-
-
Notifications
You must be signed in to change notification settings - Fork 1.5k
/
GSDeviceOGL.cpp
2150 lines (1760 loc) · 70.4 KB
/
GSDeviceOGL.cpp
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
/*
* Copyright (C) 2011-2016 PCSX2 Dev Team
* Copyright (C) 2007-2009 Gabest
*
* This Program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2, or (at your option)
* any later version.
*
* This Program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with GNU Make; see the file COPYING. If not, write to
* the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA USA.
* http://www.gnu.org/copyleft/gpl.html
*
*/
#include "stdafx.h"
#include "GSState.h"
#include "GSDeviceOGL.h"
#include "GLState.h"
#include "GSUtil.h"
#include "GSOsdManager.h"
#include <fstream>
//#define ONLY_LINES
#ifdef _WIN32
#include "resource.h"
#else
#include "GSdxResources.h"
#endif
// TODO port those value into PerfMon API
#ifdef ENABLE_OGL_DEBUG_MEM_BW
uint64 g_real_texture_upload_byte = 0;
uint64 g_vertex_upload_byte = 0;
uint64 g_uniform_upload_byte = 0;
#endif
static const uint32 g_merge_cb_index = 10;
static const uint32 g_interlace_cb_index = 11;
static const uint32 g_fx_cb_index = 14;
static const uint32 g_convert_index = 15;
static const uint32 g_vs_cb_index = 20;
static const uint32 g_ps_cb_index = 21;
static const uint32 g_gs_cb_index = 22;
bool GSDeviceOGL::m_debug_gl_call = false;
int GSDeviceOGL::m_shader_inst = 0;
int GSDeviceOGL::m_shader_reg = 0;
FILE* GSDeviceOGL::m_debug_gl_file = NULL;
GSDeviceOGL::GSDeviceOGL()
: m_msaa(0)
, m_force_texture_clear(0)
, m_fbo(0)
, m_fbo_read(0)
, m_va(NULL)
, m_apitrace(0)
, m_palette_ss(0)
, m_vs_cb(NULL)
, m_ps_cb(NULL)
, m_font(NULL)
, m_shader(NULL)
{
memset(&m_merge_obj, 0, sizeof(m_merge_obj));
memset(&m_interlace, 0, sizeof(m_interlace));
memset(&m_convert, 0, sizeof(m_convert));
memset(&m_fxaa, 0, sizeof(m_fxaa));
memset(&m_shaderfx, 0, sizeof(m_shaderfx));
memset(&m_date, 0, sizeof(m_date));
memset(&m_shadeboost, 0, sizeof(m_shadeboost));
memset(&m_om_dss, 0, sizeof(m_om_dss));
memset(&m_profiler, 0 , sizeof(m_profiler));
GLState::Clear();
m_mipmap = theApp.GetConfigI("mipmap");
m_filter = static_cast<TriFiltering>(theApp.GetConfigI("UserHacks_TriFilter"));
if (!theApp.GetConfigB("UserHacks"))
m_filter = TriFiltering::None;
// Reset the debug file
#ifdef ENABLE_OGL_DEBUG
if (theApp.GetCurrentRendererType() == GSRendererType::OGL_SW)
m_debug_gl_file = fopen("GSdx_opengl_debug_sw.txt","w");
else
m_debug_gl_file = fopen("GSdx_opengl_debug_hw.txt","w");
#endif
m_debug_gl_call = theApp.GetConfigB("debug_opengl");
m_disable_hw_gl_draw = theApp.GetConfigB("disable_hw_gl_draw");
}
GSDeviceOGL::~GSDeviceOGL()
{
if (m_debug_gl_file) {
fclose(m_debug_gl_file);
m_debug_gl_file = NULL;
}
// If the create function wasn't called nothing to do.
if (m_shader == NULL)
return;
GL_PUSH("GSDeviceOGL destructor");
// Clean vertex buffer state
delete m_va;
// Clean m_merge_obj
delete m_merge_obj.cb;
// Clean m_interlace
delete m_interlace.cb;
// Clean m_convert
delete m_convert.dss;
delete m_convert.dss_write;
delete m_convert.cb;
// Clean m_fxaa
delete m_fxaa.cb;
// Clean m_shaderfx
delete m_shaderfx.cb;
// Clean m_date
delete m_date.dss;
// Clean various opengl allocation
glDeleteFramebuffers(1, &m_fbo);
glDeleteFramebuffers(1, &m_fbo_read);
// Delete HW FX
delete m_vs_cb;
delete m_ps_cb;
glDeleteSamplers(1, &m_palette_ss);
m_ps.clear();
glDeleteSamplers(countof(m_ps_ss), m_ps_ss);
for (uint32 key = 0; key < countof(m_om_dss); key++) delete m_om_dss[key];
PboPool::Destroy();
// Must be done after the destruction of all shader/program objects
delete m_shader;
m_shader = NULL;
// Purge any pending message to reduce noise in Valgrind (potential memory leak
// in Mesa driver that doesn't free internal buffer when the context is destroyed)
CheckDebugLog();
}
void GSDeviceOGL::GenerateProfilerData()
{
if (m_profiler.last_query < 3) {
glDeleteQueries(1 << 16, m_profiler.timer_query);
return;
}
// Wait latest quey to get valid result
GLuint available = 0;
while (!available) {
glGetQueryObjectuiv(m_profiler.timer(), GL_QUERY_RESULT_AVAILABLE, &available);
}
GLuint64 time_start;
GLuint64 time_end;
std::vector<double> times;
double ms = 0.000001;
int replay = theApp.GetConfigI("linux_replay");
int first_query = replay > 1 ? m_profiler.last_query / replay : 0;
glGetQueryObjectui64v(m_profiler.timer_query[first_query], GL_QUERY_RESULT, &time_start);
for (uint32 q = first_query + 1; q < m_profiler.last_query; q++) {
glGetQueryObjectui64v(m_profiler.timer_query[q], GL_QUERY_RESULT, &time_end);
uint64 t = time_end - time_start;
times.push_back((double)t * ms);
time_start = time_end;
}
// Latest value is often silly, just drop it
times.pop_back();
glDeleteQueries(1 << 16, m_profiler.timer_query);
double frames = times.size();
double mean = 0.0;
double sd = 0.0;
auto minmax_time = std::minmax_element(times.begin(), times.end());
for (auto t : times) mean += t;
mean = mean / frames;
for (auto t : times) sd += pow(t-mean, 2);
sd = sqrt(sd / frames);
uint32 time_repartition[16] = {0};
for (auto t : times) {
uint32 slot = (uint32)(t/2.0);
if (slot >= countof(time_repartition)) {
slot = countof(time_repartition) - 1;
}
time_repartition[slot]++;
}
fprintf(stderr, "\nPerformance Profile for %.0f frames:\n", frames);
fprintf(stderr, "Min %4.2f ms\t(%4.2f fps)\n", *minmax_time.first, 1000.0 / *minmax_time.first);
fprintf(stderr, "Mean %4.2f ms\t(%4.2f fps)\n", mean, 1000.0 / mean);
fprintf(stderr, "Max %4.2f ms\t(%4.2f fps)\n", *minmax_time.second, 1000.0 / *minmax_time.second);
fprintf(stderr, "SD %4.2f ms\n", sd);
fprintf(stderr, "\n");
fprintf(stderr, "Frame Repartition\n");
for (uint32 i = 0; i < countof(time_repartition); i ++) {
fprintf(stderr, "%3u ms => %3u ms\t%4u\n", 2 * i, 2 * (i+1), time_repartition[i]);
}
FILE* csv = fopen("GSdx_profile.csv", "w");
if (csv) {
for (size_t i = 0; i < times.size(); i++) {
fprintf(csv, "%zu,%lf\n", i, times[i]);
}
fclose(csv);
}
}
GSTexture* GSDeviceOGL::CreateSurface(int type, int w, int h, bool msaa, int fmt)
{
GL_PUSH("Create surface");
// A wrapper to call GSTextureOGL, with the different kind of parameter
GSTextureOGL* t = new GSTextureOGL(type, w, h, fmt, m_fbo_read, m_mipmap > 1 || m_filter != TriFiltering::None);
// NOTE: I'm not sure RenderTarget always need to be cleared. It could be costly for big upscale.
// FIXME: it will be more logical to do it in FetchSurface. This code is only called at first creation
// of the texture. However we could reuse a deleted texture.
if (m_force_texture_clear == 0) {
switch(type)
{
case GSTexture::RenderTarget:
ClearRenderTarget(t, 0);
break;
case GSTexture::DepthStencil:
ClearDepth(t);
// No need to clear the stencil now.
break;
}
}
return t;
}
GSTexture* GSDeviceOGL::FetchSurface(int type, int w, int h, bool msaa, int format)
{
GSTexture* t = GSDevice::FetchSurface(type, w, h, false, format);
if (m_force_texture_clear) {
GSVector4 red(1.0f, 0.0f, 0.0f, 1.0f);
switch(type)
{
case GSTexture::RenderTarget:
ClearRenderTarget(t, 0);
break;
case GSTexture::DepthStencil:
ClearDepth(t);
// No need to clear the stencil now.
break;
case GSTexture::Texture:
if (m_force_texture_clear > 1)
static_cast<GSTextureOGL*>(t)->Clear((void*)&red);
else if (m_force_texture_clear)
static_cast<GSTextureOGL*>(t)->Clear(NULL);
break;
}
}
return t;
}
bool GSDeviceOGL::Create(const std::shared_ptr<GSWnd> &wnd)
{
std::vector<char> shader;
// ****************************************************************
// Debug helper
// ****************************************************************
#ifdef ENABLE_OGL_DEBUG
if (theApp.GetConfigB("debug_opengl")) {
glDebugMessageCallback((GLDEBUGPROC)DebugOutputToFile, NULL);
glEnable(GL_DEBUG_OUTPUT_SYNCHRONOUS_ARB);
glDebugMessageControl(GL_DONT_CARE, GL_DONT_CARE, GL_DONT_CARE, 0, NULL, true);
// Useless info message on Nvidia driver
GLuint ids[] = {0x20004};
glDebugMessageControl(GL_DEBUG_SOURCE_API_ARB, GL_DEBUG_TYPE_OTHER_ARB, GL_DONT_CARE, countof(ids), ids, false);
}
#endif
m_force_texture_clear = theApp.GetConfigI("force_texture_clear");
// WARNING it must be done after the control setup (at least on MESA)
GL_PUSH("GSDeviceOGL::Create");
// ****************************************************************
// Various object
// ****************************************************************
{
GL_PUSH("GSDeviceOGL::Various");
m_shader = new GSShaderOGL(theApp.GetConfigB("debug_glsl_shader"));
glGenFramebuffers(1, &m_fbo);
// Always write to the first buffer
OMSetFBO(m_fbo);
GLenum target[1] = {GL_COLOR_ATTACHMENT0};
glDrawBuffers(1, target);
OMSetFBO(0);
glGenFramebuffers(1, &m_fbo_read);
// Always read from the first buffer
glBindFramebuffer(GL_READ_FRAMEBUFFER, m_fbo_read);
glReadBuffer(GL_COLOR_ATTACHMENT0);
glBindFramebuffer(GL_READ_FRAMEBUFFER, 0);
// Some timers to help profiling
if (GLLoader::in_replayer) {
glCreateQueries(GL_TIMESTAMP, 1 << 16, m_profiler.timer_query);
}
}
// ****************************************************************
// Vertex buffer state
// ****************************************************************
{
GL_PUSH("GSDeviceOGL::Vertex Buffer");
static_assert(sizeof(GSVertexPT1) == sizeof(GSVertex), "wrong GSVertex size");
GSInputLayoutOGL il_convert[] =
{
{2 , GL_FLOAT , GL_FALSE , sizeof(GSVertexPT1) , (const GLvoid*)(0) } ,
{2 , GL_FLOAT , GL_FALSE , sizeof(GSVertexPT1) , (const GLvoid*)(16) } ,
{4 , GL_UNSIGNED_BYTE , GL_FALSE , sizeof(GSVertex) , (const GLvoid*)(8) } ,
{1 , GL_FLOAT , GL_FALSE , sizeof(GSVertex) , (const GLvoid*)(12) } ,
{2 , GL_UNSIGNED_SHORT , GL_FALSE , sizeof(GSVertex) , (const GLvoid*)(16) } ,
{1 , GL_UNSIGNED_INT , GL_FALSE , sizeof(GSVertex) , (const GLvoid*)(20) } ,
{2 , GL_UNSIGNED_SHORT , GL_FALSE , sizeof(GSVertex) , (const GLvoid*)(24) } ,
{4 , GL_UNSIGNED_BYTE , GL_TRUE , sizeof(GSVertex) , (const GLvoid*)(28) } , // Only 1 byte is useful but hardware unit only support 4B
};
m_va = new GSVertexBufferStateOGL(il_convert, countof(il_convert));
}
// ****************************************************************
// Pre Generate the different sampler object
// ****************************************************************
{
GL_PUSH("GSDeviceOGL::Sampler");
for (uint32 key = 0; key < countof(m_ps_ss); key++) {
m_ps_ss[key] = CreateSampler(PSSamplerSelector(key));
}
}
// ****************************************************************
// convert
// ****************************************************************
GLuint vs = 0;
GLuint ps = 0;
{
GL_PUSH("GSDeviceOGL::Convert");
m_convert.cb = new GSUniformBufferOGL("Misc UBO", g_convert_index, sizeof(MiscConstantBuffer));
// Upload once and forget about it
m_misc_cb_cache.ScalingFactor = GSVector4i(theApp.GetConfigI("upscale_multiplier"));
m_convert.cb->cache_upload(&m_misc_cb_cache);
theApp.LoadResource(IDR_CONVERT_GLSL, shader);
vs = m_shader->Compile("convert.glsl", "vs_main", GL_VERTEX_SHADER, shader.data());
m_convert.vs = vs;
for(size_t i = 0; i < countof(m_convert.ps); i++) {
ps = m_shader->Compile("convert.glsl", format("ps_main%d", i), GL_FRAGMENT_SHADER, shader.data());
std::string pretty_name = "Convert pipe " + std::to_string(i);
m_convert.ps[i] = m_shader->LinkPipeline(pretty_name, vs, 0, ps);
}
PSSamplerSelector point;
m_convert.pt = GetSamplerID(point);
PSSamplerSelector bilinear;
bilinear.biln = true;
m_convert.ln = GetSamplerID(bilinear);
m_convert.dss = new GSDepthStencilOGL();
m_convert.dss_write = new GSDepthStencilOGL();
m_convert.dss_write->EnableDepth();
m_convert.dss_write->SetDepth(GL_ALWAYS, true);
}
// ****************************************************************
// merge
// ****************************************************************
{
GL_PUSH("GSDeviceOGL::Merge");
m_merge_obj.cb = new GSUniformBufferOGL("Merge UBO", g_merge_cb_index, sizeof(MergeConstantBuffer));
theApp.LoadResource(IDR_MERGE_GLSL, shader);
for(size_t i = 0; i < countof(m_merge_obj.ps); i++) {
ps = m_shader->Compile("merge.glsl", format("ps_main%d", i), GL_FRAGMENT_SHADER, shader.data());
std::string pretty_name = "Merge pipe " + std::to_string(i);
m_merge_obj.ps[i] = m_shader->LinkPipeline(pretty_name, vs, 0, ps);
}
}
// ****************************************************************
// interlace
// ****************************************************************
{
GL_PUSH("GSDeviceOGL::Interlace");
m_interlace.cb = new GSUniformBufferOGL("Interlace UBO", g_interlace_cb_index, sizeof(InterlaceConstantBuffer));
theApp.LoadResource(IDR_INTERLACE_GLSL, shader);
for(size_t i = 0; i < countof(m_interlace.ps); i++) {
ps = m_shader->Compile("interlace.glsl", format("ps_main%d", i), GL_FRAGMENT_SHADER, shader.data());
std::string pretty_name = "Interlace pipe " + std::to_string(i);
m_interlace.ps[i] = m_shader->LinkPipeline(pretty_name, vs, 0, ps);
}
}
// ****************************************************************
// Shade boost
// ****************************************************************
{
GL_PUSH("GSDeviceOGL::Shadeboost");
int ShadeBoost_Contrast = theApp.GetConfigI("ShadeBoost_Contrast");
int ShadeBoost_Brightness = theApp.GetConfigI("ShadeBoost_Brightness");
int ShadeBoost_Saturation = theApp.GetConfigI("ShadeBoost_Saturation");
std::string shade_macro = format("#define SB_SATURATION %d.0\n", ShadeBoost_Saturation)
+ format("#define SB_BRIGHTNESS %d.0\n", ShadeBoost_Brightness)
+ format("#define SB_CONTRAST %d.0\n", ShadeBoost_Contrast);
theApp.LoadResource(IDR_SHADEBOOST_GLSL, shader);
ps = m_shader->Compile("shadeboost.glsl", "ps_main", GL_FRAGMENT_SHADER, shader.data(), shade_macro);
m_shadeboost.ps = m_shader->LinkPipeline("ShadeBoost pipe", vs, 0, ps);
}
// ****************************************************************
// rasterization configuration
// ****************************************************************
{
GL_PUSH("GSDeviceOGL::Rasterization");
#ifdef ONLY_LINES
glLineWidth(5.0);
glPolygonMode(GL_FRONT_AND_BACK, GL_LINE);
#else
glPolygonMode(GL_FRONT_AND_BACK, GL_FILL);
#endif
glDisable(GL_CULL_FACE);
glEnable(GL_SCISSOR_TEST);
glDisable(GL_MULTISAMPLE);
glDisable(GL_DITHER); // Honestly I don't know!
}
// ****************************************************************
// DATE
// ****************************************************************
{
GL_PUSH("GSDeviceOGL::Date");
m_date.dss = new GSDepthStencilOGL();
m_date.dss->EnableStencil();
m_date.dss->SetStencil(GL_ALWAYS, GL_REPLACE);
}
// ****************************************************************
// Use DX coordinate convention
// ****************************************************************
// VS gl_position.z => [-1,-1]
// FS depth => [0, 1]
// because of -1 we loose lot of precision for small GS value
// This extension allow FS depth to range from -1 to 1. So
// gl_position.z could range from [0, 1]
// Change depth convention
if (GLLoader::found_GL_ARB_clip_control && glClipControl)
glClipControl(GL_LOWER_LEFT, GL_ZERO_TO_ONE);
// ****************************************************************
// HW renderer shader
// ****************************************************************
CreateTextureFX();
// ****************************************************************
// Pbo Pool allocation
// ****************************************************************
{
GL_PUSH("GSDeviceOGL::PBO");
// Mesa seems to use it to compute the row length. In our case, we are
// tightly packed so don't bother with this parameter and set it to the
// minimum alignment (1 byte)
glPixelStorei(GL_UNPACK_ALIGNMENT, 1);
PboPool::Init();
}
// ****************************************************************
// Get Available Memory
// ****************************************************************
GLint vram[4] = {0};
if (GLLoader::vendor_id_amd) {
// Full vram, remove a small margin for others buffer
glGetIntegerv(GL_TEXTURE_FREE_MEMORY_ATI, vram);
} else if (GLLoader::found_GL_NVX_gpu_memory_info) {
// GL_GPU_MEMORY_INFO_TOTAL_AVAILABLE_MEMORY_NVX <= give full memory
// Available vram
glGetIntegerv(GL_GPU_MEMORY_INFO_CURRENT_AVAILABLE_VIDMEM_NVX, vram);
} else {
fprintf(stdout, "No extenstion supported to get available memory. Use default value !\n");
}
// When VRAM is at least 2GB, we set the limit to the default i.e. 3.8 GB
// When VRAM is below 2GB, we add a factor 2 because RAM can be used. Potentially
// low VRAM gpu can go higher but perf will be bad anyway.
if (vram[0] > 0 && vram[0] < 1800000)
GLState::available_vram = (int64)(vram[0]) * 1024ul * 2ul;
fprintf(stdout, "Available VRAM/RAM:%lldMB for textures\n", GLState::available_vram >> 20u);
// ****************************************************************
// Texture Font (OSD)
// ****************************************************************
GSVector2i tex_font = m_osd.get_texture_font_size();
m_font = new GSTextureOGL(GSTextureOGL::Texture, tex_font.x, tex_font.y, GL_R8, m_fbo_read, false);
// ****************************************************************
// Finish window setup and backbuffer
// ****************************************************************
if(!GSDevice::Create(wnd))
return false;
GSVector4i rect = wnd->GetClientRect();
Reset(rect.z, rect.w);
// Basic to ensure structures are correctly packed
static_assert(sizeof(VSSelector) == 4, "Wrong VSSelector size");
static_assert(sizeof(PSSelector) == 8, "Wrong PSSelector size");
static_assert(sizeof(PSSamplerSelector) == 4, "Wrong PSSamplerSelector size");
static_assert(sizeof(OMDepthStencilSelector) == 4, "Wrong OMDepthStencilSelector size");
static_assert(sizeof(OMColorMaskSelector) == 4, "Wrong OMColorMaskSelector size");
return true;
}
void GSDeviceOGL::CreateTextureFX()
{
GL_PUSH("GSDeviceOGL::CreateTextureFX");
m_vs_cb = new GSUniformBufferOGL("HW VS UBO", g_vs_cb_index, sizeof(VSConstantBuffer));
m_ps_cb = new GSUniformBufferOGL("HW PS UBO", g_ps_cb_index, sizeof(PSConstantBuffer));
theApp.LoadResource(IDR_TFX_VGS_GLSL, m_shader_tfx_vgs);
theApp.LoadResource(IDR_TFX_FS_GLSL, m_shader_tfx_fs);
// warning 1 sampler by image unit. So you cannot reuse m_ps_ss...
m_palette_ss = CreateSampler(PSSamplerSelector(0));
glBindSampler(1, m_palette_ss);
// Pre compile the (remaining) Geometry & Vertex Shader
// One-Hot encoding
memset(m_gs, 0, sizeof(m_gs));
m_gs[1] = CompileGS(GSSelector(1));
m_gs[2] = CompileGS(GSSelector(2));
m_gs[4] = CompileGS(GSSelector(4));
for (uint32 key = 0; key < countof(m_vs); key++)
m_vs[key] = CompileVS(VSSelector(key));
// Enable all bits for stencil operations. Technically 1 bit is
// enough but buffer is polluted with noise. Clear will be limited
// to the mask.
glStencilMask(0xFF);
for (uint32 key = 0; key < countof(m_om_dss); key++) {
m_om_dss[key] = CreateDepthStencil(OMDepthStencilSelector(key));
}
// Help to debug FS in apitrace
m_apitrace = CompilePS(PSSelector());
}
bool GSDeviceOGL::Reset(int w, int h)
{
if(!GSDevice::Reset(w, h))
return false;
// Opengl allocate the backbuffer with the window. The render is done in the backbuffer when
// there isn't any FBO. Only a dummy texture is created to easily detect when the rendering is done
// in the backbuffer
m_backbuffer = new GSTextureOGL(GSTextureOGL::Backbuffer, w, h, 0, m_fbo_read, false);
return true;
}
void GSDeviceOGL::SetVSync(int vsync)
{
m_wnd->SetVSync(vsync);
}
void GSDeviceOGL::Flip()
{
#ifdef ENABLE_OGL_DEBUG
CheckDebugLog();
#endif
m_wnd->Flip();
if (GLLoader::in_replayer) {
glQueryCounter(m_profiler.timer(), GL_TIMESTAMP);
m_profiler.last_query++;
}
}
void GSDeviceOGL::BeforeDraw()
{
}
void GSDeviceOGL::AfterDraw()
{
}
void GSDeviceOGL::DrawPrimitive()
{
BeforeDraw();
m_va->DrawPrimitive();
AfterDraw();
}
void GSDeviceOGL::DrawPrimitive(int offset, int count)
{
BeforeDraw();
m_va->DrawPrimitive(offset, count);
AfterDraw();
}
void GSDeviceOGL::DrawIndexedPrimitive()
{
BeforeDraw();
if (!m_disable_hw_gl_draw)
m_va->DrawIndexedPrimitive();
AfterDraw();
}
void GSDeviceOGL::DrawIndexedPrimitive(int offset, int count)
{
//ASSERT(offset + count <= (int)m_index.count);
BeforeDraw();
if (!m_disable_hw_gl_draw)
m_va->DrawIndexedPrimitive(offset, count);
AfterDraw();
}
void GSDeviceOGL::ClearRenderTarget(GSTexture* t, const GSVector4& c)
{
if (!t) return;
GSTextureOGL* T = static_cast<GSTextureOGL*>(t);
if (T->HasBeenCleaned() && !T->IsBackbuffer())
return;
// Performance note: potentially T->Clear() could be used. Main purpose of
// Clear() is to avoid the framebuffer setup cost. However, in this context,
// the texture 't' will be set as the render target of the framebuffer and
// therefore will require a framebuffer setup.
// So using the old/standard path is faster/better albeit verbose.
GL_PUSH("Clear RT %d", T->GetID());
// TODO: check size of scissor before toggling it
glDisable(GL_SCISSOR_TEST);
uint32 old_color_mask = GLState::wrgba;
OMSetColorMaskState();
if (T->IsBackbuffer()) {
OMSetFBO(0);
// glDrawBuffer(GL_BACK); // this is the default when there is no FB
// 0 will select the first drawbuffer ie GL_BACK
glClearBufferfv(GL_COLOR, 0, c.v);
} else {
OMSetFBO(m_fbo);
OMAttachRt(T);
glClearBufferfv(GL_COLOR, 0, c.v);
}
OMSetColorMaskState(OMColorMaskSelector(old_color_mask));
glEnable(GL_SCISSOR_TEST);
T->WasCleaned();
}
void GSDeviceOGL::ClearRenderTarget(GSTexture* t, uint32 c)
{
if (!t) return;
GSVector4 color = GSVector4::rgba32(c) * (1.0f / 255);
ClearRenderTarget(t, color);
}
void GSDeviceOGL::ClearDepth(GSTexture* t)
{
if (!t) return;
GSTextureOGL* T = static_cast<GSTextureOGL*>(t);
GL_PUSH("Clear Depth %d", T->GetID());
if (0 && GLLoader::found_GL_ARB_clear_texture) {
// I don't know what the driver does but it creates
// some slowdowns on Harry Potter PS
// Maybe it triggers some texture relocations, or maybe
// it clears also the stencil value (2 times slower)
//
// Let's disable this code for the moment.
// Don't bother with Depth_Stencil insanity
T->Clear(NULL);
} else {
OMSetFBO(m_fbo);
// RT must be detached, if RT is too small, depth won't be fully cleared
// AT tolenico 2 map clip bug
OMAttachRt(NULL);
OMAttachDs(T);
// TODO: check size of scissor before toggling it
glDisable(GL_SCISSOR_TEST);
float c = 0.0f;
if (GLState::depth_mask) {
glClearBufferfv(GL_DEPTH, 0, &c);
} else {
glDepthMask(true);
glClearBufferfv(GL_DEPTH, 0, &c);
glDepthMask(false);
}
glEnable(GL_SCISSOR_TEST);
}
}
void GSDeviceOGL::ClearStencil(GSTexture* t, uint8 c)
{
if (!t) return;
GSTextureOGL* T = static_cast<GSTextureOGL*>(t);
GL_PUSH("Clear Stencil %d", T->GetID());
// Keep SCISSOR_TEST enabled on purpose to reduce the size
// of clean in DATE (impact big upscaling)
OMSetFBO(m_fbo);
OMAttachDs(T);
GLint color = c;
glClearBufferiv(GL_STENCIL, 0, &color);
}
GLuint GSDeviceOGL::CreateSampler(PSSamplerSelector sel)
{
GL_PUSH("Create Sampler");
GLuint sampler;
glCreateSamplers(1, &sampler);
// Bilinear filtering
if (sel.biln) {
glSamplerParameteri(sampler, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
glSamplerParameteri(sampler, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
} else {
glSamplerParameteri(sampler, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
glSamplerParameteri(sampler, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
}
switch (static_cast<GS_MIN_FILTER>(sel.triln)) {
case GS_MIN_FILTER::Nearest:
// Nop based on biln
break;
case GS_MIN_FILTER::Linear:
// Nop based on biln
break;
case GS_MIN_FILTER::Nearest_Mipmap_Nearest:
glSamplerParameteri(sampler, GL_TEXTURE_MIN_FILTER, GL_NEAREST_MIPMAP_NEAREST);
break;
case GS_MIN_FILTER::Nearest_Mipmap_Linear:
glSamplerParameteri(sampler, GL_TEXTURE_MIN_FILTER, GL_NEAREST_MIPMAP_LINEAR);
break;
case GS_MIN_FILTER::Linear_Mipmap_Nearest:
glSamplerParameteri(sampler, GL_TEXTURE_MIN_FILTER, GL_LINEAR_MIPMAP_NEAREST);
break;
case GS_MIN_FILTER::Linear_Mipmap_Linear:
glSamplerParameteri(sampler, GL_TEXTURE_MIN_FILTER, GL_LINEAR_MIPMAP_LINEAR);
break;
default:
break;
}
//glSamplerParameterf(sampler, GL_TEXTURE_MIN_LOD, 0);
//glSamplerParameterf(sampler, GL_TEXTURE_MAX_LOD, 6);
if (sel.tau)
glSamplerParameteri(sampler, GL_TEXTURE_WRAP_S, GL_REPEAT);
else
glSamplerParameteri(sampler, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
if (sel.tav)
glSamplerParameteri(sampler, GL_TEXTURE_WRAP_T, GL_REPEAT);
else
glSamplerParameteri(sampler, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
glSamplerParameteri(sampler, GL_TEXTURE_WRAP_R, GL_CLAMP_TO_EDGE);
int anisotropy = theApp.GetConfigI("MaxAnisotropy");
if (GLLoader::found_GL_EXT_texture_filter_anisotropic && anisotropy && sel.aniso)
glSamplerParameterf(sampler, GL_TEXTURE_MAX_ANISOTROPY_EXT, (float)anisotropy);
return sampler;
}
GLuint GSDeviceOGL::GetSamplerID(PSSamplerSelector ssel)
{
return m_ps_ss[ssel];
}
GSDepthStencilOGL* GSDeviceOGL::CreateDepthStencil(OMDepthStencilSelector dssel)
{
GSDepthStencilOGL* dss = new GSDepthStencilOGL();
if (dssel.date)
{
dss->EnableStencil();
if (dssel.date_one)
dss->SetStencil(GL_EQUAL, GL_ZERO);
else
dss->SetStencil(GL_EQUAL, GL_KEEP);
}
if(dssel.ztst != ZTST_ALWAYS || dssel.zwe)
{
static const GLenum ztst[] =
{
GL_NEVER,
GL_ALWAYS,
GL_GEQUAL,
GL_GREATER
};
dss->EnableDepth();
dss->SetDepth(ztst[dssel.ztst], dssel.zwe);
}
return dss;
}
void GSDeviceOGL::InitPrimDateTexture(GSTexture* rt, const GSVector4i& area)
{
const GSVector2i& rtsize = rt->GetSize();
// Create a texture to avoid the useless clean@0
if (m_date.t == NULL)
m_date.t = CreateTexture(rtsize.x, rtsize.y, GL_R32I);
// Clean with the max signed value
int max_int = 0x7FFFFFFF;
static_cast<GSTextureOGL*>(m_date.t)->Clear(&max_int, area);
glBindImageTexture(2, static_cast<GSTextureOGL*>(m_date.t)->GetID(), 0, false, 0, GL_READ_WRITE, GL_R32I);
#ifdef ENABLE_OGL_DEBUG
// Help to see the texture in apitrace
PSSetShaderResource(2, m_date.t);
#endif
}
void GSDeviceOGL::RecycleDateTexture()
{
if (m_date.t) {
//static_cast<GSTextureOGL*>(m_date.t)->Save(format("/tmp/date_adv_%04ld.csv", GSState::s_n));
Recycle(m_date.t);
m_date.t = NULL;
}
}
void GSDeviceOGL::Barrier(GLbitfield b)
{
glMemoryBarrier(b);
}
GLuint GSDeviceOGL::CompileVS(VSSelector sel)
{
std::string macro = format("#define VS_INT_FST %d\n", sel.int_fst);
if (GLLoader::buggy_sso_dual_src)
return m_shader->CompileShader("tfx_vgs.glsl", "vs_main", GL_VERTEX_SHADER, m_shader_tfx_vgs.data(), macro);
else
return m_shader->Compile("tfx_vgs.glsl", "vs_main", GL_VERTEX_SHADER, m_shader_tfx_vgs.data(), macro);
}
GLuint GSDeviceOGL::CompileGS(GSSelector sel)
{
std::string macro = format("#define GS_POINT %d\n", sel.point)
+ format("#define GS_LINE %d\n", sel.line);
if (GLLoader::buggy_sso_dual_src)
return m_shader->CompileShader("tfx_vgs.glsl", "gs_main", GL_GEOMETRY_SHADER, m_shader_tfx_vgs.data(), macro);
else
return m_shader->Compile("tfx_vgs.glsl", "gs_main", GL_GEOMETRY_SHADER, m_shader_tfx_vgs.data(), macro);
}
GLuint GSDeviceOGL::CompilePS(PSSelector sel)
{
std::string macro = format("#define PS_FST %d\n", sel.fst)
+ format("#define PS_WMS %d\n", sel.wms)
+ format("#define PS_WMT %d\n", sel.wmt)
+ format("#define PS_TEX_FMT %d\n", sel.tex_fmt)
+ format("#define PS_DFMT %d\n", sel.dfmt)
+ format("#define PS_DEPTH_FMT %d\n", sel.depth_fmt)
+ format("#define PS_CHANNEL_FETCH %d\n", sel.channel)
+ format("#define PS_URBAN_CHAOS_HLE %d\n", sel.urban_chaos_hle)
+ format("#define PS_TALES_OF_ABYSS_HLE %d\n", sel.tales_of_abyss_hle)
+ format("#define PS_TEX_IS_FB %d\n", sel.tex_is_fb)
+ format("#define PS_AEM %d\n", sel.aem)
+ format("#define PS_TFX %d\n", sel.tfx)
+ format("#define PS_TCC %d\n", sel.tcc)
+ format("#define PS_ATST %d\n", sel.atst)
+ format("#define PS_FOG %d\n", sel.fog)
+ format("#define PS_CLR1 %d\n", sel.clr1)
+ format("#define PS_FBA %d\n", sel.fba)
+ format("#define PS_LTF %d\n", sel.ltf)
+ format("#define PS_AUTOMATIC_LOD %d\n", sel.automatic_lod)
+ format("#define PS_MANUAL_LOD %d\n", sel.manual_lod)
+ format("#define PS_COLCLIP %d\n", sel.colclip)
+ format("#define PS_DATE %d\n", sel.date)
+ format("#define PS_TCOFFSETHACK %d\n", sel.tcoffsethack)
//+ format("#define PS_POINT_SAMPLER %d\n", sel.point_sampler)
+ format("#define PS_BLEND_A %d\n", sel.blend_a)
+ format("#define PS_BLEND_B %d\n", sel.blend_b)
+ format("#define PS_BLEND_C %d\n", sel.blend_c)
+ format("#define PS_BLEND_D %d\n", sel.blend_d)
+ format("#define PS_IIP %d\n", sel.iip)
+ format("#define PS_SHUFFLE %d\n", sel.shuffle)
+ format("#define PS_READ_BA %d\n", sel.read_ba)
+ format("#define PS_WRITE_RG %d\n", sel.write_rg)
+ format("#define PS_FBMASK %d\n", sel.fbmask)
+ format("#define PS_HDR %d\n", sel.hdr)
+ format("#define PS_PABE %d\n", sel.pabe);
;
if (GLLoader::buggy_sso_dual_src)
return m_shader->CompileShader("tfx.glsl", "ps_main", GL_FRAGMENT_SHADER, m_shader_tfx_fs.data(), macro);
else
return m_shader->Compile("tfx.glsl", "ps_main", GL_FRAGMENT_SHADER, m_shader_tfx_fs.data(), macro);
}
void GSDeviceOGL::SelfShaderTestRun(const std::string& dir, const std::string& file, const PSSelector& sel, int& nb_shader)
{
#ifdef __unix__
std::string out = "/tmp/GSdx_Shader/";
GSmkdir(out.c_str());
out += dir + "/";
GSmkdir(out.c_str());
out += file;
#else
std::string out = file;
#endif
#ifdef __linux__
// Nouveau actually