summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
-rw-r--r--src/shader_recompiler/backend/glsl/emit_context.cpp2
-rw-r--r--src/shader_recompiler/backend/glsl/emit_glsl.cpp2
-rw-r--r--src/shader_recompiler/backend/glsl/emit_glsl_image.cpp46
-rw-r--r--src/shader_recompiler/profile.h2
-rw-r--r--src/video_core/renderer_opengl/gl_device.cpp1
-rw-r--r--src/video_core/renderer_opengl/gl_device.h5
-rw-r--r--src/video_core/renderer_opengl/gl_shader_cache.cpp2
7 files changed, 47 insertions, 13 deletions
diff --git a/src/shader_recompiler/backend/glsl/emit_context.cpp b/src/shader_recompiler/backend/glsl/emit_context.cpp
index f0e9dffc2..d0880bdcb 100644
--- a/src/shader_recompiler/backend/glsl/emit_context.cpp
+++ b/src/shader_recompiler/backend/glsl/emit_context.cpp
@@ -384,7 +384,7 @@ void EmitContext::SetupExtensions() {
384 profile.support_viewport_index_layer_non_geometry && stage != Stage::Geometry) { 384 profile.support_viewport_index_layer_non_geometry && stage != Stage::Geometry) {
385 header += "#extension GL_ARB_shader_viewport_layer_array : enable\n"; 385 header += "#extension GL_ARB_shader_viewport_layer_array : enable\n";
386 } 386 }
387 if (info.uses_sparse_residency) { 387 if (info.uses_sparse_residency && profile.support_gl_sparse_textures) {
388 header += "#extension GL_ARB_sparse_texture2 : enable\n"; 388 header += "#extension GL_ARB_sparse_texture2 : enable\n";
389 } 389 }
390 if (info.stores_viewport_mask && profile.support_viewport_mask) { 390 if (info.stores_viewport_mask && profile.support_viewport_mask) {
diff --git a/src/shader_recompiler/backend/glsl/emit_glsl.cpp b/src/shader_recompiler/backend/glsl/emit_glsl.cpp
index d76b63b2d..6d64913bb 100644
--- a/src/shader_recompiler/backend/glsl/emit_glsl.cpp
+++ b/src/shader_recompiler/backend/glsl/emit_glsl.cpp
@@ -215,7 +215,7 @@ std::string EmitGLSL(const Profile& profile, const RuntimeInfo& runtime_info, IR
215 ctx.header += 215 ctx.header +=
216 fmt::format("shared uint smem[{}];", Common::AlignUp(program.shared_memory_size, 4)); 216 fmt::format("shared uint smem[{}];", Common::AlignUp(program.shared_memory_size, 4));
217 } 217 }
218 ctx.header += "\nvoid main(){\n"; 218 ctx.header += "void main(){\n";
219 if (program.stage == Stage::VertexA || program.stage == Stage::VertexB) { 219 if (program.stage == Stage::VertexA || program.stage == Stage::VertexB) {
220 ctx.header += "gl_Position = vec4(0.0f, 0.0f, 0.0f, 1.0f);"; 220 ctx.header += "gl_Position = vec4(0.0f, 0.0f, 0.0f, 1.0f);";
221 // TODO: Properly resolve attribute issues 221 // TODO: Properly resolve attribute issues
diff --git a/src/shader_recompiler/backend/glsl/emit_glsl_image.cpp b/src/shader_recompiler/backend/glsl/emit_glsl_image.cpp
index 51181d1c1..c6b3df9c9 100644
--- a/src/shader_recompiler/backend/glsl/emit_glsl_image.cpp
+++ b/src/shader_recompiler/backend/glsl/emit_glsl_image.cpp
@@ -94,7 +94,11 @@ std::string GetOffsetVec(EmitContext& ctx, const IR::Value& offset) {
94 break; 94 break;
95 } 95 }
96 } 96 }
97 const auto offset_str{ctx.var_alloc.Consume(offset)}; 97 const bool has_var_aoffi{ctx.profile.support_gl_variable_aoffi};
98 if (!has_var_aoffi) {
99 // LOG_WARNING("Device does not support variable texture offsets, STUBBING");
100 }
101 const auto offset_str{has_var_aoffi ? ctx.var_alloc.Consume(offset) : "0"};
98 switch (offset.Type()) { 102 switch (offset.Type()) {
99 case IR::Type::U32: 103 case IR::Type::U32:
100 return fmt::format("int({})", offset_str); 104 return fmt::format("int({})", offset_str);
@@ -146,7 +150,12 @@ void EmitImageSampleImplicitLod(EmitContext& ctx, IR::Inst& inst, const IR::Valu
146 const auto bias{info.has_bias ? fmt::format(",{}", bias_lc) : ""}; 150 const auto bias{info.has_bias ? fmt::format(",{}", bias_lc) : ""};
147 const auto texel{ctx.var_alloc.Define(inst, GlslVarType::F32x4)}; 151 const auto texel{ctx.var_alloc.Define(inst, GlslVarType::F32x4)};
148 const auto sparse_inst{PrepareSparse(inst)}; 152 const auto sparse_inst{PrepareSparse(inst)};
149 if (!sparse_inst) { 153 const bool supports_sparse{ctx.profile.support_gl_sparse_textures};
154 if (sparse_inst && !supports_sparse) {
155 // LOG_WARNING(..., "Device does not support sparse texture queries. STUBBING");
156 ctx.AddU1("{}=true;", *sparse_inst);
157 }
158 if (!sparse_inst || !supports_sparse) {
150 if (!offset.IsEmpty()) { 159 if (!offset.IsEmpty()) {
151 const auto offset_str{GetOffsetVec(ctx, offset)}; 160 const auto offset_str{GetOffsetVec(ctx, offset)};
152 if (ctx.stage == Stage::Fragment) { 161 if (ctx.stage == Stage::Fragment) {
@@ -163,7 +172,6 @@ void EmitImageSampleImplicitLod(EmitContext& ctx, IR::Inst& inst, const IR::Valu
163 } 172 }
164 return; 173 return;
165 } 174 }
166 // TODO: Query sparseTexels extension support
167 if (!offset.IsEmpty()) { 175 if (!offset.IsEmpty()) {
168 ctx.AddU1("{}=sparseTexelsResidentARB(sparseTextureOffsetARB({},{},{},{}{}));", 176 ctx.AddU1("{}=sparseTexelsResidentARB(sparseTextureOffsetARB({},{},{},{}{}));",
169 *sparse_inst, texture, coords, GetOffsetVec(ctx, offset), texel, bias); 177 *sparse_inst, texture, coords, GetOffsetVec(ctx, offset), texel, bias);
@@ -186,7 +194,12 @@ void EmitImageSampleExplicitLod(EmitContext& ctx, IR::Inst& inst, const IR::Valu
186 const auto texture{Texture(ctx, info, index)}; 194 const auto texture{Texture(ctx, info, index)};
187 const auto texel{ctx.var_alloc.Define(inst, GlslVarType::F32x4)}; 195 const auto texel{ctx.var_alloc.Define(inst, GlslVarType::F32x4)};
188 const auto sparse_inst{PrepareSparse(inst)}; 196 const auto sparse_inst{PrepareSparse(inst)};
189 if (!sparse_inst) { 197 const bool supports_sparse{ctx.profile.support_gl_sparse_textures};
198 if (sparse_inst && !supports_sparse) {
199 // LOG_WARNING(..., "Device does not support sparse texture queries. STUBBING");
200 ctx.AddU1("{}=true;", *sparse_inst);
201 }
202 if (!sparse_inst || !supports_sparse) {
190 if (!offset.IsEmpty()) { 203 if (!offset.IsEmpty()) {
191 ctx.Add("{}=textureLodOffset({},{},{},{});", texel, texture, coords, lod_lc, 204 ctx.Add("{}=textureLodOffset({},{},{},{});", texel, texture, coords, lod_lc,
192 GetOffsetVec(ctx, offset)); 205 GetOffsetVec(ctx, offset));
@@ -195,7 +208,6 @@ void EmitImageSampleExplicitLod(EmitContext& ctx, IR::Inst& inst, const IR::Valu
195 } 208 }
196 return; 209 return;
197 } 210 }
198 // TODO: Query sparseTexels extension support
199 if (!offset.IsEmpty()) { 211 if (!offset.IsEmpty()) {
200 ctx.AddU1("{}=sparseTexelsResidentARB(sparseTexelFetchOffsetARB({},{},int({}),{},{}));", 212 ctx.AddU1("{}=sparseTexelsResidentARB(sparseTexelFetchOffsetARB({},{},int({}),{},{}));",
201 *sparse_inst, texture, CastToIntVec(coords, info), lod_lc, 213 *sparse_inst, texture, CastToIntVec(coords, info), lod_lc,
@@ -315,7 +327,12 @@ void EmitImageGather(EmitContext& ctx, IR::Inst& inst, const IR::Value& index,
315 const auto texture{Texture(ctx, info, index)}; 327 const auto texture{Texture(ctx, info, index)};
316 const auto texel{ctx.var_alloc.Define(inst, GlslVarType::F32x4)}; 328 const auto texel{ctx.var_alloc.Define(inst, GlslVarType::F32x4)};
317 const auto sparse_inst{PrepareSparse(inst)}; 329 const auto sparse_inst{PrepareSparse(inst)};
318 if (!sparse_inst) { 330 const bool supports_sparse{ctx.profile.support_gl_sparse_textures};
331 if (sparse_inst && !supports_sparse) {
332 // LOG_WARNING(..., "Device does not support sparse texture queries. STUBBING");
333 ctx.AddU1("{}=true;", *sparse_inst);
334 }
335 if (!sparse_inst || !supports_sparse) {
319 if (offset.IsEmpty()) { 336 if (offset.IsEmpty()) {
320 ctx.Add("{}=textureGather({},{},int({}));", texel, texture, coords, 337 ctx.Add("{}=textureGather({},{},int({}));", texel, texture, coords,
321 info.gather_component); 338 info.gather_component);
@@ -332,7 +349,6 @@ void EmitImageGather(EmitContext& ctx, IR::Inst& inst, const IR::Value& index,
332 info.gather_component); 349 info.gather_component);
333 return; 350 return;
334 } 351 }
335 // TODO: Query sparseTexels extension support
336 if (offset.IsEmpty()) { 352 if (offset.IsEmpty()) {
337 ctx.AddU1("{}=sparseTexelsResidentARB(sparseTextureGatherARB({},{},{},int({})));", 353 ctx.AddU1("{}=sparseTexelsResidentARB(sparseTextureGatherARB({},{},{},int({})));",
338 *sparse_inst, texture, coords, texel, info.gather_component); 354 *sparse_inst, texture, coords, texel, info.gather_component);
@@ -358,7 +374,12 @@ void EmitImageGatherDref(EmitContext& ctx, IR::Inst& inst, const IR::Value& inde
358 const auto texture{Texture(ctx, info, index)}; 374 const auto texture{Texture(ctx, info, index)};
359 const auto texel{ctx.var_alloc.Define(inst, GlslVarType::F32x4)}; 375 const auto texel{ctx.var_alloc.Define(inst, GlslVarType::F32x4)};
360 const auto sparse_inst{PrepareSparse(inst)}; 376 const auto sparse_inst{PrepareSparse(inst)};
361 if (!sparse_inst) { 377 const bool supports_sparse{ctx.profile.support_gl_sparse_textures};
378 if (sparse_inst && !supports_sparse) {
379 // LOG_WARNING(..., "Device does not support sparse texture queries. STUBBING");
380 ctx.AddU1("{}=true;", *sparse_inst);
381 }
382 if (!sparse_inst || !supports_sparse) {
362 if (offset.IsEmpty()) { 383 if (offset.IsEmpty()) {
363 ctx.Add("{}=textureGather({},{},{});", texel, texture, coords, dref); 384 ctx.Add("{}=textureGather({},{},{});", texel, texture, coords, dref);
364 return; 385 return;
@@ -373,7 +394,6 @@ void EmitImageGatherDref(EmitContext& ctx, IR::Inst& inst, const IR::Value& inde
373 ctx.Add("{}=textureGatherOffsets({},{},{},{});", texel, texture, coords, dref, offsets); 394 ctx.Add("{}=textureGatherOffsets({},{},{},{});", texel, texture, coords, dref, offsets);
374 return; 395 return;
375 } 396 }
376 // TODO: Query sparseTexels extension support
377 if (offset.IsEmpty()) { 397 if (offset.IsEmpty()) {
378 ctx.AddU1("{}=sparseTexelsResidentARB(sparseTextureGatherARB({},{},{},{}));", *sparse_inst, 398 ctx.AddU1("{}=sparseTexelsResidentARB(sparseTextureGatherARB({},{},{},{}));", *sparse_inst,
379 texture, coords, dref, texel); 399 texture, coords, dref, texel);
@@ -404,7 +424,12 @@ void EmitImageFetch(EmitContext& ctx, IR::Inst& inst, const IR::Value& index,
404 const auto texture{Texture(ctx, info, index)}; 424 const auto texture{Texture(ctx, info, index)};
405 const auto sparse_inst{PrepareSparse(inst)}; 425 const auto sparse_inst{PrepareSparse(inst)};
406 const auto texel{ctx.var_alloc.Define(inst, GlslVarType::F32x4)}; 426 const auto texel{ctx.var_alloc.Define(inst, GlslVarType::F32x4)};
407 if (!sparse_inst) { 427 const bool supports_sparse{ctx.profile.support_gl_sparse_textures};
428 if (sparse_inst && !supports_sparse) {
429 // LOG_WARNING(..., "Device does not support sparse texture queries. STUBBING");
430 ctx.AddU1("{}=true;", *sparse_inst);
431 }
432 if (!sparse_inst || !supports_sparse) {
408 if (!offset.empty()) { 433 if (!offset.empty()) {
409 ctx.Add("{}=texelFetchOffset({},{},int({}),{});", texel, texture, 434 ctx.Add("{}=texelFetchOffset({},{},int({}),{});", texel, texture,
410 CoordsCastToInt(coords, info), lod, CoordsCastToInt(offset, info)); 435 CoordsCastToInt(coords, info), lod, CoordsCastToInt(offset, info));
@@ -418,7 +443,6 @@ void EmitImageFetch(EmitContext& ctx, IR::Inst& inst, const IR::Value& index,
418 } 443 }
419 return; 444 return;
420 } 445 }
421 // TODO: Query sparseTexels extension support
422 if (!offset.empty()) { 446 if (!offset.empty()) {
423 ctx.AddU1("{}=sparseTexelsResidentARB(sparseTexelFetchOffsetARB({},{},int({}),{},{}));", 447 ctx.AddU1("{}=sparseTexelsResidentARB(sparseTexelFetchOffsetARB({},{},int({}),{},{}));",
424 *sparse_inst, texture, CastToIntVec(coords, info), lod, 448 *sparse_inst, texture, CastToIntVec(coords, info), lod,
diff --git a/src/shader_recompiler/profile.h b/src/shader_recompiler/profile.h
index 246995190..236c79a0a 100644
--- a/src/shader_recompiler/profile.h
+++ b/src/shader_recompiler/profile.h
@@ -87,6 +87,8 @@ struct Profile {
87 bool support_gl_amd_gpu_shader_half_float{}; 87 bool support_gl_amd_gpu_shader_half_float{};
88 bool support_gl_texture_shadow_lod{}; 88 bool support_gl_texture_shadow_lod{};
89 bool support_gl_warp_intrinsics{}; 89 bool support_gl_warp_intrinsics{};
90 bool support_gl_variable_aoffi{};
91 bool support_gl_sparse_textures{};
90 92
91 bool warp_size_potentially_larger_than_guest{}; 93 bool warp_size_potentially_larger_than_guest{};
92 94
diff --git a/src/video_core/renderer_opengl/gl_device.cpp b/src/video_core/renderer_opengl/gl_device.cpp
index 20ea42cff..bf08a6d93 100644
--- a/src/video_core/renderer_opengl/gl_device.cpp
+++ b/src/video_core/renderer_opengl/gl_device.cpp
@@ -160,6 +160,7 @@ Device::Device() {
160 has_depth_buffer_float = HasExtension(extensions, "GL_NV_depth_buffer_float"); 160 has_depth_buffer_float = HasExtension(extensions, "GL_NV_depth_buffer_float");
161 has_nv_gpu_shader_5 = GLAD_GL_NV_gpu_shader5; 161 has_nv_gpu_shader_5 = GLAD_GL_NV_gpu_shader5;
162 has_amd_shader_half_float = GLAD_GL_AMD_gpu_shader_half_float; 162 has_amd_shader_half_float = GLAD_GL_AMD_gpu_shader_half_float;
163 has_sparse_texture_2 = GLAD_GL_ARB_sparse_texture2;
163 warp_size_potentially_larger_than_guest = !is_nvidia && !is_intel; 164 warp_size_potentially_larger_than_guest = !is_nvidia && !is_intel;
164 165
165 // At the moment of writing this, only Nvidia's driver optimizes BufferSubData on exclusive 166 // At the moment of writing this, only Nvidia's driver optimizes BufferSubData on exclusive
diff --git a/src/video_core/renderer_opengl/gl_device.h b/src/video_core/renderer_opengl/gl_device.h
index ff0ff2b08..0b59c9df0 100644
--- a/src/video_core/renderer_opengl/gl_device.h
+++ b/src/video_core/renderer_opengl/gl_device.h
@@ -128,6 +128,10 @@ public:
128 return has_amd_shader_half_float; 128 return has_amd_shader_half_float;
129 } 129 }
130 130
131 bool HasSparseTexture2() const {
132 return has_sparse_texture_2;
133 }
134
131 bool IsWarpSizePotentiallyLargerThanGuest() const { 135 bool IsWarpSizePotentiallyLargerThanGuest() const {
132 return warp_size_potentially_larger_than_guest; 136 return warp_size_potentially_larger_than_guest;
133 } 137 }
@@ -165,6 +169,7 @@ private:
165 bool has_depth_buffer_float{}; 169 bool has_depth_buffer_float{};
166 bool has_nv_gpu_shader_5{}; 170 bool has_nv_gpu_shader_5{};
167 bool has_amd_shader_half_float{}; 171 bool has_amd_shader_half_float{};
172 bool has_sparse_texture_2{};
168 bool warp_size_potentially_larger_than_guest{}; 173 bool warp_size_potentially_larger_than_guest{};
169}; 174};
170 175
diff --git a/src/video_core/renderer_opengl/gl_shader_cache.cpp b/src/video_core/renderer_opengl/gl_shader_cache.cpp
index 884739aec..3d59d34d7 100644
--- a/src/video_core/renderer_opengl/gl_shader_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_cache.cpp
@@ -193,6 +193,8 @@ ShaderCache::ShaderCache(RasterizerOpenGL& rasterizer_, Core::Frontend::EmuWindo
193 .support_gl_amd_gpu_shader_half_float = device.HasAmdShaderHalfFloat(), 193 .support_gl_amd_gpu_shader_half_float = device.HasAmdShaderHalfFloat(),
194 .support_gl_texture_shadow_lod = device.HasTextureShadowLod(), 194 .support_gl_texture_shadow_lod = device.HasTextureShadowLod(),
195 .support_gl_warp_intrinsics = false, 195 .support_gl_warp_intrinsics = false,
196 .support_gl_variable_aoffi = device.HasVariableAoffi(),
197 .support_gl_sparse_textures = device.HasSparseTexture2(),
196 198
197 .warp_size_potentially_larger_than_guest = device.IsWarpSizePotentiallyLargerThanGuest(), 199 .warp_size_potentially_larger_than_guest = device.IsWarpSizePotentiallyLargerThanGuest(),
198 200