v->ReturnBWPerState[i][j] |
4832 | > (v->RoundTripPingLatencyCycles + 32) / v->DCFCLKState[i][j] + ReorderingBytes / v->ReturnBWPerState[i][j]) { |
4833 | v->ROBSupport[i][j] = true1; |
4834 | } else { |
4835 | v->ROBSupport[i][j] = false0; |
4836 | } |
4837 | } |
4838 | } |
4839 | |
4840 | //Vertical Active BW support check |
4841 | |
4842 | MaxTotalVActiveRDBandwidth = 0; |
4843 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
4844 | MaxTotalVActiveRDBandwidth = MaxTotalVActiveRDBandwidth + v->ReadBandwidthLuma[k] + v->ReadBandwidthChroma[k]; |
4845 | } |
4846 | |
4847 | for (i = 0; i < mode_lib->soc.num_states; ++i) { |
4848 | for (j = 0; j <= 1; ++j) { |
4849 | v->MaxTotalVerticalActiveAvailableBandwidth[i][j] = dml_min( |
4850 | v->IdealSDPPortBandwidthPerState[i][j] * v->MaxAveragePercentOfIdealSDPPortBWDisplayCanUseInNormalSystemOperation / 100, |
4851 | v->DRAMSpeedPerState[i] * v->NumberOfChannels * v->DRAMChannelWidth * v->MaxAveragePercentOfIdealDRAMBWDisplayCanUseInNormalSystemOperation |
4852 | / 100); |
4853 | if (MaxTotalVActiveRDBandwidth <= v->MaxTotalVerticalActiveAvailableBandwidth[i][j]) { |
4854 | v->TotalVerticalActiveBandwidthSupport[i][j] = true1; |
4855 | } else { |
4856 | v->TotalVerticalActiveBandwidthSupport[i][j] = false0; |
4857 | } |
4858 | } |
4859 | } |
4860 | |
4861 | //Prefetch Check |
4862 | |
4863 | for (i = 0; i < mode_lib->soc.num_states; ++i) { |
4864 | for (j = 0; j <= 1; ++j) { |
4865 | int NextPrefetchModeState = MinPrefetchMode; |
4866 | |
4867 | v->TimeCalc = 24 / v->ProjectedDCFCLKDeepSleep[i][j]; |
4868 | |
4869 | v->BandwidthWithoutPrefetchSupported[i][j] = true1; |
4870 | if (v->TotalVActivePixelBandwidth[i][j] + v->TotalVActiveCursorBandwidth[i][j] + v->TotalMetaRowBandwidth[i][j] + v->TotalDPTERowBandwidth[i][j] |
4871 | > v->ReturnBWPerState[i][j] || v->NotUrgentLatencyHiding[i][j]) { |
4872 | v->BandwidthWithoutPrefetchSupported[i][j] = false0; |
4873 | } |
4874 | |
4875 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
4876 | v->NoOfDPPThisState[k] = v->NoOfDPP[i][j][k]; |
4877 | v->swath_width_luma_ub_this_state[k] = v->swath_width_luma_ub_all_states[i][j][k]; |
4878 | v->swath_width_chroma_ub_this_state[k] = v->swath_width_chroma_ub_all_states[i][j][k]; |
4879 | v->SwathWidthYThisState[k] = v->SwathWidthYAllStates[i][j][k]; |
4880 | v->SwathWidthCThisState[k] = v->SwathWidthCAllStates[i][j][k]; |
4881 | v->SwathHeightYThisState[k] = v->SwathHeightYAllStates[i][j][k]; |
4882 | v->SwathHeightCThisState[k] = v->SwathHeightCAllStates[i][j][k]; |
4883 | v->DETBufferSizeYThisState[k] = v->DETBufferSizeYAllStates[i][j][k]; |
4884 | v->DETBufferSizeCThisState[k] = v->DETBufferSizeCAllStates[i][j][k]; |
4885 | v->ODMCombineEnabled[k] = v->ODMCombineEnablePerState[i][k]; |
4886 | } |
4887 | |
4888 | v->ExtraLatency = CalculateExtraLatency( |
4889 | v->RoundTripPingLatencyCycles, |
4890 | ReorderingBytes, |
4891 | v->DCFCLKState[i][j], |
4892 | v->TotalNumberOfActiveDPP[i][j], |
4893 | v->PixelChunkSizeInKByte, |
4894 | v->TotalNumberOfDCCActiveDPP[i][j], |
4895 | v->MetaChunkSize, |
4896 | v->ReturnBWPerState[i][j], |
4897 | v->GPUVMEnable, |
4898 | v->HostVMEnable, |
4899 | v->NumberOfActivePlanes, |
4900 | v->NoOfDPPThisState, |
4901 | v->dpte_group_bytes, |
4902 | v->PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData, |
4903 | v->PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly, |
4904 | v->HostVMMinPageSize, |
4905 | v->HostVMMaxNonCachedPageTableLevels); |
4906 | |
4907 | v->NextMaxVStartup = v->MaxMaxVStartup[i][j]; |
4908 | do { |
4909 | v->PrefetchModePerState[i][j] = NextPrefetchModeState; |
4910 | v->MaxVStartup = v->NextMaxVStartup; |
4911 | |
4912 | v->TWait = CalculateTWait(v->PrefetchModePerState[i][j], v->FinalDRAMClockChangeLatency, v->UrgLatency[i], v->SREnterPlusExitTime); |
4913 | |
4914 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
4915 | Pipe myPipe = { 0 }; |
4916 | |
4917 | myPipe.DPPCLK = v->RequiredDPPCLK[i][j][k]; |
4918 | myPipe.DISPCLK = v->RequiredDISPCLK[i][j]; |
4919 | myPipe.PixelClock = v->PixelClock[k]; |
4920 | myPipe.DCFCLKDeepSleep = v->ProjectedDCFCLKDeepSleep[i][j]; |
4921 | myPipe.DPPPerPlane = v->NoOfDPP[i][j][k]; |
4922 | myPipe.ScalerEnabled = v->ScalerEnabled[k]; |
4923 | myPipe.SourceScan = v->SourceScan[k]; |
4924 | myPipe.BlockWidth256BytesY = v->Read256BlockWidthY[k]; |
4925 | myPipe.BlockHeight256BytesY = v->Read256BlockHeightY[k]; |
4926 | myPipe.BlockWidth256BytesC = v->Read256BlockWidthC[k]; |
4927 | myPipe.BlockHeight256BytesC = v->Read256BlockHeightC[k]; |
4928 | myPipe.InterlaceEnable = v->Interlace[k]; |
4929 | myPipe.NumberOfCursors = v->NumberOfCursors[k]; |
4930 | myPipe.VBlank = v->VTotal[k] - v->VActive[k]; |
4931 | myPipe.HTotal = v->HTotal[k]; |
4932 | myPipe.DCCEnable = v->DCCEnable[k]; |
4933 | myPipe.ODMCombineEnabled = !!v->ODMCombineEnabled[k]; |
4934 | |
4935 | v->NoTimeForPrefetch[i][j][k] = CalculatePrefetchSchedule( |
4936 | mode_lib, |
4937 | v->PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData, |
4938 | v->PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly, |
4939 | &myPipe, |
4940 | v->DSCDelayPerState[i][k], |
4941 | v->DPPCLKDelaySubtotal + v->DPPCLKDelayCNVCFormater, |
4942 | v->DPPCLKDelaySCL, |
4943 | v->DPPCLKDelaySCLLBOnly, |
4944 | v->DPPCLKDelayCNVCCursor, |
4945 | v->DISPCLKDelaySubtotal, |
4946 | v->SwathWidthYThisState[k] / v->HRatio[k], |
4947 | v->OutputFormat[k], |
4948 | v->MaxInterDCNTileRepeaters, |
4949 | dml_min(v->MaxVStartup, v->MaximumVStartup[i][j][k]), |
4950 | v->MaximumVStartup[i][j][k], |
4951 | v->GPUVMMaxPageTableLevels, |
4952 | v->GPUVMEnable, |
4953 | v->HostVMEnable, |
4954 | v->HostVMMaxNonCachedPageTableLevels, |
4955 | v->HostVMMinPageSize, |
4956 | v->DynamicMetadataEnable[k], |
4957 | v->DynamicMetadataVMEnabled, |
4958 | v->DynamicMetadataLinesBeforeActiveRequired[k], |
4959 | v->DynamicMetadataTransmittedBytes[k], |
4960 | v->UrgLatency[i], |
4961 | v->ExtraLatency, |
4962 | v->TimeCalc, |
4963 | v->PDEAndMetaPTEBytesPerFrame[i][j][k], |
4964 | v->MetaRowBytes[i][j][k], |
4965 | v->DPTEBytesPerRow[i][j][k], |
4966 | v->PrefetchLinesY[i][j][k], |
4967 | v->SwathWidthYThisState[k], |
4968 | v->BytePerPixelY[k], |
4969 | v->PrefillY[k], |
4970 | v->MaxNumSwY[k], |
4971 | v->PrefetchLinesC[i][j][k], |
4972 | v->SwathWidthCThisState[k], |
4973 | v->BytePerPixelC[k], |
4974 | v->PrefillC[k], |
4975 | v->MaxNumSwC[k], |
4976 | v->swath_width_luma_ub_this_state[k], |
4977 | v->swath_width_chroma_ub_this_state[k], |
4978 | v->SwathHeightYThisState[k], |
4979 | v->SwathHeightCThisState[k], |
4980 | v->TWait, |
4981 | v->ProgressiveToInterlaceUnitInOPP, |
4982 | &v->DSTXAfterScaler[k], |
4983 | &v->DSTYAfterScaler[k], |
4984 | &v->LineTimesForPrefetch[k], |
4985 | &v->PrefetchBW[k], |
4986 | &v->LinesForMetaPTE[k], |
4987 | &v->LinesForMetaAndDPTERow[k], |
4988 | &v->VRatioPreY[i][j][k], |
4989 | &v->VRatioPreC[i][j][k], |
4990 | &v->RequiredPrefetchPixelDataBWLuma[i][j][k], |
4991 | &v->RequiredPrefetchPixelDataBWChroma[i][j][k], |
4992 | &v->NoTimeForDynamicMetadata[i][j][k], |
4993 | &v->Tno_bw[k], |
4994 | &v->prefetch_vmrow_bw[k], |
4995 | &v->Tdmdl_vm[k], |
4996 | &v->Tdmdl[k], |
4997 | &v->VUpdateOffsetPix[k], |
4998 | &v->VUpdateWidthPix[k], |
4999 | &v->VReadyOffsetPix[k]); |
5000 | } |
5001 | |
5002 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
5003 | CalculateUrgentBurstFactor( |
5004 | v->swath_width_luma_ub_this_state[k], |
5005 | v->swath_width_chroma_ub_this_state[k], |
5006 | v->DETBufferSizeInKByte, |
5007 | v->SwathHeightYThisState[k], |
5008 | v->SwathHeightCThisState[k], |
5009 | v->HTotal[k] / v->PixelClock[k], |
5010 | v->UrgentLatency, |
5011 | v->CursorBufferSize, |
5012 | v->CursorWidth[k][0], |
5013 | v->CursorBPP[k][0], |
5014 | v->VRatioPreY[i][j][k], |
5015 | v->VRatioPreC[i][j][k], |
5016 | v->BytePerPixelInDETY[k], |
5017 | v->BytePerPixelInDETC[k], |
5018 | v->DETBufferSizeYThisState[k], |
5019 | v->DETBufferSizeCThisState[k], |
5020 | &v->UrgentBurstFactorCursorPre[k], |
5021 | &v->UrgentBurstFactorLumaPre[k], |
5022 | &v->UrgentBurstFactorChroma[k], |
5023 | &v->NoUrgentLatencyHidingPre[k]); |
5024 | } |
5025 | |
5026 | v->MaximumReadBandwidthWithPrefetch = 0.0; |
5027 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
5028 | v->cursor_bw_pre[k] = v->NumberOfCursors[k] * v->CursorWidth[k][0] * v->CursorBPP[k][0] / 8.0 / (v->HTotal[k] / v->PixelClock[k]) |
5029 | * v->VRatioPreY[i][j][k]; |
5030 | |
5031 | v->MaximumReadBandwidthWithPrefetch = v->MaximumReadBandwidthWithPrefetch |
5032 | + dml_max4( |
5033 | v->VActivePixelBandwidth[i][j][k], |
5034 | v->VActiveCursorBandwidth[i][j][k] |
5035 | + v->NoOfDPP[i][j][k] * (v->meta_row_bandwidth[i][j][k] + v->dpte_row_bandwidth[i][j][k]), |
5036 | v->NoOfDPP[i][j][k] * v->prefetch_vmrow_bw[k], |
5037 | v->NoOfDPP[i][j][k] |
5038 | * (v->RequiredPrefetchPixelDataBWLuma[i][j][k] * v->UrgentBurstFactorLumaPre[k] |
5039 | + v->RequiredPrefetchPixelDataBWChroma[i][j][k] |
5040 | * v->UrgentBurstFactorChromaPre[k]) |
5041 | + v->cursor_bw_pre[k] * v->UrgentBurstFactorCursorPre[k]); |
5042 | } |
5043 | |
5044 | v->NotEnoughUrgentLatencyHidingPre = false0; |
5045 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
5046 | if (v->NoUrgentLatencyHidingPre[k] == true1) { |
5047 | v->NotEnoughUrgentLatencyHidingPre = true1; |
5048 | } |
5049 | } |
5050 | |
5051 | v->PrefetchSupported[i][j] = true1; |
5052 | if (v->BandwidthWithoutPrefetchSupported[i][j] == false0 || v->MaximumReadBandwidthWithPrefetch > v->ReturnBWPerState[i][j] |
5053 | || v->NotEnoughUrgentLatencyHidingPre == 1) { |
5054 | v->PrefetchSupported[i][j] = false0; |
5055 | } |
5056 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
5057 | if (v->LineTimesForPrefetch[k] < 2.0 || v->LinesForMetaPTE[k] >= 32.0 || v->LinesForMetaAndDPTERow[k] >= 16.0 |
5058 | || v->NoTimeForPrefetch[i][j][k] == true1) { |
5059 | v->PrefetchSupported[i][j] = false0; |
5060 | } |
5061 | } |
5062 | |
5063 | v->DynamicMetadataSupported[i][j] = true1; |
5064 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
5065 | if (v->NoTimeForDynamicMetadata[i][j][k] == true1) { |
5066 | v->DynamicMetadataSupported[i][j] = false0; |
5067 | } |
5068 | } |
5069 | |
5070 | v->VRatioInPrefetchSupported[i][j] = true1; |
5071 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
5072 | if (v->VRatioPreY[i][j][k] > 4.0 || v->VRatioPreC[i][j][k] > 4.0 || v->NoTimeForPrefetch[i][j][k] == true1) { |
5073 | v->VRatioInPrefetchSupported[i][j] = false0; |
5074 | } |
5075 | } |
5076 | v->AnyLinesForVMOrRowTooLarge = false0; |
5077 | for (k = 0; k < v->NumberOfActivePlanes; ++k) { |
5078 | if (v->LinesForMetaAndDPTERow[k] >= 16 || v->LinesForMetaPTE[k] >= 32) { |
5079 | v->AnyLinesForVMOrRowTooLarge = true1; |
5080 | } |
5081 | } |
5082 | |
5083 | if (v->PrefetchSupported[i][j] == true1 && v->VRatioInPrefetchSupported[i][j] == true1) { |
5084 | v->BandwidthAvailableForImmediateFlip = v->ReturnBWPerState[i][j]; |
5085 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
5086 | v->BandwidthAvailableForImmediateFlip = v->BandwidthAvailableForImmediateFlip |
5087 | - dml_max( |
5088 | v->VActivePixelBandwidth[i][j][k] + v->VActiveCursorBandwidth[i][j][k], |
5089 | v->NoOfDPP[i][j][k] |
5090 | * (v->RequiredPrefetchPixelDataBWLuma[i][j][k] * v->UrgentBurstFactorLumaPre[k] |
5091 | + v->RequiredPrefetchPixelDataBWChroma[i][j][k] |
5092 | * v->UrgentBurstFactorChromaPre[k]) |
5093 | + v->cursor_bw_pre[k] * v->UrgentBurstFactorCursorPre[k]); |
5094 | } |
5095 | v->TotImmediateFlipBytes = 0.0; |
5096 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
5097 | v->TotImmediateFlipBytes = v->TotImmediateFlipBytes + v->NoOfDPP[i][j][k] * v->PDEAndMetaPTEBytesPerFrame[i][j][k] |
5098 | + v->MetaRowBytes[i][j][k] + v->DPTEBytesPerRow[i][j][k]; |
5099 | } |
5100 | |
5101 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
5102 | CalculateFlipSchedule( |
5103 | mode_lib, |
5104 | v->PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelMixedWithVMData, |
5105 | v->PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyVMDataOnly, |
5106 | v->ExtraLatency, |
5107 | v->UrgLatency[i], |
5108 | v->GPUVMMaxPageTableLevels, |
5109 | v->HostVMEnable, |
5110 | v->HostVMMaxNonCachedPageTableLevels, |
5111 | v->GPUVMEnable, |
5112 | v->HostVMMinPageSize, |
5113 | v->PDEAndMetaPTEBytesPerFrame[i][j][k], |
5114 | v->MetaRowBytes[i][j][k], |
5115 | v->DPTEBytesPerRow[i][j][k], |
5116 | v->BandwidthAvailableForImmediateFlip, |
5117 | v->TotImmediateFlipBytes, |
5118 | v->SourcePixelFormat[k], |
5119 | v->HTotal[k] / v->PixelClock[k], |
5120 | v->VRatio[k], |
5121 | v->VRatioChroma[k], |
5122 | v->Tno_bw[k], |
5123 | v->DCCEnable[k], |
5124 | v->dpte_row_height[k], |
5125 | v->meta_row_height[k], |
5126 | v->dpte_row_height_chroma[k], |
5127 | v->meta_row_height_chroma[k], |
5128 | &v->DestinationLinesToRequestVMInImmediateFlip[k], |
5129 | &v->DestinationLinesToRequestRowInImmediateFlip[k], |
5130 | &v->final_flip_bw[k], |
5131 | &v->ImmediateFlipSupportedForPipe[k]); |
5132 | } |
5133 | v->total_dcn_read_bw_with_flip = 0.0; |
5134 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
5135 | v->total_dcn_read_bw_with_flip = v->total_dcn_read_bw_with_flip |
5136 | + dml_max3( |
5137 | v->NoOfDPP[i][j][k] * v->prefetch_vmrow_bw[k], |
5138 | v->NoOfDPP[i][j][k] * v->final_flip_bw[k] + v->VActivePixelBandwidth[i][j][k] |
5139 | + v->VActiveCursorBandwidth[i][j][k], |
5140 | v->NoOfDPP[i][j][k] |
5141 | * (v->final_flip_bw[k] |
5142 | + v->RequiredPrefetchPixelDataBWLuma[i][j][k] |
5143 | * v->UrgentBurstFactorLumaPre[k] |
5144 | + v->RequiredPrefetchPixelDataBWChroma[i][j][k] |
5145 | * v->UrgentBurstFactorChromaPre[k]) |
5146 | + v->cursor_bw_pre[k] * v->UrgentBurstFactorCursorPre[k]); |
5147 | } |
5148 | v->ImmediateFlipSupportedForState[i][j] = true1; |
5149 | if (v->total_dcn_read_bw_with_flip > v->ReturnBWPerState[i][j]) { |
5150 | v->ImmediateFlipSupportedForState[i][j] = false0; |
5151 | } |
5152 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
5153 | if (v->ImmediateFlipSupportedForPipe[k] == false0) { |
5154 | v->ImmediateFlipSupportedForState[i][j] = false0; |
5155 | } |
5156 | } |
5157 | } else { |
5158 | v->ImmediateFlipSupportedForState[i][j] = false0; |
5159 | } |
5160 | if (v->MaxVStartup <= 13 || v->AnyLinesForVMOrRowTooLarge == false0) { |
5161 | v->NextMaxVStartup = v->MaxMaxVStartup[i][j]; |
5162 | NextPrefetchModeState = NextPrefetchModeState + 1; |
5163 | } else { |
5164 | v->NextMaxVStartup = v->NextMaxVStartup - 1; |
5165 | } |
5166 | } while (!((v->PrefetchSupported[i][j] == true1 && v->DynamicMetadataSupported[i][j] == true1 && v->VRatioInPrefetchSupported[i][j] == true1 |
5167 | && ((v->HostVMEnable == false0 && v->ImmediateFlipRequirement != dm_immediate_flip_required) |
5168 | || v->ImmediateFlipSupportedForState[i][j] == true1)) |
5169 | || (v->NextMaxVStartup == v->MaxMaxVStartup[i][j] && NextPrefetchModeState > MaxPrefetchMode))); |
5170 | |
5171 | CalculateWatermarksAndDRAMSpeedChangeSupport( |
5172 | mode_lib, |
5173 | v->PrefetchModePerState[i][j], |
5174 | v->NumberOfActivePlanes, |
5175 | v->MaxLineBufferLines, |
5176 | v->LineBufferSize, |
5177 | v->DPPOutputBufferPixels, |
5178 | v->DETBufferSizeInKByte, |
5179 | v->WritebackInterfaceBufferSize, |
5180 | v->DCFCLKState[i][j], |
5181 | v->ReturnBWPerState[i][j], |
5182 | v->GPUVMEnable, |
5183 | v->dpte_group_bytes, |
5184 | v->MetaChunkSize, |
5185 | v->UrgLatency[i], |
5186 | v->ExtraLatency, |
5187 | v->WritebackLatency, |
5188 | v->WritebackChunkSize, |
5189 | v->SOCCLKPerState[i], |
5190 | v->FinalDRAMClockChangeLatency, |
5191 | v->SRExitTime, |
5192 | v->SREnterPlusExitTime, |
5193 | v->ProjectedDCFCLKDeepSleep[i][j], |
5194 | v->NoOfDPPThisState, |
5195 | v->DCCEnable, |
5196 | v->RequiredDPPCLKThisState, |
5197 | v->DETBufferSizeYThisState, |
5198 | v->DETBufferSizeCThisState, |
5199 | v->SwathHeightYThisState, |
5200 | v->SwathHeightCThisState, |
5201 | v->LBBitPerPixel, |
5202 | v->SwathWidthYThisState, |
5203 | v->SwathWidthCThisState, |
5204 | v->HRatio, |
5205 | v->HRatioChroma, |
5206 | v->vtaps, |
5207 | v->VTAPsChroma, |
5208 | v->VRatio, |
5209 | v->VRatioChroma, |
5210 | v->HTotal, |
5211 | v->PixelClock, |
5212 | v->BlendingAndTiming, |
5213 | v->BytePerPixelInDETY, |
5214 | v->BytePerPixelInDETC, |
5215 | v->DSTXAfterScaler, |
5216 | v->DSTYAfterScaler, |
5217 | v->WritebackEnable, |
5218 | v->WritebackPixelFormat, |
5219 | v->WritebackDestinationWidth, |
5220 | v->WritebackDestinationHeight, |
5221 | v->WritebackSourceHeight, |
5222 | &v->DRAMClockChangeSupport[i][j], |
5223 | &v->UrgentWatermark, |
5224 | &v->WritebackUrgentWatermark, |
5225 | &v->DRAMClockChangeWatermark, |
5226 | &v->WritebackDRAMClockChangeWatermark, |
5227 | &v->StutterExitWatermark, |
5228 | &v->StutterEnterPlusExitWatermark, |
5229 | &v->MinActiveDRAMClockChangeLatencySupported); |
5230 | } |
5231 | } |
5232 | |
5233 | /*PTE Buffer Size Check*/ |
5234 | |
5235 | for (i = 0; i < v->soc.num_states; i++) { |
5236 | for (j = 0; j < 2; j++) { |
5237 | v->PTEBufferSizeNotExceeded[i][j] = true1; |
5238 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
5239 | if (v->PTEBufferSizeNotExceededY[i][j][k] == false0 || v->PTEBufferSizeNotExceededC[i][j][k] == false0) { |
5240 | v->PTEBufferSizeNotExceeded[i][j] = false0; |
5241 | } |
5242 | } |
5243 | } |
5244 | } |
5245 | /*Cursor Support Check*/ |
5246 | |
5247 | v->CursorSupport = true1; |
5248 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
5249 | if (v->CursorWidth[k][0] > 0.0) { |
5250 | if (v->CursorBPP[k][0] == 64 && v->Cursor64BppSupport == false0) { |
5251 | v->CursorSupport = false0; |
5252 | } |
5253 | } |
5254 | } |
5255 | /*Valid Pitch Check*/ |
5256 | |
5257 | v->PitchSupport = true1; |
5258 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
5259 | v->AlignedYPitch[k] = dml_ceil(dml_max(v->PitchY[k], v->SurfaceWidthY[k]), v->MacroTileWidthY[k]); |
5260 | if (v->DCCEnable[k] == true1) { |
5261 | v->AlignedDCCMetaPitchY[k] = dml_ceil(dml_max(v->DCCMetaPitchY[k], v->SurfaceWidthY[k]), 64.0 * v->Read256BlockWidthY[k]); |
5262 | } else { |
5263 | v->AlignedDCCMetaPitchY[k] = v->DCCMetaPitchY[k]; |
5264 | } |
5265 | if (v->SourcePixelFormat[k] != dm_444_64 && v->SourcePixelFormat[k] != dm_444_32 && v->SourcePixelFormat[k] != dm_444_16 && v->SourcePixelFormat[k] != dm_mono_16 |
5266 | && v->SourcePixelFormat[k] != dm_rgbe && v->SourcePixelFormat[k] != dm_mono_8) { |
5267 | v->AlignedCPitch[k] = dml_ceil(dml_max(v->PitchC[k], v->SurfaceWidthC[k]), v->MacroTileWidthC[k]); |
5268 | if (v->DCCEnable[k] == true1) { |
5269 | v->AlignedDCCMetaPitchC[k] = dml_ceil(dml_max(v->DCCMetaPitchC[k], v->SurfaceWidthC[k]), 64.0 * v->Read256BlockWidthC[k]); |
5270 | } else { |
5271 | v->AlignedDCCMetaPitchC[k] = v->DCCMetaPitchC[k]; |
5272 | } |
5273 | } else { |
5274 | v->AlignedCPitch[k] = v->PitchC[k]; |
5275 | v->AlignedDCCMetaPitchC[k] = v->DCCMetaPitchC[k]; |
5276 | } |
5277 | if (v->AlignedYPitch[k] > v->PitchY[k] || v->AlignedCPitch[k] > v->PitchC[k] || v->AlignedDCCMetaPitchY[k] > v->DCCMetaPitchY[k] |
5278 | || v->AlignedDCCMetaPitchC[k] > v->DCCMetaPitchC[k]) { |
5279 | v->PitchSupport = false0; |
5280 | } |
5281 | } |
5282 | |
5283 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
5284 | if (v->ViewportWidth[k] > v->SurfaceWidthY[k] || v->ViewportHeight[k] > v->SurfaceHeightY[k]) |
5285 | ViewportExceedsSurface = true1; |
5286 | |
5287 | if (v->SourcePixelFormat[k] != dm_444_64 && v->SourcePixelFormat[k] != dm_444_32 && v->SourcePixelFormat[k] != dm_444_16 |
5288 | && v->SourcePixelFormat[k] != dm_444_8 && v->SourcePixelFormat[k] != dm_rgbe) { |
5289 | if (v->ViewportWidthChroma[k] > v->SurfaceWidthC[k] || v->ViewportHeightChroma[k] > v->SurfaceHeightC[k]) { |
5290 | ViewportExceedsSurface = true1; |
5291 | } |
5292 | } |
5293 | } |
5294 | /*Mode Support, Voltage State and SOC Configuration*/ |
5295 | |
5296 | for (i = v->soc.num_states - 1; i >= 0; i--) { |
5297 | for (j = 0; j < 2; j++) { |
5298 | if (v->ScaleRatioAndTapsSupport == 1 && v->SourceFormatPixelAndScanSupport == 1 && v->ViewportSizeSupport[i][j] == 1 |
5299 | && v->DIOSupport[i] == 1 && v->ODMCombine4To1SupportCheckOK[i] == 1 |
5300 | && v->NotEnoughDSCUnits[i] == 0 |
5301 | && v->DTBCLKRequiredMoreThanSupported[i] == 0 |
5302 | && v->ROBSupport[i][j] == 1 && v->DISPCLK_DPPCLK_Support[i][j] == 1 && v->TotalAvailablePipesSupport[i][j] == 1 |
5303 | && EnoughWritebackUnits == 1 && WritebackModeSupport == 1 |
5304 | && v->WritebackLatencySupport == 1 && v->WritebackScaleRatioAndTapsSupport == 1 && v->CursorSupport == 1 && v->PitchSupport == 1 |
5305 | && ViewportExceedsSurface == 0 && v->PrefetchSupported[i][j] == 1 && v->DynamicMetadataSupported[i][j] == 1 |
5306 | && v->TotalVerticalActiveBandwidthSupport[i][j] == 1 && v->VRatioInPrefetchSupported[i][j] == 1 |
5307 | && v->PTEBufferSizeNotExceeded[i][j] == 1 && v->NonsupportedDSCInputBPC == 0 |
5308 | && ((v->HostVMEnable == 0 && v->ImmediateFlipRequirement != dm_immediate_flip_required) |
5309 | || v->ImmediateFlipSupportedForState[i][j] == true1)) { |
5310 | v->ModeSupport[i][j] = true1; |
5311 | } else { |
5312 | v->ModeSupport[i][j] = false0; |
5313 | } |
5314 | } |
5315 | } |
5316 | { |
5317 | unsigned int MaximumMPCCombine = 0; |
5318 | for (i = v->soc.num_states; i >= 0; i--) { |
5319 | if (i == v->soc.num_states || v->ModeSupport[i][0] == true1 || v->ModeSupport[i][1] == true1) { |
5320 | v->VoltageLevel = i; |
5321 | v->ModeIsSupported = v->ModeSupport[i][0] == true1 || v->ModeSupport[i][1] == true1; |
5322 | if (v->ModeSupport[i][1] == true1) { |
5323 | MaximumMPCCombine = 1; |
5324 | } else { |
5325 | MaximumMPCCombine = 0; |
5326 | } |
5327 | } |
5328 | } |
5329 | v->ImmediateFlipSupport = v->ImmediateFlipSupportedForState[v->VoltageLevel][MaximumMPCCombine]; |
5330 | for (k = 0; k <= v->NumberOfActivePlanes - 1; k++) { |
5331 | v->MPCCombineEnable[k] = v->MPCCombine[v->VoltageLevel][MaximumMPCCombine][k]; |
5332 | v->DPPPerPlane[k] = v->NoOfDPP[v->VoltageLevel][MaximumMPCCombine][k]; |
5333 | } |
5334 | v->DCFCLK = v->DCFCLKState[v->VoltageLevel][MaximumMPCCombine]; |
5335 | v->DRAMSpeed = v->DRAMSpeedPerState[v->VoltageLevel]; |
5336 | v->FabricClock = v->FabricClockPerState[v->VoltageLevel]; |
5337 | v->SOCCLK = v->SOCCLKPerState[v->VoltageLevel]; |
5338 | v->ReturnBW = v->ReturnBWPerState[v->VoltageLevel][MaximumMPCCombine]; |
5339 | v->maxMpcComb = MaximumMPCCombine; |
5340 | } |
5341 | } |
5342 | |
5343 | static void CalculateWatermarksAndDRAMSpeedChangeSupport( |
5344 | struct display_mode_lib *mode_lib, |
5345 | unsigned int PrefetchMode, |
5346 | unsigned int NumberOfActivePlanes, |
5347 | unsigned int MaxLineBufferLines, |
5348 | unsigned int LineBufferSize, |
5349 | unsigned int DPPOutputBufferPixels, |
5350 | double DETBufferSizeInKByte, |
5351 | unsigned int WritebackInterfaceBufferSize, |
5352 | double DCFCLK, |
5353 | double ReturnBW, |
5354 | bool_Bool GPUVMEnable, |
5355 | unsigned int dpte_group_bytes[], |
5356 | unsigned int MetaChunkSize, |
5357 | double UrgentLatency, |
5358 | double ExtraLatency, |
5359 | double WritebackLatency, |
5360 | double WritebackChunkSize, |
5361 | double SOCCLK, |
5362 | double DRAMClockChangeLatency, |
5363 | double SRExitTime, |
5364 | double SREnterPlusExitTime, |
5365 | double DCFCLKDeepSleep, |
5366 | unsigned int DPPPerPlane[], |
5367 | bool_Bool DCCEnable[], |
5368 | double DPPCLK[], |
5369 | double DETBufferSizeY[], |
5370 | double DETBufferSizeC[], |
5371 | unsigned int SwathHeightY[], |
5372 | unsigned int SwathHeightC[], |
5373 | unsigned int LBBitPerPixel[], |
5374 | double SwathWidthY[], |
5375 | double SwathWidthC[], |
5376 | double HRatio[], |
5377 | double HRatioChroma[], |
5378 | unsigned int vtaps[], |
5379 | unsigned int VTAPsChroma[], |
5380 | double VRatio[], |
5381 | double VRatioChroma[], |
5382 | unsigned int HTotal[], |
5383 | double PixelClock[], |
5384 | unsigned int BlendingAndTiming[], |
5385 | double BytePerPixelDETY[], |
5386 | double BytePerPixelDETC[], |
5387 | double DSTXAfterScaler[], |
5388 | double DSTYAfterScaler[], |
5389 | bool_Bool WritebackEnable[], |
5390 | enum source_format_class WritebackPixelFormat[], |
5391 | double WritebackDestinationWidth[], |
5392 | double WritebackDestinationHeight[], |
5393 | double WritebackSourceHeight[], |
5394 | enum clock_change_support *DRAMClockChangeSupport, |
5395 | double *UrgentWatermark, |
5396 | double *WritebackUrgentWatermark, |
5397 | double *DRAMClockChangeWatermark, |
5398 | double *WritebackDRAMClockChangeWatermark, |
5399 | double *StutterExitWatermark, |
5400 | double *StutterEnterPlusExitWatermark, |
5401 | double *MinActiveDRAMClockChangeLatencySupported) |
5402 | { |
5403 | double EffectiveLBLatencyHidingY = 0; |
5404 | double EffectiveLBLatencyHidingC = 0; |
5405 | double LinesInDETY[DC__NUM_DPP__MAX8] = { 0 }; |
5406 | double LinesInDETC = 0; |
5407 | unsigned int LinesInDETYRoundedDownToSwath[DC__NUM_DPP__MAX8] = { 0 }; |
5408 | unsigned int LinesInDETCRoundedDownToSwath = 0; |
5409 | double FullDETBufferingTimeY[DC__NUM_DPP__MAX8] = { 0 }; |
5410 | double FullDETBufferingTimeC = 0; |
5411 | double ActiveDRAMClockChangeLatencyMarginY = 0; |
5412 | double ActiveDRAMClockChangeLatencyMarginC = 0; |
5413 | double WritebackDRAMClockChangeLatencyMargin = 0; |
5414 | double PlaneWithMinActiveDRAMClockChangeMargin = 0; |
5415 | double SecondMinActiveDRAMClockChangeMarginOneDisplayInVBLank = 0; |
5416 | double FullDETBufferingTimeYStutterCriticalPlane = 0; |
5417 | double TimeToFinishSwathTransferStutterCriticalPlane = 0; |
5418 | double WritebackDRAMClockChangeLatencyHiding = 0; |
5419 | unsigned int k, j; |
5420 | |
5421 | mode_lib->vba.TotalActiveDPP = 0; |
5422 | mode_lib->vba.TotalDCCActiveDPP = 0; |
5423 | for (k = 0; k < NumberOfActivePlanes; ++k) { |
5424 | mode_lib->vba.TotalActiveDPP = mode_lib->vba.TotalActiveDPP + DPPPerPlane[k]; |
5425 | if (DCCEnable[k] == true1) { |
5426 | mode_lib->vba.TotalDCCActiveDPP = mode_lib->vba.TotalDCCActiveDPP + DPPPerPlane[k]; |
5427 | } |
5428 | } |
5429 | |
5430 | *UrgentWatermark = UrgentLatency + ExtraLatency; |
5431 | |
5432 | *DRAMClockChangeWatermark = DRAMClockChangeLatency + *UrgentWatermark; |
5433 | |
5434 | mode_lib->vba.TotalActiveWriteback = 0; |
5435 | for (k = 0; k < NumberOfActivePlanes; ++k) { |
5436 | if (WritebackEnable[k] == true1) { |
5437 | mode_lib->vba.TotalActiveWriteback = mode_lib->vba.TotalActiveWriteback + 1; |
5438 | } |
5439 | } |
5440 | |
5441 | if (mode_lib->vba.TotalActiveWriteback <= 1) { |
5442 | *WritebackUrgentWatermark = WritebackLatency; |
5443 | } else { |
5444 | *WritebackUrgentWatermark = WritebackLatency + WritebackChunkSize * 1024.0 / 32.0 / SOCCLK; |
5445 | } |
5446 | |
5447 | if (mode_lib->vba.TotalActiveWriteback <= 1) { |
5448 | *WritebackDRAMClockChangeWatermark = DRAMClockChangeLatency + WritebackLatency; |
5449 | } else { |
5450 | *WritebackDRAMClockChangeWatermark = DRAMClockChangeLatency + WritebackLatency + WritebackChunkSize * 1024.0 / 32.0 / SOCCLK; |
5451 | } |
5452 | |
5453 | for (k = 0; k < NumberOfActivePlanes; ++k) { |
5454 | |
5455 | mode_lib->vba.LBLatencyHidingSourceLinesY = dml_min((double) MaxLineBufferLines, dml_floor(LineBufferSize / LBBitPerPixel[k] / (SwathWidthY[k] / dml_max(HRatio[k], 1.0)), 1)) - (vtaps[k] - 1); |
5456 | |
5457 | mode_lib->vba.LBLatencyHidingSourceLinesC = dml_min((double) MaxLineBufferLines, dml_floor(LineBufferSize / LBBitPerPixel[k] / (SwathWidthC[k] / dml_max(HRatioChroma[k], 1.0)), 1)) - (VTAPsChroma[k] - 1); |
5458 | |
5459 | EffectiveLBLatencyHidingY = mode_lib->vba.LBLatencyHidingSourceLinesY / VRatio[k] * (HTotal[k] / PixelClock[k]); |
5460 | |
5461 | EffectiveLBLatencyHidingC = mode_lib->vba.LBLatencyHidingSourceLinesC / VRatioChroma[k] * (HTotal[k] / PixelClock[k]); |
5462 | |
5463 | LinesInDETY[k] = (double) DETBufferSizeY[k] / BytePerPixelDETY[k] / SwathWidthY[k]; |
5464 | LinesInDETYRoundedDownToSwath[k] = dml_floor(LinesInDETY[k], SwathHeightY[k]); |
5465 | FullDETBufferingTimeY[k] = LinesInDETYRoundedDownToSwath[k] * (HTotal[k] / PixelClock[k]) / VRatio[k]; |
5466 | if (BytePerPixelDETC[k] > 0) { |
5467 | LinesInDETC = mode_lib->vba.DETBufferSizeC[k] / BytePerPixelDETC[k] / SwathWidthC[k]; |
5468 | LinesInDETCRoundedDownToSwath = dml_floor(LinesInDETC, SwathHeightC[k]); |
5469 | FullDETBufferingTimeC = LinesInDETCRoundedDownToSwath * (HTotal[k] / PixelClock[k]) / VRatioChroma[k]; |
5470 | } else { |
5471 | LinesInDETC = 0; |
5472 | FullDETBufferingTimeC = 999999; |
5473 | } |
5474 | |
5475 | ActiveDRAMClockChangeLatencyMarginY = EffectiveLBLatencyHidingY + FullDETBufferingTimeY[k] - *UrgentWatermark - (HTotal[k] / PixelClock[k]) * (DSTXAfterScaler[k] / HTotal[k] + DSTYAfterScaler[k]) - *DRAMClockChangeWatermark; |
5476 | |
5477 | if (NumberOfActivePlanes > 1) { |
5478 | ActiveDRAMClockChangeLatencyMarginY = ActiveDRAMClockChangeLatencyMarginY - (1 - 1.0 / NumberOfActivePlanes) * SwathHeightY[k] * HTotal[k] / PixelClock[k] / VRatio[k]; |
5479 | } |
5480 | |
5481 | if (BytePerPixelDETC[k] > 0) { |
5482 | ActiveDRAMClockChangeLatencyMarginC = EffectiveLBLatencyHidingC + FullDETBufferingTimeC - *UrgentWatermark - (HTotal[k] / PixelClock[k]) * (DSTXAfterScaler[k] / HTotal[k] + DSTYAfterScaler[k]) - *DRAMClockChangeWatermark; |
5483 | |
5484 | if (NumberOfActivePlanes > 1) { |
5485 | ActiveDRAMClockChangeLatencyMarginC = ActiveDRAMClockChangeLatencyMarginC - (1 - 1.0 / NumberOfActivePlanes) * SwathHeightC[k] * HTotal[k] / PixelClock[k] / VRatioChroma[k]; |
5486 | } |
5487 | mode_lib->vba.ActiveDRAMClockChangeLatencyMargin[k] = dml_min(ActiveDRAMClockChangeLatencyMarginY, ActiveDRAMClockChangeLatencyMarginC); |
5488 | } else { |
5489 | mode_lib->vba.ActiveDRAMClockChangeLatencyMargin[k] = ActiveDRAMClockChangeLatencyMarginY; |
5490 | } |
5491 | |
5492 | if (WritebackEnable[k] == true1) { |
5493 | |
5494 | WritebackDRAMClockChangeLatencyHiding = WritebackInterfaceBufferSize * 1024 / (WritebackDestinationWidth[k] * WritebackDestinationHeight[k] / (WritebackSourceHeight[k] * HTotal[k] / PixelClock[k]) * 4); |
5495 | if (WritebackPixelFormat[k] == dm_444_64) { |
5496 | WritebackDRAMClockChangeLatencyHiding = WritebackDRAMClockChangeLatencyHiding / 2; |
5497 | } |
5498 | if (mode_lib->vba.WritebackConfiguration == dm_whole_buffer_for_single_stream_interleave) { |
5499 | WritebackDRAMClockChangeLatencyHiding = WritebackDRAMClockChangeLatencyHiding * 2; |
5500 | } |
5501 | WritebackDRAMClockChangeLatencyMargin = WritebackDRAMClockChangeLatencyHiding - mode_lib->vba.WritebackDRAMClockChangeWatermark; |
5502 | mode_lib->vba.ActiveDRAMClockChangeLatencyMargin[k] = dml_min(mode_lib->vba.ActiveDRAMClockChangeLatencyMargin[k], WritebackDRAMClockChangeLatencyMargin); |
5503 | } |
5504 | } |
5505 | |
5506 | mode_lib->vba.MinActiveDRAMClockChangeMargin = 999999; |
5507 | PlaneWithMinActiveDRAMClockChangeMargin = 0; |
5508 | for (k = 0; k < NumberOfActivePlanes; ++k) { |
5509 | if (mode_lib->vba.ActiveDRAMClockChangeLatencyMargin[k] < mode_lib->vba.MinActiveDRAMClockChangeMargin) { |
5510 | mode_lib->vba.MinActiveDRAMClockChangeMargin = mode_lib->vba.ActiveDRAMClockChangeLatencyMargin[k]; |
5511 | if (BlendingAndTiming[k] == k) { |
5512 | PlaneWithMinActiveDRAMClockChangeMargin = k; |
5513 | } else { |
5514 | for (j = 0; j < NumberOfActivePlanes; ++j) { |
5515 | if (BlendingAndTiming[k] == j) { |
5516 | PlaneWithMinActiveDRAMClockChangeMargin = j; |
5517 | } |
5518 | } |
5519 | } |
5520 | } |
5521 | } |
5522 | |
5523 | *MinActiveDRAMClockChangeLatencySupported = mode_lib->vba.MinActiveDRAMClockChangeMargin + DRAMClockChangeLatency; |
5524 | |
5525 | SecondMinActiveDRAMClockChangeMarginOneDisplayInVBLank = 999999; |
5526 | for (k = 0; k < NumberOfActivePlanes; ++k) { |
5527 | if (!((k == PlaneWithMinActiveDRAMClockChangeMargin) && (BlendingAndTiming[k] == k)) && !(BlendingAndTiming[k] == PlaneWithMinActiveDRAMClockChangeMargin) && mode_lib->vba.ActiveDRAMClockChangeLatencyMargin[k] < SecondMinActiveDRAMClockChangeMarginOneDisplayInVBLank) { |
5528 | SecondMinActiveDRAMClockChangeMarginOneDisplayInVBLank = mode_lib->vba.ActiveDRAMClockChangeLatencyMargin[k]; |
5529 | } |
5530 | } |
5531 | |
5532 | mode_lib->vba.TotalNumberOfActiveOTG = 0; |
5533 | for (k = 0; k < NumberOfActivePlanes; ++k) { |
5534 | if (BlendingAndTiming[k] == k) { |
5535 | mode_lib->vba.TotalNumberOfActiveOTG = mode_lib->vba.TotalNumberOfActiveOTG + 1; |
5536 | } |
5537 | } |
5538 | |
5539 | if (mode_lib->vba.MinActiveDRAMClockChangeMargin > 0 && PrefetchMode == 0) { |
5540 | *DRAMClockChangeSupport = dm_dram_clock_change_vactive; |
5541 | } else if (((mode_lib->vba.SynchronizedVBlank == true1 || mode_lib->vba.TotalNumberOfActiveOTG == 1 || SecondMinActiveDRAMClockChangeMarginOneDisplayInVBLank > 0) && PrefetchMode == 0)) { |
5542 | *DRAMClockChangeSupport = dm_dram_clock_change_vblank; |
5543 | } else { |
5544 | *DRAMClockChangeSupport = dm_dram_clock_change_unsupported; |
5545 | } |
5546 | |
5547 | FullDETBufferingTimeYStutterCriticalPlane = FullDETBufferingTimeY[0]; |
5548 | for (k = 0; k < NumberOfActivePlanes; ++k) { |
5549 | if (FullDETBufferingTimeY[k] <= FullDETBufferingTimeYStutterCriticalPlane) { |
5550 | FullDETBufferingTimeYStutterCriticalPlane = FullDETBufferingTimeY[k]; |
5551 | TimeToFinishSwathTransferStutterCriticalPlane = (SwathHeightY[k] - (LinesInDETY[k] - LinesInDETYRoundedDownToSwath[k])) * (HTotal[k] / PixelClock[k]) / VRatio[k]; |
5552 | } |
5553 | } |
5554 | |
5555 | *StutterExitWatermark = SRExitTime + ExtraLatency + 10 / DCFCLKDeepSleep; |
5556 | *StutterEnterPlusExitWatermark = dml_max(SREnterPlusExitTime + ExtraLatency + 10 / DCFCLKDeepSleep, TimeToFinishSwathTransferStutterCriticalPlane); |
5557 | |
5558 | } |
5559 | |
5560 | static void CalculateDCFCLKDeepSleep( |
5561 | struct display_mode_lib *mode_lib, |
5562 | unsigned int NumberOfActivePlanes, |
5563 | int BytePerPixelY[], |
5564 | int BytePerPixelC[], |
5565 | double VRatio[], |
5566 | double VRatioChroma[], |
5567 | double SwathWidthY[], |
5568 | double SwathWidthC[], |
5569 | unsigned int DPPPerPlane[], |
5570 | double HRatio[], |
5571 | double HRatioChroma[], |
5572 | double PixelClock[], |
5573 | double PSCL_THROUGHPUT[], |
5574 | double PSCL_THROUGHPUT_CHROMA[], |
5575 | double DPPCLK[], |
5576 | double ReadBandwidthLuma[], |
5577 | double ReadBandwidthChroma[], |
5578 | int ReturnBusWidth, |
5579 | double *DCFCLKDeepSleep) |
5580 | { |
5581 | double DisplayPipeLineDeliveryTimeLuma = 0; |
5582 | double DisplayPipeLineDeliveryTimeChroma = 0; |
5583 | unsigned int k; |
5584 | double ReadBandwidth = 0.0; |
5585 | |
5586 | //double DCFCLKDeepSleepPerPlane[DC__NUM_DPP__MAX]; |
5587 | for (k = 0; k < NumberOfActivePlanes; ++k) { |
5588 | |
5589 | if (VRatio[k] <= 1) { |
5590 | DisplayPipeLineDeliveryTimeLuma = SwathWidthY[k] * DPPPerPlane[k] / HRatio[k] / PixelClock[k]; |
5591 | } else { |
5592 | DisplayPipeLineDeliveryTimeLuma = SwathWidthY[k] / PSCL_THROUGHPUT[k] / DPPCLK[k]; |
5593 | } |
5594 | if (BytePerPixelC[k] == 0) { |
5595 | DisplayPipeLineDeliveryTimeChroma = 0; |
5596 | } else { |
5597 | if (VRatioChroma[k] <= 1) { |
5598 | DisplayPipeLineDeliveryTimeChroma = SwathWidthC[k] * DPPPerPlane[k] / HRatioChroma[k] / PixelClock[k]; |
5599 | } else { |
5600 | DisplayPipeLineDeliveryTimeChroma = SwathWidthC[k] / PSCL_THROUGHPUT_CHROMA[k] / DPPCLK[k]; |
5601 | } |
5602 | } |
5603 | |
5604 | if (BytePerPixelC[k] > 0) { |
5605 | mode_lib->vba.DCFCLKDeepSleepPerPlane[k] = dml_max(1.1 * SwathWidthY[k] * BytePerPixelY[k] / 32.0 / DisplayPipeLineDeliveryTimeLuma, 1.1 * SwathWidthC[k] * BytePerPixelC[k] / 32.0 / DisplayPipeLineDeliveryTimeChroma); |
5606 | } else { |
5607 | mode_lib->vba.DCFCLKDeepSleepPerPlane[k] = 1.1 * SwathWidthY[k] * BytePerPixelY[k] / 64.0 / DisplayPipeLineDeliveryTimeLuma; |
5608 | } |
5609 | mode_lib->vba.DCFCLKDeepSleepPerPlane[k] = dml_max(mode_lib->vba.DCFCLKDeepSleepPerPlane[k], PixelClock[k] / 16); |
5610 | |
5611 | } |
5612 | |
5613 | for (k = 0; k < NumberOfActivePlanes; ++k) { |
5614 | ReadBandwidth = ReadBandwidth + ReadBandwidthLuma[k] + ReadBandwidthChroma[k]; |
5615 | } |
5616 | |
5617 | *DCFCLKDeepSleep = dml_max(8.0, ReadBandwidth / ReturnBusWidth); |
5618 | |
5619 | for (k = 0; k < NumberOfActivePlanes; ++k) { |
5620 | *DCFCLKDeepSleep = dml_max(*DCFCLKDeepSleep, mode_lib->vba.DCFCLKDeepSleepPerPlane[k]); |
5621 | } |
5622 | } |
5623 | |
5624 | static void CalculateUrgentBurstFactor( |
5625 | long swath_width_luma_ub, |
5626 | long swath_width_chroma_ub, |
5627 | unsigned int DETBufferSizeInKByte, |
5628 | unsigned int SwathHeightY, |
5629 | unsigned int SwathHeightC, |
5630 | double LineTime, |
5631 | double UrgentLatency, |
5632 | double CursorBufferSize, |
5633 | unsigned int CursorWidth, |
5634 | unsigned int CursorBPP, |
5635 | double VRatio, |
5636 | double VRatioC, |
5637 | double BytePerPixelInDETY, |
5638 | double BytePerPixelInDETC, |
5639 | double DETBufferSizeY, |
5640 | double DETBufferSizeC, |
5641 | double *UrgentBurstFactorCursor, |
5642 | double *UrgentBurstFactorLuma, |
5643 | double *UrgentBurstFactorChroma, |
5644 | bool_Bool *NotEnoughUrgentLatencyHiding) |
5645 | { |
5646 | double LinesInDETLuma = 0; |
5647 | double LinesInDETChroma = 0; |
5648 | unsigned int LinesInCursorBuffer = 0; |
5649 | double CursorBufferSizeInTime = 0; |
5650 | double DETBufferSizeInTimeLuma = 0; |
5651 | double DETBufferSizeInTimeChroma = 0; |
5652 | |
5653 | *NotEnoughUrgentLatencyHiding = 0; |
5654 | |
5655 | if (CursorWidth > 0) { |
5656 | LinesInCursorBuffer = 1 << (unsigned int) dml_floor(dml_log2(CursorBufferSize * 1024.0 / (CursorWidth * CursorBPP / 8.0)), 1.0); |
5657 | if (VRatio > 0) { |
5658 | CursorBufferSizeInTime = LinesInCursorBuffer * LineTime / VRatio; |
5659 | if (CursorBufferSizeInTime - UrgentLatency <= 0) { |
5660 | *NotEnoughUrgentLatencyHiding = 1; |
5661 | *UrgentBurstFactorCursor = 0; |
5662 | } else { |
5663 | *UrgentBurstFactorCursor = CursorBufferSizeInTime / (CursorBufferSizeInTime - UrgentLatency); |
5664 | } |
5665 | } else { |
5666 | *UrgentBurstFactorCursor = 1; |
5667 | } |
5668 | } |
5669 | |
5670 | LinesInDETLuma = DETBufferSizeY / BytePerPixelInDETY / swath_width_luma_ub; |
5671 | if (VRatio > 0) { |
5672 | DETBufferSizeInTimeLuma = dml_floor(LinesInDETLuma, SwathHeightY) * LineTime / VRatio; |
5673 | if (DETBufferSizeInTimeLuma - UrgentLatency <= 0) { |
5674 | *NotEnoughUrgentLatencyHiding = 1; |
5675 | *UrgentBurstFactorLuma = 0; |
5676 | } else { |
5677 | *UrgentBurstFactorLuma = DETBufferSizeInTimeLuma / (DETBufferSizeInTimeLuma - UrgentLatency); |
5678 | } |
5679 | } else { |
5680 | *UrgentBurstFactorLuma = 1; |
5681 | } |
5682 | |
5683 | if (BytePerPixelInDETC > 0) { |
5684 | LinesInDETChroma = DETBufferSizeC / BytePerPixelInDETC / swath_width_chroma_ub; |
5685 | if (VRatio > 0) { |
5686 | DETBufferSizeInTimeChroma = dml_floor(LinesInDETChroma, SwathHeightC) * LineTime / VRatio; |
5687 | if (DETBufferSizeInTimeChroma - UrgentLatency <= 0) { |
5688 | *NotEnoughUrgentLatencyHiding = 1; |
5689 | *UrgentBurstFactorChroma = 0; |
5690 | } else { |
5691 | *UrgentBurstFactorChroma = DETBufferSizeInTimeChroma / (DETBufferSizeInTimeChroma - UrgentLatency); |
5692 | } |
5693 | } else { |
5694 | *UrgentBurstFactorChroma = 1; |
5695 | } |
5696 | } |
5697 | } |
5698 | |
5699 | static void CalculatePixelDeliveryTimes( |
5700 | unsigned int NumberOfActivePlanes, |
5701 | double VRatio[], |
5702 | double VRatioChroma[], |
5703 | double VRatioPrefetchY[], |
5704 | double VRatioPrefetchC[], |
5705 | unsigned int swath_width_luma_ub[], |
5706 | unsigned int swath_width_chroma_ub[], |
5707 | unsigned int DPPPerPlane[], |
5708 | double HRatio[], |
5709 | double HRatioChroma[], |
5710 | double PixelClock[], |
5711 | double PSCL_THROUGHPUT[], |
5712 | double PSCL_THROUGHPUT_CHROMA[], |
5713 | double DPPCLK[], |
5714 | int BytePerPixelC[], |
5715 | enum scan_direction_class SourceScan[], |
5716 | unsigned int NumberOfCursors[], |
5717 | unsigned int CursorWidth[][2], |
5718 | unsigned int CursorBPP[][2], |
5719 | unsigned int BlockWidth256BytesY[], |
5720 | unsigned int BlockHeight256BytesY[], |
5721 | unsigned int BlockWidth256BytesC[], |
5722 | unsigned int BlockHeight256BytesC[], |
5723 | double DisplayPipeLineDeliveryTimeLuma[], |
5724 | double DisplayPipeLineDeliveryTimeChroma[], |
5725 | double DisplayPipeLineDeliveryTimeLumaPrefetch[], |
5726 | double DisplayPipeLineDeliveryTimeChromaPrefetch[], |
5727 | double DisplayPipeRequestDeliveryTimeLuma[], |
5728 | double DisplayPipeRequestDeliveryTimeChroma[], |
5729 | double DisplayPipeRequestDeliveryTimeLumaPrefetch[], |
5730 | double DisplayPipeRequestDeliveryTimeChromaPrefetch[], |
5731 | double CursorRequestDeliveryTime[], |
5732 | double CursorRequestDeliveryTimePrefetch[]) |
5733 | { |
5734 | double req_per_swath_ub = 0; |
5735 | unsigned int k; |
5736 | |
5737 | for (k = 0; k < NumberOfActivePlanes; ++k) { |
5738 | if (VRatio[k] <= 1) { |
5739 | DisplayPipeLineDeliveryTimeLuma[k] = swath_width_luma_ub[k] * DPPPerPlane[k] / HRatio[k] / PixelClock[k]; |
5740 | } else { |
5741 | DisplayPipeLineDeliveryTimeLuma[k] = swath_width_luma_ub[k] / PSCL_THROUGHPUT[k] / DPPCLK[k]; |
5742 | } |
5743 | |
5744 | if (BytePerPixelC[k] == 0) { |
5745 | DisplayPipeLineDeliveryTimeChroma[k] = 0; |
5746 | } else { |
5747 | if (VRatioChroma[k] <= 1) { |
5748 | DisplayPipeLineDeliveryTimeChroma[k] = swath_width_chroma_ub[k] * DPPPerPlane[k] / HRatioChroma[k] / PixelClock[k]; |
5749 | } else { |
5750 | DisplayPipeLineDeliveryTimeChroma[k] = swath_width_chroma_ub[k] / PSCL_THROUGHPUT_CHROMA[k] / DPPCLK[k]; |
5751 | } |
5752 | } |
5753 | |
5754 | if (VRatioPrefetchY[k] <= 1) { |
5755 | DisplayPipeLineDeliveryTimeLumaPrefetch[k] = swath_width_luma_ub[k] * DPPPerPlane[k] / HRatio[k] / PixelClock[k]; |
5756 | } else { |
5757 | DisplayPipeLineDeliveryTimeLumaPrefetch[k] = swath_width_luma_ub[k] / PSCL_THROUGHPUT[k] / DPPCLK[k]; |
5758 | } |
5759 | |
5760 | if (BytePerPixelC[k] == 0) { |
5761 | DisplayPipeLineDeliveryTimeChromaPrefetch[k] = 0; |
5762 | } else { |
5763 | if (VRatioPrefetchC[k] <= 1) { |
5764 | DisplayPipeLineDeliveryTimeChromaPrefetch[k] = swath_width_chroma_ub[k] * DPPPerPlane[k] / HRatioChroma[k] / PixelClock[k]; |
5765 | } else { |
5766 | DisplayPipeLineDeliveryTimeChromaPrefetch[k] = swath_width_chroma_ub[k] / PSCL_THROUGHPUT_CHROMA[k] / DPPCLK[k]; |
5767 | } |
5768 | } |
5769 | } |
5770 | |
5771 | for (k = 0; k < NumberOfActivePlanes; ++k) { |
5772 | if (SourceScan[k] != dm_vert) { |
5773 | req_per_swath_ub = swath_width_luma_ub[k] / BlockWidth256BytesY[k]; |
5774 | } else { |
5775 | req_per_swath_ub = swath_width_luma_ub[k] / BlockHeight256BytesY[k]; |
5776 | } |
5777 | DisplayPipeRequestDeliveryTimeLuma[k] = DisplayPipeLineDeliveryTimeLuma[k] / req_per_swath_ub; |
5778 | DisplayPipeRequestDeliveryTimeLumaPrefetch[k] = DisplayPipeLineDeliveryTimeLumaPrefetch[k] / req_per_swath_ub; |
5779 | if (BytePerPixelC[k] == 0) { |
5780 | DisplayPipeRequestDeliveryTimeChroma[k] = 0; |
5781 | DisplayPipeRequestDeliveryTimeChromaPrefetch[k] = 0; |
5782 | } else { |
5783 | if (SourceScan[k] != dm_vert) { |
5784 | req_per_swath_ub = swath_width_chroma_ub[k] / BlockWidth256BytesC[k]; |
5785 | } else { |
5786 | req_per_swath_ub = swath_width_chroma_ub[k] / BlockHeight256BytesC[k]; |
5787 | } |
5788 | DisplayPipeRequestDeliveryTimeChroma[k] = DisplayPipeLineDeliveryTimeChroma[k] / req_per_swath_ub; |
5789 | DisplayPipeRequestDeliveryTimeChromaPrefetch[k] = DisplayPipeLineDeliveryTimeChromaPrefetch[k] / req_per_swath_ub; |
5790 | } |
5791 | } |
5792 | |
5793 | for (k = 0; k < NumberOfActivePlanes; ++k) { |
5794 | int cursor_req_per_width = 0; |
5795 | cursor_req_per_width = dml_ceil(CursorWidth[k][0] * CursorBPP[k][0] / 256 / 8, 1); |
5796 | if (NumberOfCursors[k] > 0) { |
5797 | if (VRatio[k] <= 1) { |
5798 | CursorRequestDeliveryTime[k] = CursorWidth[k][0] / HRatio[k] / PixelClock[k] / cursor_req_per_width; |
5799 | } else { |
5800 | CursorRequestDeliveryTime[k] = CursorWidth[k][0] / PSCL_THROUGHPUT[k] / DPPCLK[k] / cursor_req_per_width; |
5801 | } |
5802 | if (VRatioPrefetchY[k] <= 1) { |
5803 | CursorRequestDeliveryTimePrefetch[k] = CursorWidth[k][0] / HRatio[k] / PixelClock[k] / cursor_req_per_width; |
5804 | } else { |
5805 | CursorRequestDeliveryTimePrefetch[k] = CursorWidth[k][0] / PSCL_THROUGHPUT[k] / DPPCLK[k] / cursor_req_per_width; |
5806 | } |
5807 | } else { |
5808 | CursorRequestDeliveryTime[k] = 0; |
5809 | CursorRequestDeliveryTimePrefetch[k] = 0; |
5810 | } |
5811 | } |
5812 | } |
5813 | |
5814 | static void CalculateMetaAndPTETimes( |
5815 | int NumberOfActivePlanes, |
5816 | bool_Bool GPUVMEnable, |
5817 | int MetaChunkSize, |
5818 | int MinMetaChunkSizeBytes, |
5819 | int HTotal[], |
5820 | double VRatio[], |
5821 | double VRatioChroma[], |
5822 | double DestinationLinesToRequestRowInVBlank[], |
5823 | double DestinationLinesToRequestRowInImmediateFlip[], |
5824 | bool_Bool DCCEnable[], |
5825 | double PixelClock[], |
5826 | int BytePerPixelY[], |
5827 | int BytePerPixelC[], |
5828 | enum scan_direction_class SourceScan[], |
5829 | int dpte_row_height[], |
5830 | int dpte_row_height_chroma[], |
5831 | int meta_row_width[], |
5832 | int meta_row_width_chroma[], |
5833 | int meta_row_height[], |
5834 | int meta_row_height_chroma[], |
5835 | int meta_req_width[], |
5836 | int meta_req_width_chroma[], |
5837 | int meta_req_height[], |
5838 | int meta_req_height_chroma[], |
5839 | int dpte_group_bytes[], |
5840 | int PTERequestSizeY[], |
5841 | int PTERequestSizeC[], |
5842 | int PixelPTEReqWidthY[], |
5843 | int PixelPTEReqHeightY[], |
5844 | int PixelPTEReqWidthC[], |
5845 | int PixelPTEReqHeightC[], |
5846 | int dpte_row_width_luma_ub[], |
5847 | int dpte_row_width_chroma_ub[], |
5848 | double DST_Y_PER_PTE_ROW_NOM_L[], |
5849 | double DST_Y_PER_PTE_ROW_NOM_C[], |
5850 | double DST_Y_PER_META_ROW_NOM_L[], |
5851 | double DST_Y_PER_META_ROW_NOM_C[], |
5852 | double TimePerMetaChunkNominal[], |
5853 | double TimePerChromaMetaChunkNominal[], |
5854 | double TimePerMetaChunkVBlank[], |
5855 | double TimePerChromaMetaChunkVBlank[], |
5856 | double TimePerMetaChunkFlip[], |
5857 | double TimePerChromaMetaChunkFlip[], |
5858 | double time_per_pte_group_nom_luma[], |
5859 | double time_per_pte_group_vblank_luma[], |
5860 | double time_per_pte_group_flip_luma[], |
5861 | double time_per_pte_group_nom_chroma[], |
5862 | double time_per_pte_group_vblank_chroma[], |
5863 | double time_per_pte_group_flip_chroma[]) |
5864 | { |
5865 | unsigned int meta_chunk_width = 0; |
5866 | unsigned int min_meta_chunk_width = 0; |
5867 | unsigned int meta_chunk_per_row_int = 0; |
5868 | unsigned int meta_row_remainder = 0; |
5869 | unsigned int meta_chunk_threshold = 0; |
5870 | unsigned int meta_chunks_per_row_ub = 0; |
5871 | unsigned int meta_chunk_width_chroma = 0; |
5872 | unsigned int min_meta_chunk_width_chroma = 0; |
5873 | unsigned int meta_chunk_per_row_int_chroma = 0; |
5874 | unsigned int meta_row_remainder_chroma = 0; |
5875 | unsigned int meta_chunk_threshold_chroma = 0; |
5876 | unsigned int meta_chunks_per_row_ub_chroma = 0; |
5877 | unsigned int dpte_group_width_luma = 0; |
5878 | unsigned int dpte_groups_per_row_luma_ub = 0; |
5879 | unsigned int dpte_group_width_chroma = 0; |
5880 | unsigned int dpte_groups_per_row_chroma_ub = 0; |
5881 | unsigned int k; |
5882 | |
5883 | for (k = 0; k < NumberOfActivePlanes; ++k) { |
5884 | DST_Y_PER_PTE_ROW_NOM_L[k] = dpte_row_height[k] / VRatio[k]; |
5885 | if (BytePerPixelC[k] == 0) { |
5886 | DST_Y_PER_PTE_ROW_NOM_C[k] = 0; |
5887 | } else { |
5888 | DST_Y_PER_PTE_ROW_NOM_C[k] = dpte_row_height_chroma[k] / VRatioChroma[k]; |
5889 | } |
5890 | DST_Y_PER_META_ROW_NOM_L[k] = meta_row_height[k] / VRatio[k]; |
5891 | if (BytePerPixelC[k] == 0) { |
5892 | DST_Y_PER_META_ROW_NOM_C[k] = 0; |
5893 | } else { |
5894 | DST_Y_PER_META_ROW_NOM_C[k] = meta_row_height_chroma[k] / VRatioChroma[k]; |
5895 | } |
5896 | } |
5897 | |
5898 | for (k = 0; k < NumberOfActivePlanes; ++k) { |
5899 | if (DCCEnable[k] == true1) { |
5900 | meta_chunk_width = MetaChunkSize * 1024 * 256 / BytePerPixelY[k] / meta_row_height[k]; |
5901 | min_meta_chunk_width = MinMetaChunkSizeBytes * 256 / BytePerPixelY[k] / meta_row_height[k]; |
5902 | meta_chunk_per_row_int = meta_row_width[k] / meta_chunk_width; |
5903 | meta_row_remainder = meta_row_width[k] % meta_chunk_width; |
5904 | if (SourceScan[k] != dm_vert) { |
5905 | meta_chunk_threshold = 2 * min_meta_chunk_width - meta_req_width[k]; |
5906 | } else { |
5907 | meta_chunk_threshold = 2 * min_meta_chunk_width - meta_req_height[k]; |
5908 | } |
5909 | if (meta_row_remainder <= meta_chunk_threshold) { |
5910 | meta_chunks_per_row_ub = meta_chunk_per_row_int + 1; |
5911 | } else { |
5912 | meta_chunks_per_row_ub = meta_chunk_per_row_int + 2; |
5913 | } |
5914 | TimePerMetaChunkNominal[k] = meta_row_height[k] / VRatio[k] * HTotal[k] / PixelClock[k] / meta_chunks_per_row_ub; |
5915 | TimePerMetaChunkVBlank[k] = DestinationLinesToRequestRowInVBlank[k] * HTotal[k] / PixelClock[k] / meta_chunks_per_row_ub; |
5916 | TimePerMetaChunkFlip[k] = DestinationLinesToRequestRowInImmediateFlip[k] * HTotal[k] / PixelClock[k] / meta_chunks_per_row_ub; |
5917 | if (BytePerPixelC[k] == 0) { |
5918 | TimePerChromaMetaChunkNominal[k] = 0; |
5919 | TimePerChromaMetaChunkVBlank[k] = 0; |
5920 | TimePerChromaMetaChunkFlip[k] = 0; |
5921 | } else { |
5922 | meta_chunk_width_chroma = MetaChunkSize * 1024 * 256 / BytePerPixelC[k] / meta_row_height_chroma[k]; |
5923 | min_meta_chunk_width_chroma = MinMetaChunkSizeBytes * 256 / BytePerPixelC[k] / meta_row_height_chroma[k]; |
5924 | meta_chunk_per_row_int_chroma = (double) meta_row_width_chroma[k] / meta_chunk_width_chroma; |
5925 | meta_row_remainder_chroma = meta_row_width_chroma[k] % meta_chunk_width_chroma; |
5926 | if (SourceScan[k] != dm_vert) { |
5927 | meta_chunk_threshold_chroma = 2 * min_meta_chunk_width_chroma - meta_req_width_chroma[k]; |
5928 | } else { |
5929 | meta_chunk_threshold_chroma = 2 * min_meta_chunk_width_chroma - meta_req_height_chroma[k]; |
5930 | } |
5931 | if (meta_row_remainder_chroma <= meta_chunk_threshold_chroma) { |
5932 | meta_chunks_per_row_ub_chroma = meta_chunk_per_row_int_chroma + 1; |
5933 | } else { |
5934 | meta_chunks_per_row_ub_chroma = meta_chunk_per_row_int_chroma + 2; |
5935 | } |
5936 | TimePerChromaMetaChunkNominal[k] = meta_row_height_chroma[k] / VRatioChroma[k] * HTotal[k] / PixelClock[k] / meta_chunks_per_row_ub_chroma; |
5937 | TimePerChromaMetaChunkVBlank[k] = DestinationLinesToRequestRowInVBlank[k] * HTotal[k] / PixelClock[k] / meta_chunks_per_row_ub_chroma; |
5938 | TimePerChromaMetaChunkFlip[k] = DestinationLinesToRequestRowInImmediateFlip[k] * HTotal[k] / PixelClock[k] / meta_chunks_per_row_ub_chroma; |
5939 | } |
5940 | } else { |
5941 | TimePerMetaChunkNominal[k] = 0; |
5942 | TimePerMetaChunkVBlank[k] = 0; |
5943 | TimePerMetaChunkFlip[k] = 0; |
5944 | TimePerChromaMetaChunkNominal[k] = 0; |
5945 | TimePerChromaMetaChunkVBlank[k] = 0; |
5946 | TimePerChromaMetaChunkFlip[k] = 0; |
5947 | } |
5948 | } |
5949 | |
5950 | for (k = 0; k < NumberOfActivePlanes; ++k) { |
5951 | if (GPUVMEnable == true1) { |
5952 | if (SourceScan[k] != dm_vert) { |
5953 | dpte_group_width_luma = dpte_group_bytes[k] / PTERequestSizeY[k] * PixelPTEReqWidthY[k]; |
5954 | } else { |
5955 | dpte_group_width_luma = dpte_group_bytes[k] / PTERequestSizeY[k] * PixelPTEReqHeightY[k]; |
5956 | } |
5957 | dpte_groups_per_row_luma_ub = dml_ceil(1.0 * dpte_row_width_luma_ub[k] / dpte_group_width_luma, 1); |
5958 | time_per_pte_group_nom_luma[k] = DST_Y_PER_PTE_ROW_NOM_L[k] * HTotal[k] / PixelClock[k] / dpte_groups_per_row_luma_ub; |
5959 | time_per_pte_group_vblank_luma[k] = DestinationLinesToRequestRowInVBlank[k] * HTotal[k] / PixelClock[k] / dpte_groups_per_row_luma_ub; |
5960 | time_per_pte_group_flip_luma[k] = DestinationLinesToRequestRowInImmediateFlip[k] * HTotal[k] / PixelClock[k] / dpte_groups_per_row_luma_ub; |
5961 | if (BytePerPixelC[k] == 0) { |
5962 | time_per_pte_group_nom_chroma[k] = 0; |
5963 | time_per_pte_group_vblank_chroma[k] = 0; |
5964 | time_per_pte_group_flip_chroma[k] = 0; |
5965 | } else { |
5966 | if (SourceScan[k] != dm_vert) { |
5967 | dpte_group_width_chroma = dpte_group_bytes[k] / PTERequestSizeC[k] * PixelPTEReqWidthC[k]; |
5968 | } else { |
5969 | dpte_group_width_chroma = dpte_group_bytes[k] / PTERequestSizeC[k] * PixelPTEReqHeightC[k]; |
5970 | } |
5971 | dpte_groups_per_row_chroma_ub = dml_ceil(1.0 * dpte_row_width_chroma_ub[k] / dpte_group_width_chroma, 1); |
5972 | time_per_pte_group_nom_chroma[k] = DST_Y_PER_PTE_ROW_NOM_C[k] * HTotal[k] / PixelClock[k] / dpte_groups_per_row_chroma_ub; |
5973 | time_per_pte_group_vblank_chroma[k] = DestinationLinesToRequestRowInVBlank[k] * HTotal[k] / PixelClock[k] / dpte_groups_per_row_chroma_ub; |
5974 | time_per_pte_group_flip_chroma[k] = DestinationLinesToRequestRowInImmediateFlip[k] * HTotal[k] / PixelClock[k] / dpte_groups_per_row_chroma_ub; |
5975 | } |
5976 | } else { |
5977 | time_per_pte_group_nom_luma[k] = 0; |
5978 | time_per_pte_group_vblank_luma[k] = 0; |
5979 | time_per_pte_group_flip_luma[k] = 0; |
5980 | time_per_pte_group_nom_chroma[k] = 0; |
5981 | time_per_pte_group_vblank_chroma[k] = 0; |
5982 | time_per_pte_group_flip_chroma[k] = 0; |
5983 | } |
5984 | } |
5985 | } |
5986 | |
5987 | static void CalculateVMGroupAndRequestTimes( |
5988 | unsigned int NumberOfActivePlanes, |
5989 | bool_Bool GPUVMEnable, |
5990 | unsigned int GPUVMMaxPageTableLevels, |
5991 | unsigned int HTotal[], |
5992 | int BytePerPixelC[], |
5993 | double DestinationLinesToRequestVMInVBlank[], |
5994 | double DestinationLinesToRequestVMInImmediateFlip[], |
5995 | bool_Bool DCCEnable[], |
5996 | double PixelClock[], |
5997 | int dpte_row_width_luma_ub[], |
5998 | int dpte_row_width_chroma_ub[], |
5999 | int vm_group_bytes[], |
6000 | unsigned int dpde0_bytes_per_frame_ub_l[], |
6001 | unsigned int dpde0_bytes_per_frame_ub_c[], |
6002 | int meta_pte_bytes_per_frame_ub_l[], |
6003 | int meta_pte_bytes_per_frame_ub_c[], |
6004 | double TimePerVMGroupVBlank[], |
6005 | double TimePerVMGroupFlip[], |
6006 | double TimePerVMRequestVBlank[], |
6007 | double TimePerVMRequestFlip[]) |
6008 | { |
6009 | int num_group_per_lower_vm_stage = 0; |
6010 | int num_req_per_lower_vm_stage = 0; |
6011 | unsigned int k; |
6012 | |
6013 | for (k = 0; k < NumberOfActivePlanes; ++k) { |
6014 | if (GPUVMEnable == true1 && (DCCEnable[k] == true1 || GPUVMMaxPageTableLevels > 1)) { |
6015 | if (DCCEnable[k] == false0) { |
6016 | if (BytePerPixelC[k] > 0) { |
6017 | num_group_per_lower_vm_stage = dml_ceil((double) (dpde0_bytes_per_frame_ub_l[k]) |
6018 | / (double) (vm_group_bytes[k]), 1) + dml_ceil((double) (dpde0_bytes_per_frame_ub_c[k]) |
6019 | / (double) (vm_group_bytes[k]), 1); |
6020 | } else { |
6021 | num_group_per_lower_vm_stage = dml_ceil((double) (dpde0_bytes_per_frame_ub_l[k]) |
6022 | / (double) (vm_group_bytes[k]), 1); |
6023 | } |
6024 | } else { |
6025 | if (GPUVMMaxPageTableLevels == 1) { |
6026 | if (BytePerPixelC[k] > 0) { |
6027 | num_group_per_lower_vm_stage = dml_ceil((double) (meta_pte_bytes_per_frame_ub_l[k]) |
6028 | / (double) (vm_group_bytes[k]), 1) + dml_ceil((double) (meta_pte_bytes_per_frame_ub_c[k]) |
6029 | / (double) (vm_group_bytes[k]), 1); |
6030 | } else { |
6031 | num_group_per_lower_vm_stage = dml_ceil((double) (meta_pte_bytes_per_frame_ub_l[k]) |
6032 | / (double) (vm_group_bytes[k]), 1); |
6033 | } |
6034 | } else { |
6035 | if (BytePerPixelC[k] > 0) { |
6036 | num_group_per_lower_vm_stage = 2 + dml_ceil((double) (dpde0_bytes_per_frame_ub_l[k]) / (double) (vm_group_bytes[k]), 1) |
6037 | + dml_ceil((double) (dpde0_bytes_per_frame_ub_c[k]) / (double) (vm_group_bytes[k]), 1) |
6038 | + dml_ceil((double) (meta_pte_bytes_per_frame_ub_l[k]) / (double) (vm_group_bytes[k]), 1) |
6039 | + dml_ceil((double) (meta_pte_bytes_per_frame_ub_c[k]) / (double) (vm_group_bytes[k]), 1); |
6040 | } else { |
6041 | num_group_per_lower_vm_stage = 1 + dml_ceil((double) (dpde0_bytes_per_frame_ub_l[k]) / (double) (vm_group_bytes[k]), 1) |
6042 | + dml_ceil((double) (meta_pte_bytes_per_frame_ub_l[k]) / (double) (vm_group_bytes[k]), 1); |
6043 | } |
6044 | } |
6045 | } |
6046 | |
6047 | if (DCCEnable[k] == false0) { |
6048 | if (BytePerPixelC[k] > 0) { |
6049 | num_req_per_lower_vm_stage = dpde0_bytes_per_frame_ub_l[k] / 64 + dpde0_bytes_per_frame_ub_c[k] / 64; |
6050 | } else { |
6051 | num_req_per_lower_vm_stage = dpde0_bytes_per_frame_ub_l[k] / 64; |
6052 | } |
6053 | } else { |
6054 | if (GPUVMMaxPageTableLevels == 1) { |
6055 | if (BytePerPixelC[k] > 0) { |
6056 | num_req_per_lower_vm_stage = meta_pte_bytes_per_frame_ub_l[k] / 64 |
6057 | + meta_pte_bytes_per_frame_ub_c[k] / 64; |
6058 | } else { |
6059 | num_req_per_lower_vm_stage = meta_pte_bytes_per_frame_ub_l[k] / 64; |
6060 | } |
6061 | } else { |
6062 | if (BytePerPixelC[k] > 0) { |
6063 | num_req_per_lower_vm_stage = dpde0_bytes_per_frame_ub_l[k] / 64 |
6064 | + dpde0_bytes_per_frame_ub_c[k] / 64 + meta_pte_bytes_per_frame_ub_l[k] |
6065 | / 64 + meta_pte_bytes_per_frame_ub_c[k] / 64; |
6066 | } else { |
6067 | num_req_per_lower_vm_stage = dpde0_bytes_per_frame_ub_l[k] / 64 |
6068 | + meta_pte_bytes_per_frame_ub_l[k] / 64; |
6069 | } |
6070 | } |
6071 | } |
6072 | |
6073 | TimePerVMGroupVBlank[k] = DestinationLinesToRequestVMInVBlank[k] * HTotal[k] / PixelClock[k] |
6074 | / num_group_per_lower_vm_stage; |
6075 | TimePerVMGroupFlip[k] = DestinationLinesToRequestVMInImmediateFlip[k] * HTotal[k] / PixelClock[k] |
6076 | / num_group_per_lower_vm_stage; |
6077 | TimePerVMRequestVBlank[k] = DestinationLinesToRequestVMInVBlank[k] * HTotal[k] / PixelClock[k] |
6078 | / num_req_per_lower_vm_stage; |
6079 | TimePerVMRequestFlip[k] = DestinationLinesToRequestVMInImmediateFlip[k] * HTotal[k] / PixelClock[k] |
6080 | / num_req_per_lower_vm_stage; |
6081 | |
6082 | if (GPUVMMaxPageTableLevels > 2) { |
6083 | TimePerVMGroupVBlank[k] = TimePerVMGroupVBlank[k] / 2; |
6084 | TimePerVMGroupFlip[k] = TimePerVMGroupFlip[k] / 2; |
6085 | TimePerVMRequestVBlank[k] = TimePerVMRequestVBlank[k] / 2; |
6086 | TimePerVMRequestFlip[k] = TimePerVMRequestFlip[k] / 2; |
6087 | } |
6088 | |
6089 | } else { |
6090 | TimePerVMGroupVBlank[k] = 0; |
6091 | TimePerVMGroupFlip[k] = 0; |
6092 | TimePerVMRequestVBlank[k] = 0; |
6093 | TimePerVMRequestFlip[k] = 0; |
6094 | } |
6095 | } |
6096 | } |
6097 | |
6098 | static void CalculateStutterEfficiency( |
6099 | int NumberOfActivePlanes, |
6100 | long ROBBufferSizeInKByte, |
6101 | double TotalDataReadBandwidth, |
6102 | double DCFCLK, |
6103 | double ReturnBW, |
6104 | double SRExitTime, |
6105 | bool_Bool SynchronizedVBlank, |
6106 | int DPPPerPlane[], |
6107 | double DETBufferSizeY[], |
6108 | int BytePerPixelY[], |
6109 | double BytePerPixelDETY[], |
6110 | double SwathWidthY[], |
6111 | int SwathHeightY[], |
6112 | int SwathHeightC[], |
6113 | double DCCRateLuma[], |
6114 | double DCCRateChroma[], |
6115 | int HTotal[], |
6116 | int VTotal[], |
6117 | double PixelClock[], |
6118 | double VRatio[], |
6119 | enum scan_direction_class SourceScan[], |
6120 | int BlockHeight256BytesY[], |
6121 | int BlockWidth256BytesY[], |
6122 | int BlockHeight256BytesC[], |
6123 | int BlockWidth256BytesC[], |
6124 | int DCCYMaxUncompressedBlock[], |
6125 | int DCCCMaxUncompressedBlock[], |
6126 | int VActive[], |
6127 | bool_Bool DCCEnable[], |
6128 | bool_Bool WritebackEnable[], |
6129 | double ReadBandwidthPlaneLuma[], |
6130 | double ReadBandwidthPlaneChroma[], |
6131 | double meta_row_bw[], |
6132 | double dpte_row_bw[], |
6133 | double *StutterEfficiencyNotIncludingVBlank, |
6134 | double *StutterEfficiency, |
6135 | double *StutterPeriodOut) |
6136 | { |
6137 | double FullDETBufferingTimeY[DC__NUM_DPP__MAX8] = { 0 }; |
6138 | double FrameTimeForMinFullDETBufferingTime = 0; |
6139 | double StutterPeriod = 0; |
6140 | double AverageReadBandwidth = 0; |
6141 | double TotalRowReadBandwidth = 0; |
6142 | double AverageDCCCompressionRate = 0; |
6143 | double PartOfBurstThatFitsInROB = 0; |
6144 | double StutterBurstTime = 0; |
6145 | int TotalActiveWriteback = 0; |
6146 | double VBlankTime = 0; |
6147 | double SmallestVBlank = 0; |
6148 | int BytePerPixelYCriticalPlane = 0; |
6149 | double SwathWidthYCriticalPlane = 0; |
6150 | double LinesInDETY[DC__NUM_DPP__MAX8] = { 0 }; |
6151 | double LinesInDETYRoundedDownToSwath[DC__NUM_DPP__MAX8] = { 0 }; |
6152 | double LinesToFinishSwathTransferStutterCriticalPlane = 0; |
6153 | double MaximumEffectiveCompressionLuma = 0; |
6154 | double MaximumEffectiveCompressionChroma = 0; |
6155 | unsigned int k; |
6156 | |
6157 | for (k = 0; k < NumberOfActivePlanes; ++k) { |
6158 | LinesInDETY[k] = DETBufferSizeY[k] / BytePerPixelDETY[k] / SwathWidthY[k]; |
6159 | LinesInDETYRoundedDownToSwath[k] = dml_floor(LinesInDETY[k], SwathHeightY[k]); |
6160 | FullDETBufferingTimeY[k] = LinesInDETYRoundedDownToSwath[k] * (HTotal[k] / PixelClock[k]) / VRatio[k]; |
6161 | } |
6162 | |
6163 | StutterPeriod = FullDETBufferingTimeY[0]; |
6164 | FrameTimeForMinFullDETBufferingTime = VTotal[0] * HTotal[0] / PixelClock[0]; |
6165 | BytePerPixelYCriticalPlane = BytePerPixelY[0]; |
6166 | SwathWidthYCriticalPlane = SwathWidthY[0]; |
6167 | LinesToFinishSwathTransferStutterCriticalPlane = SwathHeightY[0] |
6168 | - (LinesInDETY[0] - LinesInDETYRoundedDownToSwath[0]); |
6169 | |
6170 | for (k = 0; k < NumberOfActivePlanes; ++k) { |
6171 | if (FullDETBufferingTimeY[k] < StutterPeriod) { |
6172 | StutterPeriod = FullDETBufferingTimeY[k]; |
6173 | FrameTimeForMinFullDETBufferingTime = VTotal[k] * HTotal[k] / PixelClock[k]; |
6174 | BytePerPixelYCriticalPlane = BytePerPixelY[k]; |
6175 | SwathWidthYCriticalPlane = SwathWidthY[k]; |
6176 | LinesToFinishSwathTransferStutterCriticalPlane = SwathHeightY[k] |
6177 | - (LinesInDETY[k] - LinesInDETYRoundedDownToSwath[k]); |
6178 | } |
6179 | } |
6180 | |
6181 | AverageReadBandwidth = 0; |
6182 | TotalRowReadBandwidth = 0; |
6183 | for (k = 0; k < NumberOfActivePlanes; ++k) { |
6184 | if (DCCEnable[k] == true1) { |
6185 | if ((SourceScan[k] == dm_vert && BlockWidth256BytesY[k] > SwathHeightY[k]) |
6186 | || (SourceScan[k] != dm_vert |
6187 | && BlockHeight256BytesY[k] > SwathHeightY[k]) |
6188 | || DCCYMaxUncompressedBlock[k] < 256) { |
6189 | MaximumEffectiveCompressionLuma = 2; |
6190 | } else { |
6191 | MaximumEffectiveCompressionLuma = 4; |
6192 | } |
6193 | AverageReadBandwidth = AverageReadBandwidth + ReadBandwidthPlaneLuma[k] / dml_min(DCCRateLuma[k], MaximumEffectiveCompressionLuma); |
6194 | |
6195 | if (ReadBandwidthPlaneChroma[k] > 0) { |
6196 | if ((SourceScan[k] == dm_vert && BlockWidth256BytesC[k] > SwathHeightC[k]) |
6197 | || (SourceScan[k] != dm_vert && BlockHeight256BytesC[k] > SwathHeightC[k]) |
6198 | || DCCCMaxUncompressedBlock[k] < 256) { |
6199 | MaximumEffectiveCompressionChroma = 2; |
6200 | } else { |
6201 | MaximumEffectiveCompressionChroma = 4; |
6202 | } |
6203 | AverageReadBandwidth = AverageReadBandwidth + ReadBandwidthPlaneChroma[k] / dml_min(DCCRateChroma[k], MaximumEffectiveCompressionChroma); |
6204 | } |
6205 | } else { |
6206 | AverageReadBandwidth = AverageReadBandwidth + ReadBandwidthPlaneLuma[k] + ReadBandwidthPlaneChroma[k]; |
6207 | } |
6208 | TotalRowReadBandwidth = TotalRowReadBandwidth + DPPPerPlane[k] * (meta_row_bw[k] + dpte_row_bw[k]); |
6209 | } |
6210 | |
6211 | AverageDCCCompressionRate = TotalDataReadBandwidth / AverageReadBandwidth; |
6212 | PartOfBurstThatFitsInROB = dml_min(StutterPeriod * TotalDataReadBandwidth, ROBBufferSizeInKByte * 1024 * AverageDCCCompressionRate); |
6213 | StutterBurstTime = PartOfBurstThatFitsInROB / AverageDCCCompressionRate / ReturnBW + (StutterPeriod * TotalDataReadBandwidth |
6214 | - PartOfBurstThatFitsInROB) / (DCFCLK * 64) + StutterPeriod * TotalRowReadBandwidth / ReturnBW; |
6215 | StutterBurstTime = dml_max(StutterBurstTime, LinesToFinishSwathTransferStutterCriticalPlane * BytePerPixelYCriticalPlane * SwathWidthYCriticalPlane / ReturnBW); |
6216 | |
6217 | TotalActiveWriteback = 0; |
6218 | for (k = 0; k < NumberOfActivePlanes; ++k) { |
6219 | if (WritebackEnable[k] == true1) { |
6220 | TotalActiveWriteback = TotalActiveWriteback + 1; |
6221 | } |
6222 | } |
6223 | |
6224 | if (TotalActiveWriteback == 0) { |
6225 | *StutterEfficiencyNotIncludingVBlank = (1 |
6226 | - (SRExitTime + StutterBurstTime) / StutterPeriod) * 100; |
6227 | } else { |
6228 | *StutterEfficiencyNotIncludingVBlank = 0; |
6229 | } |
6230 | |
6231 | if (SynchronizedVBlank == true1 || NumberOfActivePlanes == 1) { |
6232 | SmallestVBlank = (VTotal[0] - VActive[0]) * HTotal[0] / PixelClock[0]; |
6233 | } else { |
6234 | SmallestVBlank = 0; |
6235 | } |
6236 | for (k = 0; k < NumberOfActivePlanes; ++k) { |
6237 | if (SynchronizedVBlank == true1 || NumberOfActivePlanes == 1) { |
6238 | VBlankTime = (VTotal[k] - VActive[k]) * HTotal[k] / PixelClock[k]; |
6239 | } else { |
6240 | VBlankTime = 0; |
6241 | } |
6242 | SmallestVBlank = dml_min(SmallestVBlank, VBlankTime); |
6243 | } |
6244 | |
6245 | *StutterEfficiency = (*StutterEfficiencyNotIncludingVBlank / 100.0 * (FrameTimeForMinFullDETBufferingTime - SmallestVBlank) + SmallestVBlank) / FrameTimeForMinFullDETBufferingTime * 100; |
6246 | |
6247 | if (StutterPeriodOut) |
6248 | *StutterPeriodOut = StutterPeriod; |
6249 | } |
6250 | |
6251 | static void CalculateSwathAndDETConfiguration( |
6252 | bool_Bool ForceSingleDPP, |
6253 | int NumberOfActivePlanes, |
6254 | long DETBufferSizeInKByte, |
6255 | double MaximumSwathWidthLuma[], |
6256 | double MaximumSwathWidthChroma[], |
6257 | enum scan_direction_class SourceScan[], |
6258 | enum source_format_class SourcePixelFormat[], |
6259 | enum dm_swizzle_mode SurfaceTiling[], |
6260 | int ViewportWidth[], |
6261 | int ViewportHeight[], |
6262 | int SurfaceWidthY[], |
6263 | int SurfaceWidthC[], |
6264 | int SurfaceHeightY[], |
6265 | int SurfaceHeightC[], |
6266 | int Read256BytesBlockHeightY[], |
6267 | int Read256BytesBlockHeightC[], |
6268 | int Read256BytesBlockWidthY[], |
6269 | int Read256BytesBlockWidthC[], |
6270 | enum odm_combine_mode ODMCombineEnabled[], |
6271 | int BlendingAndTiming[], |
6272 | int BytePerPixY[], |
6273 | int BytePerPixC[], |
6274 | double BytePerPixDETY[], |
6275 | double BytePerPixDETC[], |
6276 | int HActive[], |
6277 | double HRatio[], |
6278 | double HRatioChroma[], |
6279 | int DPPPerPlane[], |
6280 | int swath_width_luma_ub[], |
6281 | int swath_width_chroma_ub[], |
6282 | double SwathWidth[], |
6283 | double SwathWidthChroma[], |
6284 | int SwathHeightY[], |
6285 | int SwathHeightC[], |
6286 | double DETBufferSizeY[], |
6287 | double DETBufferSizeC[], |
6288 | bool_Bool ViewportSizeSupportPerPlane[], |
6289 | bool_Bool *ViewportSizeSupport) |
6290 | { |
6291 | int MaximumSwathHeightY[DC__NUM_DPP__MAX8] = { 0 }; |
6292 | int MaximumSwathHeightC[DC__NUM_DPP__MAX8] = { 0 }; |
6293 | int MinimumSwathHeightY = 0; |
6294 | int MinimumSwathHeightC = 0; |
6295 | long RoundedUpMaxSwathSizeBytesY = 0; |