From 32b78a85726eb1d82c461ceefa066fc723cf4413 Mon Sep 17 00:00:00 2001 From: NeoBrainX Date: Wed, 1 May 2013 11:39:30 +0200 Subject: [PATCH] First steps of making the pixel shader uid struct more compact. --- Source/Core/VideoCommon/Src/BPMemory.h | 2 +- .../Core/VideoCommon/Src/LightingShaderGen.h | 30 +-- .../Core/VideoCommon/Src/PixelShaderGen.cpp | 91 ++++--- Source/Core/VideoCommon/Src/PixelShaderGen.h | 246 ++++++++---------- Source/Core/VideoCommon/Src/ShaderGenCommon.h | 17 +- 5 files changed, 181 insertions(+), 205 deletions(-) diff --git a/Source/Core/VideoCommon/Src/BPMemory.h b/Source/Core/VideoCommon/Src/BPMemory.h index 0ae29a3022..5254ba7bd7 100644 --- a/Source/Core/VideoCommon/Src/BPMemory.h +++ b/Source/Core/VideoCommon/Src/BPMemory.h @@ -376,7 +376,7 @@ struct TevStageCombiner union TwoTevStageOrders { - struct + struct { u32 texmap0 : 3; // indirect tex stage texmap u32 texcoord0 : 3; diff --git a/Source/Core/VideoCommon/Src/LightingShaderGen.h b/Source/Core/VideoCommon/Src/LightingShaderGen.h index fde60be891..36bc52749a 100644 --- a/Source/Core/VideoCommon/Src/LightingShaderGen.h +++ b/Source/Core/VideoCommon/Src/LightingShaderGen.h @@ -54,8 +54,8 @@ static void GenerateLightShader(T& object, LightingUidData& uid_data, int index, else if (coloralpha == 2) swizzle = "w"; - uid_data.lit_chans[litchan_index].attnfunc = chan.attnfunc; - uid_data.lit_chans[litchan_index].diffusefunc = chan.diffusefunc; + uid_data.attnfunc |= chan.attnfunc << (2*litchan_index); + uid_data.diffusefunc |= chan.diffusefunc << (2*litchan_index); if (!(chan.attnfunc & 1)) { // atten disabled @@ -124,7 +124,7 @@ static void GenerateLightingShader(T& object, LightingUidData& uid_data, int com object.Write("{\n"); - uid_data.lit_chans[j].matsource = xfregs.color[j].matsource; + uid_data.matsource |= xfregs.color[j].matsource << j; if (color.matsource) // from vertex { if (components & (VB_HAS_COL0 << j)) @@ -139,10 +139,10 @@ static void GenerateLightingShader(T& object, LightingUidData& uid_data, int com object.Write("mat = %s[%d];\n", materialsName, j+2); } - uid_data.lit_chans[j].enablelighting = xfregs.color[j].enablelighting; + uid_data.enablelighting |= xfregs.color[j].enablelighting << j; if (color.enablelighting) { - uid_data.lit_chans[j].ambsource = xfregs.color[j].ambsource; + uid_data.ambsource |= xfregs.color[j].ambsource << j; if (color.ambsource) // from vertex { if (components & (VB_HAS_COL0<() != NULL) ? out.template GetUidData() : dummy_data; + ERROR_LOG(VIDEO, "%lu", sizeof(pixel_shader_uid_data)); + out.SetBuffer(text); if (out.GetBuffer() != NULL) setlocale(LC_NUMERIC, "C"); // Reset locale for compilation @@ -257,9 +259,9 @@ static void GeneratePixelShader(T& out, DSTALPHA_MODE dstAlphaMode, API_TYPE Api numStages, numTexgen, bpmem.genMode.numindstages); uid_data.dstAlphaMode = dstAlphaMode; - uid_data.genMode.numindstages = bpmem.genMode.numindstages; - uid_data.genMode.numtevstages = bpmem.genMode.numtevstages; - uid_data.genMode.numtexgens = bpmem.genMode.numtexgens; + uid_data.genMode_numindstages = bpmem.genMode.numindstages; + uid_data.genMode_numtevstages = bpmem.genMode.numtevstages; + uid_data.genMode_numtexgens = bpmem.genMode.numtexgens; if (ApiType == API_OPENGL) { @@ -466,7 +468,7 @@ static void GeneratePixelShader(T& out, DSTALPHA_MODE dstAlphaMode, API_TYPE Api for (unsigned int i = 0; i < numTexgen; ++i) { // optional perspective divides - uid_data.texMtxInfo[i].projection = xfregs.texMtxInfo[i].projection; + uid_data.texMtxInfo_n_projection |= xfregs.texMtxInfo[i].projection << i; if (xfregs.texMtxInfo[i].projection == XF_TEXPROJ_STQ) { out.Write("\tif (uv%d.z != 0.0f)", i); @@ -496,7 +498,7 @@ static void GeneratePixelShader(T& out, DSTALPHA_MODE dstAlphaMode, API_TYPE Api unsigned int texcoord = bpmem.tevindref.getTexCoord(i); unsigned int texmap = bpmem.tevindref.getTexMap(i); - uid_data.tevindref.SetValues(i, texcoord, texmap); + uid_data.SetTevindrefValues(i, texcoord, texmap); if (texcoord < numTexgen) { out.SetConstantsUsed(C_INDTEXSCALE+i/2,C_INDTEXSCALE+i/2); @@ -566,7 +568,7 @@ static void GeneratePixelShader(T& out, DSTALPHA_MODE dstAlphaMode, API_TYPE Api // Note: depth textures are disabled if early depth test is enabled uid_data.ztex_op = bpmem.ztex2.op; uid_data.per_pixel_depth = per_pixel_depth; - uid_data.fog.fsel = bpmem.fog.c_proj_fsel.fsel; + uid_data.fog_fsel = bpmem.fog.c_proj_fsel.fsel; // depth texture can safely be ignored if the result won't be written to the depth buffer (early_ztest) and isn't used for fog either bool skip_ztexture = !per_pixel_depth && !bpmem.fog.c_proj_fsel.fsel; @@ -687,13 +689,12 @@ static void WriteStage(T& out, pixel_shader_uid_data& uid_data, int n, API_TYPE out.Write("// TEV stage %d\n", n); uid_data.bHasIndStage |= bHasIndStage << n; - if (n < 8) { uid_data.tevorders_n_texcoord1 |= texcoord << (3 * n); } - else uid_data.tevorders_n_texcoord2 |= texcoord << (3 * n - 24); + uid_data.tevorders_n_texcoord |= texcoord << (3 * n); if (bHasIndStage) { - uid_data.tevind_n.bs |= bpmem.tevind[n].bs << (2*n); - uid_data.tevind_n.bt |= bpmem.tevind[n].bt << (2*n); - uid_data.tevind_n.fmt |= bpmem.tevind[n].fmt << (2*n); + uid_data.tevind_n_bs |= bpmem.tevind[n].bs << (2*n); + uid_data.tevind_n_bt |= bpmem.tevind[n].bt << (2*n); + uid_data.tevind_n_fmt |= bpmem.tevind[n].fmt << (2*n); out.Write("// indirect op\n"); // perform the indirect op on the incoming regular coordinates using indtex%d as the offset coords @@ -708,12 +709,12 @@ static void WriteStage(T& out, pixel_shader_uid_data& uid_data, int n, API_TYPE out.Write("float3 indtevcrd%d = indtex%d * %s;\n", n, bpmem.tevind[n].bt, tevIndFmtScale[bpmem.tevind[n].fmt]); // bias - uid_data.tevind_n.Set_bias(n, bpmem.tevind[n].bias); + uid_data.Set_tevind_bias(n, bpmem.tevind[n].bias); if (bpmem.tevind[n].bias != ITB_NONE ) out.Write("indtevcrd%d.%s += %s;\n", n, tevIndBiasField[bpmem.tevind[n].bias], tevIndBiasAdd[bpmem.tevind[n].fmt]); // multiply by offset matrix and scale - uid_data.tevind_n.Set_mid(n, bpmem.tevind[n].mid); + uid_data.Set_tevind_mid(n, bpmem.tevind[n].mid); if (bpmem.tevind[n].mid != 0) { if (bpmem.tevind[n].mid <= 3) @@ -750,9 +751,9 @@ static void WriteStage(T& out, pixel_shader_uid_data& uid_data, int n, API_TYPE // --------- // Wrapping // --------- - uid_data.tevind_n.Set_sw(n, bpmem.tevind[n].sw); - uid_data.tevind_n.Set_tw(n, bpmem.tevind[n].tw); - uid_data.tevind_n.fb_addprev |= bpmem.tevind[n].fb_addprev << n; + uid_data.Set_tevind_sw(n, bpmem.tevind[n].sw); + uid_data.Set_tevind_tw(n, bpmem.tevind[n].tw); + uid_data.tevind_n_fb_addprev |= bpmem.tevind[n].fb_addprev << n; // wrap S if (bpmem.tevind[n].sw == ITW_OFF) @@ -779,8 +780,26 @@ static void WriteStage(T& out, pixel_shader_uid_data& uid_data, int n, API_TYPE TevStageCombiner::ColorCombiner &cc = bpmem.combiners[n].colorC; TevStageCombiner::AlphaCombiner &ac = bpmem.combiners[n].alphaC; - uid_data.combiners[n].colorC.hex = cc.hex & 0xFFFFFF; - uid_data.combiners[n].alphaC.hex = ac.hex & 0xFFFFFF; + uid_data.cc_n_d = cc.d; + uid_data.cc_n_c = cc.c; + uid_data.cc_n_b = cc.b; + uid_data.cc_n_a = cc.a; + uid_data.cc_n_bias = cc.bias; + uid_data.cc_n_op = cc.op; + uid_data.cc_n_clamp = cc.clamp; + uid_data.cc_n_shift = cc.shift; + uid_data.cc_n_dest = cc.dest; + uid_data.ac_n_rswap = ac.rswap; + uid_data.ac_n_tswap = ac.tswap; + uid_data.ac_n_d = ac.d; + uid_data.ac_n_c = ac.c; + uid_data.ac_n_b = ac.b; + uid_data.ac_n_a = ac.a; + uid_data.ac_n_bias = ac.bias; + uid_data.ac_n_op = ac.op; + uid_data.ac_n_clamp = ac.clamp; + uid_data.ac_n_shift = ac.shift; + uid_data.ac_n_dest = ac.dest; if(cc.a == TEVCOLORARG_RASA || cc.a == TEVCOLORARG_RASC || cc.b == TEVCOLORARG_RASA || cc.b == TEVCOLORARG_RASC @@ -790,10 +809,10 @@ static void WriteStage(T& out, pixel_shader_uid_data& uid_data, int n, API_TYPE || ac.c == TEVALPHAARG_RASA || ac.d == TEVALPHAARG_RASA) { const int i = bpmem.combiners[n].alphaC.rswap; - uid_data.tevksel[i*2 ].swap1 = bpmem.tevksel[i*2 ].swap1; - uid_data.tevksel[i*2+1].swap1 = bpmem.tevksel[i*2+1].swap1; - uid_data.tevksel[i*2 ].swap2 = bpmem.tevksel[i*2 ].swap2; - uid_data.tevksel[i*2+1].swap2 = bpmem.tevksel[i*2+1].swap2; + uid_data.tevksel_n_swap1 = bpmem.tevksel[i*2 ].swap1 << (2 * (i*2 )); + uid_data.tevksel_n_swap1 = bpmem.tevksel[i*2+1].swap1 << (2 * (i*2+1)); + uid_data.tevksel_n_swap2 = bpmem.tevksel[i*2 ].swap2 << (2 * (i*2 )); + uid_data.tevksel_n_swap2 = bpmem.tevksel[i*2+1].swap2 << (2 * (i*2+1)); char *rasswap = swapModeTable[bpmem.combiners[n].alphaC.rswap]; out.Write("rastemp = %s.%s;\n", tevRasTable[bpmem.tevorders[n / 2].getColorChan(n & 1)], rasswap); @@ -813,14 +832,14 @@ static void WriteStage(T& out, pixel_shader_uid_data& uid_data, int n, API_TYPE } const int i = bpmem.combiners[n].alphaC.tswap; - uid_data.tevksel[i*2 ].swap1 = bpmem.tevksel[i*2 ].swap1; - uid_data.tevksel[i*2+1].swap1 = bpmem.tevksel[i*2+1].swap1; - uid_data.tevksel[i*2 ].swap2 = bpmem.tevksel[i*2 ].swap2; - uid_data.tevksel[i*2+1].swap2 = bpmem.tevksel[i*2+1].swap2; + uid_data.tevksel_n_swap1 = bpmem.tevksel[i*2 ].swap1 << (2 * (i*2 )); + uid_data.tevksel_n_swap1 = bpmem.tevksel[i*2+1].swap1 << (2 * (i*2+1)); + uid_data.tevksel_n_swap2 = bpmem.tevksel[i*2 ].swap2 << (2 * (i*2 )); + uid_data.tevksel_n_swap2 = bpmem.tevksel[i*2+1].swap2 << (2 * (i*2+1)); char *texswap = swapModeTable[bpmem.combiners[n].alphaC.tswap]; int texmap = bpmem.tevorders[n/2].getTexMap(n&1); - uid_data.tevindref.SetTexmap(i, texmap); + uid_data.SetTevindrefTexmap(i, texmap); SampleTexture(out, "textemp", "tevcoord", texswap, texmap, ApiType); } else @@ -834,8 +853,8 @@ static void WriteStage(T& out, pixel_shader_uid_data& uid_data, int n, API_TYPE { int kc = bpmem.tevksel[n / 2].getKC(n & 1); int ka = bpmem.tevksel[n / 2].getKA(n & 1); - uid_data.tevksel[n/2].set_kcsel(n & 1, kc); - uid_data.tevksel[n/2].set_kasel(n & 1, ka); + uid_data.set_tevksel_kcsel(n/2, n & 1, kc); + uid_data.set_tevksel_kasel(n/2, n & 1, ka); out.Write("konsttemp = float4(%s, %s);\n", tevKSelTableC[kc], tevKSelTableA[ka]); if(kc > 7 || ka > 7) { @@ -1086,9 +1105,9 @@ static void WriteAlphaTest(T& out, pixel_shader_uid_data& uid_data, API_TYPE Api // using discard then return works the same in cg and dx9 but not in dx11 out.Write("\tif(!( "); - uid_data.alpha_test.comp0 = bpmem.alpha_test.comp0; - uid_data.alpha_test.logic = bpmem.alpha_test.comp1; - uid_data.alpha_test.logic = bpmem.alpha_test.logic; + uid_data.alpha_test_comp0 = bpmem.alpha_test.comp0; + uid_data.alpha_test_logic = bpmem.alpha_test.comp1; + uid_data.alpha_test_logic = bpmem.alpha_test.logic; // Lookup the first component from the alpha function table int compindex = bpmem.alpha_test.comp0; @@ -1117,7 +1136,7 @@ static void WriteAlphaTest(T& out, pixel_shader_uid_data& uid_data, API_TYPE Api // when the alpha test fail. This is not a correct implementation because // even if the depth test fails the fragment could be alpha blended, but // we don't have a choice. - uid_data.alpha_test.use_zcomploc_hack = bpmem.zcontrol.early_ztest && bpmem.zmode.updateenable; + uid_data.alpha_test_use_zcomploc_hack = bpmem.zcontrol.early_ztest && bpmem.zmode.updateenable; if (!(bpmem.zcontrol.early_ztest && bpmem.zmode.updateenable)) { out.Write("\t\tdiscard;\n"); @@ -1143,11 +1162,11 @@ static const char *tevFogFuncsTable[] = template static void WriteFog(T& out, pixel_shader_uid_data& uid_data) { - uid_data.fog.fsel = bpmem.fog.c_proj_fsel.fsel; + uid_data.fog_fsel = bpmem.fog.c_proj_fsel.fsel; if(bpmem.fog.c_proj_fsel.fsel == 0) return; // no Fog - uid_data.fog.proj = bpmem.fog.c_proj_fsel.proj; + uid_data.fog_proj = bpmem.fog.c_proj_fsel.proj; out.SetConstantsUsed(C_FOG, C_FOG+1); if (bpmem.fog.c_proj_fsel.proj == 0) @@ -1166,7 +1185,7 @@ static void WriteFog(T& out, pixel_shader_uid_data& uid_data) // x_adjust = sqrt((x-center)^2 + k^2)/k // ze *= x_adjust // this is completely theoretical as the real hardware seems to use a table intead of calculating the values. - uid_data.fog.RangeBaseEnabled = bpmem.fogRange.Base.Enabled; + uid_data.fog_RangeBaseEnabled = bpmem.fogRange.Base.Enabled; if (bpmem.fogRange.Base.Enabled) { out.SetConstantsUsed(C_FOG+2, C_FOG+2); diff --git a/Source/Core/VideoCommon/Src/PixelShaderGen.h b/Source/Core/VideoCommon/Src/PixelShaderGen.h index 2340b96af7..5ae3febfa3 100644 --- a/Source/Core/VideoCommon/Src/PixelShaderGen.h +++ b/Source/Core/VideoCommon/Src/PixelShaderGen.h @@ -55,162 +55,119 @@ const s_svar PSVar_Loc[] = { {I_COLORS, C_COLORS, 4 }, {I_PMATERIALS, C_PMATERIALS, 4 }, }; +// TODO: Packing? +//#pragma pack(4) struct pixel_shader_uid_data { u32 components; - DSTALPHA_MODE dstAlphaMode; // TODO: as u32 :2 - AlphaTest::TEST_RESULT Pretest; // TODO: As :2 + u32 dstAlphaMode : 2; + u32 Pretest : 2; + + u32 genMode_numtexgens : 4; + u32 genMode_numtevstages : 4; + u32 genMode_numindstages : 3; + u32 nIndirectStagesUsed : 8; - struct { - u32 numtexgens : 4; - u32 numtevstages : 4; - u32 numindstages : 3; - } genMode; - struct + + u32 texMtxInfo_n_unknown : 8; // 8x1 bit + u32 texMtxInfo_n_projection : 8; // 8x1 bit + u32 texMtxInfo_n_inputform : 16; // 8x2 bits + u32 texMtxInfo_n_texgentype : 24; // 8x3 bits + u64 texMtxInfo_n_sourcerow : 40; // 8x5 bits + u32 texMtxInfo_n_embosssourceshift : 24; // 8x3 bits + u32 texMtxInfo_n_embosslightshift : 24; // 8x3 bits + + u32 tevindref_bi0 : 3; + u32 tevindref_bc0 : 3; + u32 tevindref_bi1 : 3; + u32 tevindref_bc1 : 3; + u32 tevindref_bi2 : 3; + u32 tevindref_bc3 : 3; + u32 tevindref_bi4 : 3; + u32 tevindref_bc4 : 3; + inline void SetTevindrefValues(int index, u32 texcoord, u32 texmap) { - u32 unknown : 1; - u32 projection : 1; // XF_TEXPROJ_X - u32 inputform : 2; // XF_TEXINPUT_X - u32 texgentype : 3; // XF_TEXGEN_X - u32 sourcerow : 5; // XF_SRCGEOM_X - u32 embosssourceshift : 3; // what generated texcoord to use - u32 embosslightshift : 3; // light index that is used - } texMtxInfo[8]; - struct + if (index == 0) { tevindref_bc0 = texcoord; tevindref_bi0 = texmap; } + else if (index == 1) { tevindref_bc1 = texcoord; tevindref_bi1 = texmap; } + else if (index == 2) { tevindref_bc3 = texcoord; tevindref_bi2 = texmap; } + else if (index == 3) { tevindref_bc4 = texcoord; tevindref_bi4 = texmap; } + } + inline void SetTevindrefTexmap(int index, u32 texmap) { - u32 bi0 : 3; // indirect tex stage 0 ntexmap - u32 bc0 : 3; // indirect tex stage 0 ntexcoord - u32 bi1 : 3; - u32 bc1 : 3; - u32 bi2 : 3; - u32 bc3 : 3; - u32 bi4 : 3; - u32 bc4 : 3; - inline void SetValues(int index, u32 texcoord, u32 texmap) - { - if (index == 0) { bc0 = texcoord; bi0 = texmap; } - else if (index == 1) { bc1 = texcoord; bi1 = texmap; } - else if (index == 2) { bc3 = texcoord; bi2 = texmap; } - else if (index == 3) { bc4 = texcoord; bi4 = texmap; } - } - inline void SetTexmap(int index, u32 texmap) - { - if (index == 0) { bi0 = texmap; } - else if (index == 1) { bi1 = texmap; } - else if (index == 2) { bi2 = texmap; } - else if (index == 3) { bi4 = texmap; } - } - } tevindref; + if (index == 0) { tevindref_bi0 = texmap; } + else if (index == 1) { tevindref_bi1 = texmap; } + else if (index == 2) { tevindref_bi2 = texmap; } + else if (index == 3) { tevindref_bi4 = texmap; } + } - u32 tevorders_n_texcoord1 : 24; // 8 x 3 bit - u32 tevorders_n_texcoord2 : 24; // 8 x 3 bit - struct + u64 tevorders_n_texcoord : 48; // 16 x 3 bits + + u64 tevind_n_sw : 48; // 16 x 3 bits + u64 tevind_n_tw : 48; // 16 x 3 bits + u32 tevind_n_fb_addprev : 16; // 16 x 1 bit + u32 tevind_n_bs : 32; // 16 x 2 bits + u32 tevind_n_fmt : 32; // 16 x 2 bits + u32 tevind_n_bt : 32; // 16 x 2 bits + u64 tevind_n_bias : 48; // 16 x 3 bits + u64 tevind_n_mid : 64; // 16 x 4 bits + + // NOTE: These assume that the affected bits are zero before calling + void Set_tevind_sw(int index, u64 val) { - u32 sw1 : 24; // 8 x 3 bit - u32 sw2 : 24; // 8 x 3 bit - u32 tw1 : 24; // 8 x 3 bit - u32 tw2 : 24; // 8 x 3 bit - u32 fb_addprev : 16; // 16 x 1 bit - u32 bs : 32; // 16 x 2 bit - u32 fmt : 32; // 16 x 2 bit - u32 bt : 32; // 16 x 2 bit - u32 bias1 : 24; // 8 x 3 bit - u32 bias2 : 24; // 8 x 3 bit - u32 mid1 : 32; // 8 x 4 bit - u32 mid2 : 32; // 8 x 4 bit - - // NOTE: These assume that the affected bits are zero before calling - void Set_sw(int index, u32 val) - { - if (index < 8) sw1 |= val << (3*index); - else sw2 |= val << (3*index - 24); - } - void Set_tw(int index, u32 val) - { - if (index < 8) tw1 |= val << (3*index); - else tw2 |= val << (3*index - 24); - } - void Set_bias(int index, u32 val) - { - if (index < 8) bias1 |= val << (3*index); - else bias2 |= val << (3*index - 24); - } - void Set_mid(int index, u32 val) - { - if (index < 8) mid1 |= val << (4*index); - else mid2 |= val << (4*index - 32); - } - } tevind_n; - - struct + tevind_n_sw |= val << (3*index); + } + void Set_tevind_tw(int index, u64 val) { - u32 swap1 : 2; - u32 swap2 : 2; - u32 kcsel0 : 5; - u32 kasel0 : 5; - u32 kcsel1 : 5; - u32 kasel1 : 5; - - void set_kcsel(int i, u32 value) { if (i) kcsel1 = value; else kcsel0 = value; } - void set_kasel(int i, u32 value) { if( i) kasel1 = value; else kasel0 = value; } - } tevksel[8]; - - struct + tevind_n_tw |= val << (3*index); + } + void Set_tevind_bias(int index, u64 val) { - union { - struct //abc=8bit,d=10bit - { - u32 d : 4; // TEVSELCC_X - u32 c : 4; // TEVSELCC_X - u32 b : 4; // TEVSELCC_X - u32 a : 4; // TEVSELCC_X - - u32 bias : 2; - u32 op : 1; - u32 clamp : 1; - - u32 shift : 2; - u32 dest : 2; //1,2,3 - }; - u32 hex : 24; - } colorC; - union { - struct - { - u32 rswap : 2; - u32 tswap : 2; - u32 d : 3; // TEVSELCA_ - u32 c : 3; // TEVSELCA_ - u32 b : 3; // TEVSELCA_ - u32 a : 3; // TEVSELCA_ - - u32 bias : 2; //GXTevBias - u32 op : 1; - u32 clamp : 1; - - u32 shift : 2; - u32 dest : 2; //1,2,3 - }; - u32 hex : 24; - } alphaC; - } combiners[16]; - struct + tevind_n_bias |= val << (3*index); + } + void Set_tevind_mid(int index, u64 val) { - u32 comp0 : 3; - u32 comp1 : 3; - u32 logic : 2; - u32 use_zcomploc_hack : 1; - } alpha_test; + tevind_n_mid |= val << (4*index); + } - union { - struct - { - u32 proj : 1; // 0 - perspective, 1 - orthographic - u32 fsel : 3; // 0 - off, 2 - linear, 4 - exp, 5 - exp2, 6 - backward exp, 7 - backward exp2 - u32 RangeBaseEnabled : 1; - }; - u32 hex : 4; - } fog; + u32 tevksel_n_swap1 : 16; // 8x2 bits + u32 tevksel_n_swap2 : 16; // 8x2 bits + u64 tevksel_n_kcsel0 : 40; // 8x5 bits + u64 tevksel_n_kasel0 : 40; // 8x5 bits + u64 tevksel_n_kcsel1 : 40; // 8x5 bits + u64 tevksel_n_kasel1 : 40; // 8x5 bits + void set_tevksel_kcsel(int index, int i, u32 value) { if (i) tevksel_n_kcsel1 |= value << (5*index); else tevksel_n_kcsel0 |= value << (5*index); } + void set_tevksel_kasel(int index, int i, u32 value) { if( i) tevksel_n_kasel1 |= value << (5*index); else tevksel_n_kasel0 |= value << (5*index); } + + u64 cc_n_d : 64; // 16x4 bits + u64 cc_n_c : 64; // 16x4 bits + u64 cc_n_b : 64; // 16x4 bits + u64 cc_n_a : 64; // 16x4 bits + u32 cc_n_bias : 32; // 16x2 bits + u32 cc_n_op : 16; // 16x1 bit + u32 cc_n_clamp : 16; // 16x1 bit + u32 cc_n_shift : 32; // 16x2 bits + u32 cc_n_dest : 32; // 16x2 bits + + u32 ac_n_rswap : 32; // 16x2 bits + u32 ac_n_tswap : 32; // 16x2 bits + u64 ac_n_d : 48; // 16x3 bits + u64 ac_n_c : 48; // 16x3 bits + u64 ac_n_b : 48; // 16x3 bits + u64 ac_n_a : 48; // 16x3 bits + u32 ac_n_bias : 32; // 16x2 bits + u32 ac_n_op : 16; // 16x1 bit + u32 ac_n_clamp : 16; // 16x1 bit + u32 ac_n_shift : 32; // 16x2 bits + u32 ac_n_dest : 32; // 16x2 bits + + u32 alpha_test_comp0 : 3; + u32 alpha_test_comp1 : 3; + u32 alpha_test_logic : 2; + u32 alpha_test_use_zcomploc_hack : 1; + + u32 fog_proj : 1; + u32 fog_fsel : 3; + u32 fog_RangeBaseEnabled : 1; u32 ztex_op : 2; @@ -221,6 +178,7 @@ struct pixel_shader_uid_data LightingUidData lighting; }; +//#pragma pack() typedef ShaderUid PixelShaderUid; typedef ShaderCode PixelShaderCode; // TODO: Obsolete diff --git a/Source/Core/VideoCommon/Src/ShaderGenCommon.h b/Source/Core/VideoCommon/Src/ShaderGenCommon.h index ebf2ae1b87..b8820e2e98 100644 --- a/Source/Core/VideoCommon/Src/ShaderGenCommon.h +++ b/Source/Core/VideoCommon/Src/ShaderGenCommon.h @@ -199,22 +199,21 @@ static void DeclareUniform(T& object, API_TYPE api_type, bool using_ubos, const object.Write(";\n"); } +#pragma pack(4) /** * Common uid data used for shader generators that use lighting calculations. * Expected to be stored as a member called "lighting". */ struct LightingUidData { - struct - { - u32 matsource : 1; - u32 enablelighting : 1; - u32 ambsource : 1; - u32 diffusefunc : 2; - u32 attnfunc : 2; - u32 light_mask : 8; - } lit_chans[4]; + u32 matsource : 4; // 4x1 bit + u32 enablelighting : 4; // 4x1 bit + u32 ambsource : 4; // 4x1 bit + u32 diffusefunc : 8; // 4x2 bits + u32 attnfunc : 8; // 4x2 bits + u32 light_mask : 32; // 4x8 bits }; +#pragma pack() /** * Checks if there has been