summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorReinUsesLisp <reinuseslisp@airmail.cc>2021-02-17 04:59:28 +0100
committerameerj <52414509+ameerj@users.noreply.github.com>2021-07-23 03:51:22 +0200
commit85cce78583bc2232428a8fb39e43182877c8d5ad (patch)
tree308f4ef2d145652e08dff1da31c72c2f00dad2e1
parentshader: Remove old shader management (diff)
downloadyuzu-85cce78583bc2232428a8fb39e43182877c8d5ad.tar
yuzu-85cce78583bc2232428a8fb39e43182877c8d5ad.tar.gz
yuzu-85cce78583bc2232428a8fb39e43182877c8d5ad.tar.bz2
yuzu-85cce78583bc2232428a8fb39e43182877c8d5ad.tar.lz
yuzu-85cce78583bc2232428a8fb39e43182877c8d5ad.tar.xz
yuzu-85cce78583bc2232428a8fb39e43182877c8d5ad.tar.zst
yuzu-85cce78583bc2232428a8fb39e43182877c8d5ad.zip
Diffstat (limited to '')
-rw-r--r--src/shader_recompiler/CMakeLists.txt13
-rw-r--r--src/shader_recompiler/backend/spirv/emit_context.cpp2
-rw-r--r--src/shader_recompiler/backend/spirv/emit_spirv.cpp117
-rw-r--r--src/shader_recompiler/backend/spirv/emit_spirv.h419
-rw-r--r--src/shader_recompiler/backend/spirv/emit_spirv_bitwise_conversion.cpp24
-rw-r--r--src/shader_recompiler/backend/spirv/emit_spirv_composite.cpp48
-rw-r--r--src/shader_recompiler/backend/spirv/emit_spirv_context_get_set.cpp42
-rw-r--r--src/shader_recompiler/backend/spirv/emit_spirv_control_flow.cpp10
-rw-r--r--src/shader_recompiler/backend/spirv/emit_spirv_floating_point.cpp92
-rw-r--r--src/shader_recompiler/backend/spirv/emit_spirv_integer.cpp60
-rw-r--r--src/shader_recompiler/backend/spirv/emit_spirv_logical.cpp40
-rw-r--r--src/shader_recompiler/backend/spirv/emit_spirv_memory.cpp56
-rw-r--r--src/shader_recompiler/backend/spirv/emit_spirv_select.cpp8
-rw-r--r--src/shader_recompiler/backend/spirv/emit_spirv_undefined.cpp10
-rw-r--r--src/shader_recompiler/environment.h6
-rw-r--r--src/shader_recompiler/file_environment.cpp6
-rw-r--r--src/shader_recompiler/file_environment.h4
-rw-r--r--src/shader_recompiler/frontend/ir/basic_block.cpp2
-rw-r--r--src/shader_recompiler/frontend/ir/post_order.cpp2
-rw-r--r--src/shader_recompiler/frontend/maxwell/program.cpp2
-rw-r--r--src/shader_recompiler/frontend/maxwell/translate/impl/impl.cpp8
-rw-r--r--src/shader_recompiler/frontend/maxwell/translate/impl/impl.h1
-rw-r--r--src/shader_recompiler/frontend/maxwell/translate/impl/move_register.cpp35
-rw-r--r--src/shader_recompiler/frontend/maxwell/translate/impl/not_implemented.cpp4
-rw-r--r--src/shader_recompiler/main.cpp2
-rw-r--r--src/shader_recompiler/profile.h13
-rw-r--r--src/shader_recompiler/recompiler.cpp27
-rw-r--r--src/shader_recompiler/recompiler.h18
-rw-r--r--src/video_core/CMakeLists.txt6
-rw-r--r--src/video_core/engines/kepler_compute.h1
-rw-r--r--src/video_core/engines/shader_bytecode.h2298
-rw-r--r--src/video_core/engines/shader_header.h158
-rw-r--r--src/video_core/renderer_vulkan/vk_compute_pipeline.cpp140
-rw-r--r--src/video_core/renderer_vulkan/vk_compute_pipeline.h43
-rw-r--r--src/video_core/renderer_vulkan/vk_descriptor_pool.cpp6
-rw-r--r--src/video_core/renderer_vulkan/vk_descriptor_pool.h10
-rw-r--r--src/video_core/renderer_vulkan/vk_pipeline.h36
-rw-r--r--src/video_core/renderer_vulkan/vk_pipeline_cache.cpp190
-rw-r--r--src/video_core/renderer_vulkan/vk_pipeline_cache.h30
-rw-r--r--src/video_core/renderer_vulkan/vk_rasterizer.cpp23
-rw-r--r--src/video_core/renderer_vulkan/vk_rasterizer.h3
-rw-r--r--src/video_core/renderer_vulkan/vk_resource_pool.cpp12
-rw-r--r--src/video_core/renderer_vulkan/vk_resource_pool.h12
43 files changed, 1003 insertions, 3036 deletions
diff --git a/src/shader_recompiler/CMakeLists.txt b/src/shader_recompiler/CMakeLists.txt
index 84be94a8d..b56bdd3d9 100644
--- a/src/shader_recompiler/CMakeLists.txt
+++ b/src/shader_recompiler/CMakeLists.txt
@@ -1,4 +1,4 @@
-add_executable(shader_recompiler
+add_library(shader_recompiler STATIC
backend/spirv/emit_context.cpp
backend/spirv/emit_context.h
backend/spirv/emit_spirv.cpp
@@ -85,13 +85,19 @@ add_executable(shader_recompiler
ir_opt/passes.h
ir_opt/ssa_rewrite_pass.cpp
ir_opt/verification_pass.cpp
- main.cpp
object_pool.h
+ profile.h
+ recompiler.cpp
+ recompiler.h
shader_info.h
)
-target_include_directories(video_core PRIVATE sirit)
+target_include_directories(shader_recompiler PRIVATE sirit)
target_link_libraries(shader_recompiler PRIVATE fmt::fmt sirit)
+target_link_libraries(shader_recompiler INTERFACE fmt::fmt sirit)
+
+add_executable(shader_util main.cpp)
+target_link_libraries(shader_util PRIVATE shader_recompiler)
if (MSVC)
target_compile_options(shader_recompiler PRIVATE
@@ -121,3 +127,4 @@ else()
endif()
create_target_directory_groups(shader_recompiler)
+create_target_directory_groups(shader_util)
diff --git a/src/shader_recompiler/backend/spirv/emit_context.cpp b/src/shader_recompiler/backend/spirv/emit_context.cpp
index 1c985aff8..770067d98 100644
--- a/src/shader_recompiler/backend/spirv/emit_context.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_context.cpp
@@ -115,6 +115,7 @@ void EmitContext::DefineConstantBuffers(const Info& info) {
for (const Info::ConstantBufferDescriptor& desc : info.constant_buffer_descriptors) {
const Id id{AddGlobalVariable(uniform_type, spv::StorageClass::Uniform)};
Decorate(id, spv::Decoration::Binding, binding);
+ Decorate(id, spv::Decoration::DescriptorSet, 0U);
Name(id, fmt::format("c{}", desc.index));
std::fill_n(cbufs.data() + desc.index, desc.count, id);
binding += desc.count;
@@ -143,6 +144,7 @@ void EmitContext::DefineStorageBuffers(const Info& info) {
for (const Info::StorageBufferDescriptor& desc : info.storage_buffers_descriptors) {
const Id id{AddGlobalVariable(storage_type, spv::StorageClass::StorageBuffer)};
Decorate(id, spv::Decoration::Binding, binding);
+ Decorate(id, spv::Decoration::DescriptorSet, 0U);
Name(id, fmt::format("ssbo{}", binding));
std::fill_n(ssbos.data() + binding, desc.count, id);
binding += desc.count;
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv.cpp b/src/shader_recompiler/backend/spirv/emit_spirv.cpp
index 55018332e..d59718435 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv.cpp
@@ -2,8 +2,11 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
-#include <numeric>
+#include <span>
+#include <tuple>
#include <type_traits>
+#include <utility>
+#include <vector>
#include "shader_recompiler/backend/spirv/emit_spirv.h"
#include "shader_recompiler/frontend/ir/basic_block.h"
@@ -14,10 +17,10 @@
namespace Shader::Backend::SPIRV {
namespace {
template <class Func>
-struct FuncTraits : FuncTraits<decltype(&Func::operator())> {};
+struct FuncTraits : FuncTraits<Func> {};
-template <class ClassType, class ReturnType_, class... Args>
-struct FuncTraits<ReturnType_ (ClassType::*)(Args...)> {
+template <class ReturnType_, class... Args>
+struct FuncTraits<ReturnType_ (*)(Args...)> {
using ReturnType = ReturnType_;
static constexpr size_t NUM_ARGS = sizeof...(Args);
@@ -26,15 +29,15 @@ struct FuncTraits<ReturnType_ (ClassType::*)(Args...)> {
using ArgType = std::tuple_element_t<I, std::tuple<Args...>>;
};
-template <auto method, typename... Args>
-void SetDefinition(EmitSPIRV& emit, EmitContext& ctx, IR::Inst* inst, Args... args) {
+template <auto func, typename... Args>
+void SetDefinition(EmitContext& ctx, IR::Inst* inst, Args... args) {
const Id forward_id{inst->Definition<Id>()};
const bool has_forward_id{Sirit::ValidId(forward_id)};
Id current_id{};
if (has_forward_id) {
current_id = ctx.ExchangeCurrentId(forward_id);
}
- const Id new_id{(emit.*method)(ctx, std::forward<Args>(args)...)};
+ const Id new_id{func(ctx, std::forward<Args>(args)...)};
if (has_forward_id) {
ctx.ExchangeCurrentId(current_id);
} else {
@@ -55,42 +58,62 @@ ArgType Arg(EmitContext& ctx, const IR::Value& arg) {
}
}
-template <auto method, bool is_first_arg_inst, size_t... I>
-void Invoke(EmitSPIRV& emit, EmitContext& ctx, IR::Inst* inst, std::index_sequence<I...>) {
- using Traits = FuncTraits<decltype(method)>;
+template <auto func, bool is_first_arg_inst, size_t... I>
+void Invoke(EmitContext& ctx, IR::Inst* inst, std::index_sequence<I...>) {
+ using Traits = FuncTraits<decltype(func)>;
if constexpr (std::is_same_v<Traits::ReturnType, Id>) {
if constexpr (is_first_arg_inst) {
- SetDefinition<method>(emit, ctx, inst, inst,
- Arg<Traits::ArgType<I + 2>>(ctx, inst->Arg(I))...);
+ SetDefinition<func>(ctx, inst, inst, Arg<Traits::ArgType<I + 2>>(ctx, inst->Arg(I))...);
} else {
- SetDefinition<method>(emit, ctx, inst,
- Arg<Traits::ArgType<I + 1>>(ctx, inst->Arg(I))...);
+ SetDefinition<func>(ctx, inst, Arg<Traits::ArgType<I + 1>>(ctx, inst->Arg(I))...);
}
} else {
if constexpr (is_first_arg_inst) {
- (emit.*method)(ctx, inst, Arg<Traits::ArgType<I + 2>>(ctx, inst->Arg(I))...);
+ func(ctx, inst, Arg<Traits::ArgType<I + 2>>(ctx, inst->Arg(I))...);
} else {
- (emit.*method)(ctx, Arg<Traits::ArgType<I + 1>>(ctx, inst->Arg(I))...);
+ func(ctx, Arg<Traits::ArgType<I + 1>>(ctx, inst->Arg(I))...);
}
}
}
-template <auto method>
-void Invoke(EmitSPIRV& emit, EmitContext& ctx, IR::Inst* inst) {
- using Traits = FuncTraits<decltype(method)>;
+template <auto func>
+void Invoke(EmitContext& ctx, IR::Inst* inst) {
+ using Traits = FuncTraits<decltype(func)>;
static_assert(Traits::NUM_ARGS >= 1, "Insufficient arguments");
if constexpr (Traits::NUM_ARGS == 1) {
- Invoke<method, false>(emit, ctx, inst, std::make_index_sequence<0>{});
+ Invoke<func, false>(ctx, inst, std::make_index_sequence<0>{});
} else {
using FirstArgType = typename Traits::template ArgType<1>;
static constexpr bool is_first_arg_inst = std::is_same_v<FirstArgType, IR::Inst*>;
using Indices = std::make_index_sequence<Traits::NUM_ARGS - (is_first_arg_inst ? 2 : 1)>;
- Invoke<method, is_first_arg_inst>(emit, ctx, inst, Indices{});
+ Invoke<func, is_first_arg_inst>(ctx, inst, Indices{});
+ }
+}
+
+void EmitInst(EmitContext& ctx, IR::Inst* inst) {
+ switch (inst->Opcode()) {
+#define OPCODE(name, result_type, ...) \
+ case IR::Opcode::name: \
+ return Invoke<&Emit##name>(ctx, inst);
+#include "shader_recompiler/frontend/ir/opcodes.inc"
+#undef OPCODE
+ }
+ throw LogicError("Invalid opcode {}", inst->Opcode());
+}
+
+Id TypeId(const EmitContext& ctx, IR::Type type) {
+ switch (type) {
+ case IR::Type::U1:
+ return ctx.U1;
+ case IR::Type::U32:
+ return ctx.U32[1];
+ default:
+ throw NotImplementedException("Phi node type {}", type);
}
}
} // Anonymous namespace
-EmitSPIRV::EmitSPIRV(IR::Program& program) {
+std::vector<u32> EmitSPIRV(Environment& env, IR::Program& program) {
EmitContext ctx{program};
const Id void_function{ctx.TypeFunction(ctx.void_id)};
// FIXME: Forward declare functions (needs sirit support)
@@ -112,43 +135,17 @@ EmitSPIRV::EmitSPIRV(IR::Program& program) {
if (program.info.uses_local_invocation_id) {
interfaces.push_back(ctx.local_invocation_id);
}
-
const std::span interfaces_span(interfaces.data(), interfaces.size());
- ctx.AddEntryPoint(spv::ExecutionModel::Fragment, func, "main", interfaces_span);
- ctx.AddExecutionMode(func, spv::ExecutionMode::OriginUpperLeft);
-
- std::vector<u32> result{ctx.Assemble()};
- std::FILE* file{std::fopen("D:\\shader.spv", "wb")};
- std::fwrite(result.data(), sizeof(u32), result.size(), file);
- std::fclose(file);
- std::system("spirv-dis D:\\shader.spv") == 0 &&
- std::system("spirv-val --uniform-buffer-standard-layout D:\\shader.spv") == 0 &&
- std::system("spirv-cross -V D:\\shader.spv") == 0;
-}
+ ctx.AddEntryPoint(spv::ExecutionModel::GLCompute, func, "main", interfaces_span);
-void EmitSPIRV::EmitInst(EmitContext& ctx, IR::Inst* inst) {
- switch (inst->Opcode()) {
-#define OPCODE(name, result_type, ...) \
- case IR::Opcode::name: \
- return Invoke<&EmitSPIRV::Emit##name>(*this, ctx, inst);
-#include "shader_recompiler/frontend/ir/opcodes.inc"
-#undef OPCODE
- }
- throw LogicError("Invalid opcode {}", inst->Opcode());
-}
+ const std::array<u32, 3> workgroup_size{env.WorkgroupSize()};
+ ctx.AddExecutionMode(func, spv::ExecutionMode::LocalSize, workgroup_size[0], workgroup_size[1],
+ workgroup_size[2]);
-static Id TypeId(const EmitContext& ctx, IR::Type type) {
- switch (type) {
- case IR::Type::U1:
- return ctx.U1;
- case IR::Type::U32:
- return ctx.U32[1];
- default:
- throw NotImplementedException("Phi node type {}", type);
- }
+ return ctx.Assemble();
}
-Id EmitSPIRV::EmitPhi(EmitContext& ctx, IR::Inst* inst) {
+Id EmitPhi(EmitContext& ctx, IR::Inst* inst) {
const size_t num_args{inst->NumArgs()};
boost::container::small_vector<Id, 32> operands;
operands.reserve(num_args * 2);
@@ -178,25 +175,25 @@ Id EmitSPIRV::EmitPhi(EmitContext& ctx, IR::Inst* inst) {
return ctx.OpPhi(result_type, std::span(operands.data(), operands.size()));
}
-void EmitSPIRV::EmitVoid(EmitContext&) {}
+void EmitVoid(EmitContext&) {}
-Id EmitSPIRV::EmitIdentity(EmitContext& ctx, const IR::Value& value) {
+Id EmitIdentity(EmitContext& ctx, const IR::Value& value) {
return ctx.Def(value);
}
-void EmitSPIRV::EmitGetZeroFromOp(EmitContext&) {
+void EmitGetZeroFromOp(EmitContext&) {
throw LogicError("Unreachable instruction");
}
-void EmitSPIRV::EmitGetSignFromOp(EmitContext&) {
+void EmitGetSignFromOp(EmitContext&) {
throw LogicError("Unreachable instruction");
}
-void EmitSPIRV::EmitGetCarryFromOp(EmitContext&) {
+void EmitGetCarryFromOp(EmitContext&) {
throw LogicError("Unreachable instruction");
}
-void EmitSPIRV::EmitGetOverflowFromOp(EmitContext&) {
+void EmitGetOverflowFromOp(EmitContext&) {
throw LogicError("Unreachable instruction");
}
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv.h b/src/shader_recompiler/backend/spirv/emit_spirv.h
index 8bde82613..5813f51ff 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv.h
+++ b/src/shader_recompiler/backend/spirv/emit_spirv.h
@@ -8,223 +8,218 @@
#include "common/common_types.h"
#include "shader_recompiler/backend/spirv/emit_context.h"
+#include "shader_recompiler/environment.h"
#include "shader_recompiler/frontend/ir/microinstruction.h"
#include "shader_recompiler/frontend/ir/program.h"
namespace Shader::Backend::SPIRV {
-class EmitSPIRV {
-public:
- explicit EmitSPIRV(IR::Program& program);
+[[nodiscard]] std::vector<u32> EmitSPIRV(Environment& env, IR::Program& program);
-private:
- void EmitInst(EmitContext& ctx, IR::Inst* inst);
-
- // Microinstruction emitters
- Id EmitPhi(EmitContext& ctx, IR::Inst* inst);
- void EmitVoid(EmitContext& ctx);
- Id EmitIdentity(EmitContext& ctx, const IR::Value& value);
- void EmitBranch(EmitContext& ctx, IR::Block* label);
- void EmitBranchConditional(EmitContext& ctx, Id condition, IR::Block* true_label,
- IR::Block* false_label);
- void EmitLoopMerge(EmitContext& ctx, IR::Block* merge_label, IR::Block* continue_label);
- void EmitSelectionMerge(EmitContext& ctx, IR::Block* merge_label);
- void EmitReturn(EmitContext& ctx);
- void EmitGetRegister(EmitContext& ctx);
- void EmitSetRegister(EmitContext& ctx);
- void EmitGetPred(EmitContext& ctx);
- void EmitSetPred(EmitContext& ctx);
- void EmitSetGotoVariable(EmitContext& ctx);
- void EmitGetGotoVariable(EmitContext& ctx);
- Id EmitGetCbuf(EmitContext& ctx, const IR::Value& binding, const IR::Value& offset);
- void EmitGetAttribute(EmitContext& ctx);
- void EmitSetAttribute(EmitContext& ctx);
- void EmitGetAttributeIndexed(EmitContext& ctx);
- void EmitSetAttributeIndexed(EmitContext& ctx);
- void EmitGetZFlag(EmitContext& ctx);
- void EmitGetSFlag(EmitContext& ctx);
- void EmitGetCFlag(EmitContext& ctx);
- void EmitGetOFlag(EmitContext& ctx);
- void EmitSetZFlag(EmitContext& ctx);
- void EmitSetSFlag(EmitContext& ctx);
- void EmitSetCFlag(EmitContext& ctx);
- void EmitSetOFlag(EmitContext& ctx);
- Id EmitWorkgroupId(EmitContext& ctx);
- Id EmitLocalInvocationId(EmitContext& ctx);
- Id EmitUndefU1(EmitContext& ctx);
- Id EmitUndefU8(EmitContext& ctx);
- Id EmitUndefU16(EmitContext& ctx);
- Id EmitUndefU32(EmitContext& ctx);
- Id EmitUndefU64(EmitContext& ctx);
- void EmitLoadGlobalU8(EmitContext& ctx);
- void EmitLoadGlobalS8(EmitContext& ctx);
- void EmitLoadGlobalU16(EmitContext& ctx);
- void EmitLoadGlobalS16(EmitContext& ctx);
- void EmitLoadGlobal32(EmitContext& ctx);
- void EmitLoadGlobal64(EmitContext& ctx);
- void EmitLoadGlobal128(EmitContext& ctx);
- void EmitWriteGlobalU8(EmitContext& ctx);
- void EmitWriteGlobalS8(EmitContext& ctx);
- void EmitWriteGlobalU16(EmitContext& ctx);
- void EmitWriteGlobalS16(EmitContext& ctx);
- void EmitWriteGlobal32(EmitContext& ctx);
- void EmitWriteGlobal64(EmitContext& ctx);
- void EmitWriteGlobal128(EmitContext& ctx);
- void EmitLoadStorageU8(EmitContext& ctx);
- void EmitLoadStorageS8(EmitContext& ctx);
- void EmitLoadStorageU16(EmitContext& ctx);
- void EmitLoadStorageS16(EmitContext& ctx);
- Id EmitLoadStorage32(EmitContext& ctx, const IR::Value& binding, const IR::Value& offset);
- void EmitLoadStorage64(EmitContext& ctx);
- void EmitLoadStorage128(EmitContext& ctx);
- void EmitWriteStorageU8(EmitContext& ctx);
- void EmitWriteStorageS8(EmitContext& ctx);
- void EmitWriteStorageU16(EmitContext& ctx);
- void EmitWriteStorageS16(EmitContext& ctx);
- void EmitWriteStorage32(EmitContext& ctx, const IR::Value& binding, const IR::Value& offset,
- Id value);
- void EmitWriteStorage64(EmitContext& ctx);
- void EmitWriteStorage128(EmitContext& ctx);
- void EmitCompositeConstructU32x2(EmitContext& ctx);
- void EmitCompositeConstructU32x3(EmitContext& ctx);
- void EmitCompositeConstructU32x4(EmitContext& ctx);
- void EmitCompositeExtractU32x2(EmitContext& ctx);
- Id EmitCompositeExtractU32x3(EmitContext& ctx, Id vector, u32 index);
- void EmitCompositeExtractU32x4(EmitContext& ctx);
- void EmitCompositeConstructF16x2(EmitContext& ctx);
- void EmitCompositeConstructF16x3(EmitContext& ctx);
- void EmitCompositeConstructF16x4(EmitContext& ctx);
- void EmitCompositeExtractF16x2(EmitContext& ctx);
- void EmitCompositeExtractF16x3(EmitContext& ctx);
- void EmitCompositeExtractF16x4(EmitContext& ctx);
- void EmitCompositeConstructF32x2(EmitContext& ctx);
- void EmitCompositeConstructF32x3(EmitContext& ctx);
- void EmitCompositeConstructF32x4(EmitContext& ctx);
- void EmitCompositeExtractF32x2(EmitContext& ctx);
- void EmitCompositeExtractF32x3(EmitContext& ctx);
- void EmitCompositeExtractF32x4(EmitContext& ctx);
- void EmitCompositeConstructF64x2(EmitContext& ctx);
- void EmitCompositeConstructF64x3(EmitContext& ctx);
- void EmitCompositeConstructF64x4(EmitContext& ctx);
- void EmitCompositeExtractF64x2(EmitContext& ctx);
- void EmitCompositeExtractF64x3(EmitContext& ctx);
- void EmitCompositeExtractF64x4(EmitContext& ctx);
- void EmitSelect8(EmitContext& ctx);
- void EmitSelect16(EmitContext& ctx);
- void EmitSelect32(EmitContext& ctx);
- void EmitSelect64(EmitContext& ctx);
- void EmitBitCastU16F16(EmitContext& ctx);
- Id EmitBitCastU32F32(EmitContext& ctx, Id value);
- void EmitBitCastU64F64(EmitContext& ctx);
- void EmitBitCastF16U16(EmitContext& ctx);
- Id EmitBitCastF32U32(EmitContext& ctx, Id value);
- void EmitBitCastF64U64(EmitContext& ctx);
- void EmitPackUint2x32(EmitContext& ctx);
- void EmitUnpackUint2x32(EmitContext& ctx);
- void EmitPackFloat2x16(EmitContext& ctx);
- void EmitUnpackFloat2x16(EmitContext& ctx);
- void EmitPackDouble2x32(EmitContext& ctx);
- void EmitUnpackDouble2x32(EmitContext& ctx);
- void EmitGetZeroFromOp(EmitContext& ctx);
- void EmitGetSignFromOp(EmitContext& ctx);
- void EmitGetCarryFromOp(EmitContext& ctx);
- void EmitGetOverflowFromOp(EmitContext& ctx);
- void EmitFPAbs16(EmitContext& ctx);
- void EmitFPAbs32(EmitContext& ctx);
- void EmitFPAbs64(EmitContext& ctx);
- Id EmitFPAdd16(EmitContext& ctx, IR::Inst* inst, Id a, Id b);
- Id EmitFPAdd32(EmitContext& ctx, IR::Inst* inst, Id a, Id b);
- Id EmitFPAdd64(EmitContext& ctx, IR::Inst* inst, Id a, Id b);
- Id EmitFPFma16(EmitContext& ctx, IR::Inst* inst, Id a, Id b, Id c);
- Id EmitFPFma32(EmitContext& ctx, IR::Inst* inst, Id a, Id b, Id c);
- Id EmitFPFma64(EmitContext& ctx, IR::Inst* inst, Id a, Id b, Id c);
- void EmitFPMax32(EmitContext& ctx);
- void EmitFPMax64(EmitContext& ctx);
- void EmitFPMin32(EmitContext& ctx);
- void EmitFPMin64(EmitContext& ctx);
- Id EmitFPMul16(EmitContext& ctx, IR::Inst* inst, Id a, Id b);
- Id EmitFPMul32(EmitContext& ctx, IR::Inst* inst, Id a, Id b);
- Id EmitFPMul64(EmitContext& ctx, IR::Inst* inst, Id a, Id b);
- void EmitFPNeg16(EmitContext& ctx);
- void EmitFPNeg32(EmitContext& ctx);
- void EmitFPNeg64(EmitContext& ctx);
- void EmitFPRecip32(EmitContext& ctx);
- void EmitFPRecip64(EmitContext& ctx);
- void EmitFPRecipSqrt32(EmitContext& ctx);
- void EmitFPRecipSqrt64(EmitContext& ctx);
- void EmitFPSqrt(EmitContext& ctx);
- void EmitFPSin(EmitContext& ctx);
- void EmitFPSinNotReduced(EmitContext& ctx);
- void EmitFPExp2(EmitContext& ctx);
- void EmitFPExp2NotReduced(EmitContext& ctx);
- void EmitFPCos(EmitContext& ctx);
- void EmitFPCosNotReduced(EmitContext& ctx);
- void EmitFPLog2(EmitContext& ctx);
- void EmitFPSaturate16(EmitContext& ctx);
- void EmitFPSaturate32(EmitContext& ctx);
- void EmitFPSaturate64(EmitContext& ctx);
- void EmitFPRoundEven16(EmitContext& ctx);
- void EmitFPRoundEven32(EmitContext& ctx);
- void EmitFPRoundEven64(EmitContext& ctx);
- void EmitFPFloor16(EmitContext& ctx);
- void EmitFPFloor32(EmitContext& ctx);
- void EmitFPFloor64(EmitContext& ctx);
- void EmitFPCeil16(EmitContext& ctx);
- void EmitFPCeil32(EmitContext& ctx);
- void EmitFPCeil64(EmitContext& ctx);
- void EmitFPTrunc16(EmitContext& ctx);
- void EmitFPTrunc32(EmitContext& ctx);
- void EmitFPTrunc64(EmitContext& ctx);
- Id EmitIAdd32(EmitContext& ctx, IR::Inst* inst, Id a, Id b);
- void EmitIAdd64(EmitContext& ctx);
- Id EmitISub32(EmitContext& ctx, Id a, Id b);
- void EmitISub64(EmitContext& ctx);
- Id EmitIMul32(EmitContext& ctx, Id a, Id b);
- void EmitINeg32(EmitContext& ctx);
- void EmitIAbs32(EmitContext& ctx);
- Id EmitShiftLeftLogical32(EmitContext& ctx, Id base, Id shift);
- void EmitShiftRightLogical32(EmitContext& ctx);
- void EmitShiftRightArithmetic32(EmitContext& ctx);
- void EmitBitwiseAnd32(EmitContext& ctx);
- void EmitBitwiseOr32(EmitContext& ctx);
- void EmitBitwiseXor32(EmitContext& ctx);
- void EmitBitFieldInsert(EmitContext& ctx);
- void EmitBitFieldSExtract(EmitContext& ctx);
- Id EmitBitFieldUExtract(EmitContext& ctx, Id base, Id offset, Id count);
- Id EmitSLessThan(EmitContext& ctx, Id lhs, Id rhs);
- void EmitULessThan(EmitContext& ctx);
- void EmitIEqual(EmitContext& ctx);
- void EmitSLessThanEqual(EmitContext& ctx);
- void EmitULessThanEqual(EmitContext& ctx);
- Id EmitSGreaterThan(EmitContext& ctx, Id lhs, Id rhs);
- void EmitUGreaterThan(EmitContext& ctx);
- void EmitINotEqual(EmitContext& ctx);
- void EmitSGreaterThanEqual(EmitContext& ctx);
- Id EmitUGreaterThanEqual(EmitContext& ctx, Id lhs, Id rhs);
- void EmitLogicalOr(EmitContext& ctx);
- void EmitLogicalAnd(EmitContext& ctx);
- void EmitLogicalXor(EmitContext& ctx);
- void EmitLogicalNot(EmitContext& ctx);
- void EmitConvertS16F16(EmitContext& ctx);
- void EmitConvertS16F32(EmitContext& ctx);
- void EmitConvertS16F64(EmitContext& ctx);
- void EmitConvertS32F16(EmitContext& ctx);
- void EmitConvertS32F32(EmitContext& ctx);
- void EmitConvertS32F64(EmitContext& ctx);
- void EmitConvertS64F16(EmitContext& ctx);
- void EmitConvertS64F32(EmitContext& ctx);
- void EmitConvertS64F64(EmitContext& ctx);
- void EmitConvertU16F16(EmitContext& ctx);
- void EmitConvertU16F32(EmitContext& ctx);
- void EmitConvertU16F64(EmitContext& ctx);
- void EmitConvertU32F16(EmitContext& ctx);
- void EmitConvertU32F32(EmitContext& ctx);
- void EmitConvertU32F64(EmitContext& ctx);
- void EmitConvertU64F16(EmitContext& ctx);
- void EmitConvertU64F32(EmitContext& ctx);
- void EmitConvertU64F64(EmitContext& ctx);
- void EmitConvertU64U32(EmitContext& ctx);
- void EmitConvertU32U64(EmitContext& ctx);
-};
+// Microinstruction emitters
+Id EmitPhi(EmitContext& ctx, IR::Inst* inst);
+void EmitVoid(EmitContext& ctx);
+Id EmitIdentity(EmitContext& ctx, const IR::Value& value);
+void EmitBranch(EmitContext& ctx, IR::Block* label);
+void EmitBranchConditional(EmitContext& ctx, Id condition, IR::Block* true_label,
+ IR::Block* false_label);
+void EmitLoopMerge(EmitContext& ctx, IR::Block* merge_label, IR::Block* continue_label);
+void EmitSelectionMerge(EmitContext& ctx, IR::Block* merge_label);
+void EmitReturn(EmitContext& ctx);
+void EmitGetRegister(EmitContext& ctx);
+void EmitSetRegister(EmitContext& ctx);
+void EmitGetPred(EmitContext& ctx);
+void EmitSetPred(EmitContext& ctx);
+void EmitSetGotoVariable(EmitContext& ctx);
+void EmitGetGotoVariable(EmitContext& ctx);
+Id EmitGetCbuf(EmitContext& ctx, const IR::Value& binding, const IR::Value& offset);
+void EmitGetAttribute(EmitContext& ctx);
+void EmitSetAttribute(EmitContext& ctx);
+void EmitGetAttributeIndexed(EmitContext& ctx);
+void EmitSetAttributeIndexed(EmitContext& ctx);
+void EmitGetZFlag(EmitContext& ctx);
+void EmitGetSFlag(EmitContext& ctx);
+void EmitGetCFlag(EmitContext& ctx);
+void EmitGetOFlag(EmitContext& ctx);
+void EmitSetZFlag(EmitContext& ctx);
+void EmitSetSFlag(EmitContext& ctx);
+void EmitSetCFlag(EmitContext& ctx);
+void EmitSetOFlag(EmitContext& ctx);
+Id EmitWorkgroupId(EmitContext& ctx);
+Id EmitLocalInvocationId(EmitContext& ctx);
+Id EmitUndefU1(EmitContext& ctx);
+Id EmitUndefU8(EmitContext& ctx);
+Id EmitUndefU16(EmitContext& ctx);
+Id EmitUndefU32(EmitContext& ctx);
+Id EmitUndefU64(EmitContext& ctx);
+void EmitLoadGlobalU8(EmitContext& ctx);
+void EmitLoadGlobalS8(EmitContext& ctx);
+void EmitLoadGlobalU16(EmitContext& ctx);
+void EmitLoadGlobalS16(EmitContext& ctx);
+void EmitLoadGlobal32(EmitContext& ctx);
+void EmitLoadGlobal64(EmitContext& ctx);
+void EmitLoadGlobal128(EmitContext& ctx);
+void EmitWriteGlobalU8(EmitContext& ctx);
+void EmitWriteGlobalS8(EmitContext& ctx);
+void EmitWriteGlobalU16(EmitContext& ctx);
+void EmitWriteGlobalS16(EmitContext& ctx);
+void EmitWriteGlobal32(EmitContext& ctx);
+void EmitWriteGlobal64(EmitContext& ctx);
+void EmitWriteGlobal128(EmitContext& ctx);
+void EmitLoadStorageU8(EmitContext& ctx);
+void EmitLoadStorageS8(EmitContext& ctx);
+void EmitLoadStorageU16(EmitContext& ctx);
+void EmitLoadStorageS16(EmitContext& ctx);
+Id EmitLoadStorage32(EmitContext& ctx, const IR::Value& binding, const IR::Value& offset);
+void EmitLoadStorage64(EmitContext& ctx);
+void EmitLoadStorage128(EmitContext& ctx);
+void EmitWriteStorageU8(EmitContext& ctx);
+void EmitWriteStorageS8(EmitContext& ctx);
+void EmitWriteStorageU16(EmitContext& ctx);
+void EmitWriteStorageS16(EmitContext& ctx);
+void EmitWriteStorage32(EmitContext& ctx, const IR::Value& binding, const IR::Value& offset,
+ Id value);
+void EmitWriteStorage64(EmitContext& ctx);
+void EmitWriteStorage128(EmitContext& ctx);
+void EmitCompositeConstructU32x2(EmitContext& ctx);
+void EmitCompositeConstructU32x3(EmitContext& ctx);
+void EmitCompositeConstructU32x4(EmitContext& ctx);
+void EmitCompositeExtractU32x2(EmitContext& ctx);
+Id EmitCompositeExtractU32x3(EmitContext& ctx, Id vector, u32 index);
+void EmitCompositeExtractU32x4(EmitContext& ctx);
+void EmitCompositeConstructF16x2(EmitContext& ctx);
+void EmitCompositeConstructF16x3(EmitContext& ctx);
+void EmitCompositeConstructF16x4(EmitContext& ctx);
+void EmitCompositeExtractF16x2(EmitContext& ctx);
+void EmitCompositeExtractF16x3(EmitContext& ctx);
+void EmitCompositeExtractF16x4(EmitContext& ctx);
+void EmitCompositeConstructF32x2(EmitContext& ctx);
+void EmitCompositeConstructF32x3(EmitContext& ctx);
+void EmitCompositeConstructF32x4(EmitContext& ctx);
+void EmitCompositeExtractF32x2(EmitContext& ctx);
+void EmitCompositeExtractF32x3(EmitContext& ctx);
+void EmitCompositeExtractF32x4(EmitContext& ctx);
+void EmitCompositeConstructF64x2(EmitContext& ctx);
+void EmitCompositeConstructF64x3(EmitContext& ctx);
+void EmitCompositeConstructF64x4(EmitContext& ctx);
+void EmitCompositeExtractF64x2(EmitContext& ctx);
+void EmitCompositeExtractF64x3(EmitContext& ctx);
+void EmitCompositeExtractF64x4(EmitContext& ctx);
+void EmitSelect8(EmitContext& ctx);
+void EmitSelect16(EmitContext& ctx);
+void EmitSelect32(EmitContext& ctx);
+void EmitSelect64(EmitContext& ctx);
+void EmitBitCastU16F16(EmitContext& ctx);
+Id EmitBitCastU32F32(EmitContext& ctx, Id value);
+void EmitBitCastU64F64(EmitContext& ctx);
+void EmitBitCastF16U16(EmitContext& ctx);
+Id EmitBitCastF32U32(EmitContext& ctx, Id value);
+void EmitBitCastF64U64(EmitContext& ctx);
+void EmitPackUint2x32(EmitContext& ctx);
+void EmitUnpackUint2x32(EmitContext& ctx);
+void EmitPackFloat2x16(EmitContext& ctx);
+void EmitUnpackFloat2x16(EmitContext& ctx);
+void EmitPackDouble2x32(EmitContext& ctx);
+void EmitUnpackDouble2x32(EmitContext& ctx);
+void EmitGetZeroFromOp(EmitContext& ctx);
+void EmitGetSignFromOp(EmitContext& ctx);
+void EmitGetCarryFromOp(EmitContext& ctx);
+void EmitGetOverflowFromOp(EmitContext& ctx);
+void EmitFPAbs16(EmitContext& ctx);
+void EmitFPAbs32(EmitContext& ctx);
+void EmitFPAbs64(EmitContext& ctx);
+Id EmitFPAdd16(EmitContext& ctx, IR::Inst* inst, Id a, Id b);
+Id EmitFPAdd32(EmitContext& ctx, IR::Inst* inst, Id a, Id b);
+Id EmitFPAdd64(EmitContext& ctx, IR::Inst* inst, Id a, Id b);
+Id EmitFPFma16(EmitContext& ctx, IR::Inst* inst, Id a, Id b, Id c);
+Id EmitFPFma32(EmitContext& ctx, IR::Inst* inst, Id a, Id b, Id c);
+Id EmitFPFma64(EmitContext& ctx, IR::Inst* inst, Id a, Id b, Id c);
+void EmitFPMax32(EmitContext& ctx);
+void EmitFPMax64(EmitContext& ctx);
+void EmitFPMin32(EmitContext& ctx);
+void EmitFPMin64(EmitContext& ctx);
+Id EmitFPMul16(EmitContext& ctx, IR::Inst* inst, Id a, Id b);
+Id EmitFPMul32(EmitContext& ctx, IR::Inst* inst, Id a, Id b);
+Id EmitFPMul64(EmitContext& ctx, IR::Inst* inst, Id a, Id b);
+void EmitFPNeg16(EmitContext& ctx);
+void EmitFPNeg32(EmitContext& ctx);
+void EmitFPNeg64(EmitContext& ctx);
+void EmitFPRecip32(EmitContext& ctx);
+void EmitFPRecip64(EmitContext& ctx);
+void EmitFPRecipSqrt32(EmitContext& ctx);
+void EmitFPRecipSqrt64(EmitContext& ctx);
+void EmitFPSqrt(EmitContext& ctx);
+void EmitFPSin(EmitContext& ctx);
+void EmitFPSinNotReduced(EmitContext& ctx);
+void EmitFPExp2(EmitContext& ctx);
+void EmitFPExp2NotReduced(EmitContext& ctx);
+void EmitFPCos(EmitContext& ctx);
+void EmitFPCosNotReduced(EmitContext& ctx);
+void EmitFPLog2(EmitContext& ctx);
+void EmitFPSaturate16(EmitContext& ctx);
+void EmitFPSaturate32(EmitContext& ctx);
+void EmitFPSaturate64(EmitContext& ctx);
+void EmitFPRoundEven16(EmitContext& ctx);
+void EmitFPRoundEven32(EmitContext& ctx);
+void EmitFPRoundEven64(EmitContext& ctx);
+void EmitFPFloor16(EmitContext& ctx);
+void EmitFPFloor32(EmitContext& ctx);
+void EmitFPFloor64(EmitContext& ctx);
+void EmitFPCeil16(EmitContext& ctx);
+void EmitFPCeil32(EmitContext& ctx);
+void EmitFPCeil64(EmitContext& ctx);
+void EmitFPTrunc16(EmitContext& ctx);
+void EmitFPTrunc32(EmitContext& ctx);
+void EmitFPTrunc64(EmitContext& ctx);
+Id EmitIAdd32(EmitContext& ctx, IR::Inst* inst, Id a, Id b);
+void EmitIAdd64(EmitContext& ctx);
+Id EmitISub32(EmitContext& ctx, Id a, Id b);
+void EmitISub64(EmitContext& ctx);
+Id EmitIMul32(EmitContext& ctx, Id a, Id b);
+void EmitINeg32(EmitContext& ctx);
+void EmitIAbs32(EmitContext& ctx);
+Id EmitShiftLeftLogical32(EmitContext& ctx, Id base, Id shift);
+void EmitShiftRightLogical32(EmitContext& ctx);
+void EmitShiftRightArithmetic32(EmitContext& ctx);
+void EmitBitwiseAnd32(EmitContext& ctx);
+void EmitBitwiseOr32(EmitContext& ctx);
+void EmitBitwiseXor32(EmitContext& ctx);
+void EmitBitFieldInsert(EmitContext& ctx);
+void EmitBitFieldSExtract(EmitContext& ctx);
+Id EmitBitFieldUExtract(EmitContext& ctx, Id base, Id offset, Id count);
+Id EmitSLessThan(EmitContext& ctx, Id lhs, Id rhs);
+void EmitULessThan(EmitContext& ctx);
+void EmitIEqual(EmitContext& ctx);
+void EmitSLessThanEqual(EmitContext& ctx);
+void EmitULessThanEqual(EmitContext& ctx);
+Id EmitSGreaterThan(EmitContext& ctx, Id lhs, Id rhs);
+void EmitUGreaterThan(EmitContext& ctx);
+void EmitINotEqual(EmitContext& ctx);
+void EmitSGreaterThanEqual(EmitContext& ctx);
+Id EmitUGreaterThanEqual(EmitContext& ctx, Id lhs, Id rhs);
+void EmitLogicalOr(EmitContext& ctx);
+void EmitLogicalAnd(EmitContext& ctx);
+void EmitLogicalXor(EmitContext& ctx);
+void EmitLogicalNot(EmitContext& ctx);
+void EmitConvertS16F16(EmitContext& ctx);
+void EmitConvertS16F32(EmitContext& ctx);
+void EmitConvertS16F64(EmitContext& ctx);
+void EmitConvertS32F16(EmitContext& ctx);
+void EmitConvertS32F32(EmitContext& ctx);
+void EmitConvertS32F64(EmitContext& ctx);
+void EmitConvertS64F16(EmitContext& ctx);
+void EmitConvertS64F32(EmitContext& ctx);
+void EmitConvertS64F64(EmitContext& ctx);
+void EmitConvertU16F16(EmitContext& ctx);
+void EmitConvertU16F32(EmitContext& ctx);
+void EmitConvertU16F64(EmitContext& ctx);
+void EmitConvertU32F16(EmitContext& ctx);
+void EmitConvertU32F32(EmitContext& ctx);
+void EmitConvertU32F64(EmitContext& ctx);
+void EmitConvertU64F16(EmitContext& ctx);
+void EmitConvertU64F32(EmitContext& ctx);
+void EmitConvertU64F64(EmitContext& ctx);
+void EmitConvertU64U32(EmitContext& ctx);
+void EmitConvertU32U64(EmitContext& ctx);
} // namespace Shader::Backend::SPIRV
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_bitwise_conversion.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_bitwise_conversion.cpp
index af82df99c..49c200498 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_bitwise_conversion.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_bitwise_conversion.cpp
@@ -6,51 +6,51 @@
namespace Shader::Backend::SPIRV {
-void EmitSPIRV::EmitBitCastU16F16(EmitContext&) {
+void EmitBitCastU16F16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-Id EmitSPIRV::EmitBitCastU32F32(EmitContext& ctx, Id value) {
+Id EmitBitCastU32F32(EmitContext& ctx, Id value) {
return ctx.OpBitcast(ctx.U32[1], value);
}
-void EmitSPIRV::EmitBitCastU64F64(EmitContext&) {
+void EmitBitCastU64F64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitBitCastF16U16(EmitContext&) {
+void EmitBitCastF16U16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-Id EmitSPIRV::EmitBitCastF32U32(EmitContext& ctx, Id value) {
+Id EmitBitCastF32U32(EmitContext& ctx, Id value) {
return ctx.OpBitcast(ctx.F32[1], value);
}
-void EmitSPIRV::EmitBitCastF64U64(EmitContext&) {
+void EmitBitCastF64U64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitPackUint2x32(EmitContext&) {
+void EmitPackUint2x32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitUnpackUint2x32(EmitContext&) {
+void EmitUnpackUint2x32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitPackFloat2x16(EmitContext&) {
+void EmitPackFloat2x16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitUnpackFloat2x16(EmitContext&) {
+void EmitUnpackFloat2x16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitPackDouble2x32(EmitContext&) {
+void EmitPackDouble2x32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitUnpackDouble2x32(EmitContext&) {
+void EmitUnpackDouble2x32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_composite.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_composite.cpp
index a7374c89d..348e4796d 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_composite.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_composite.cpp
@@ -6,99 +6,99 @@
namespace Shader::Backend::SPIRV {
-void EmitSPIRV::EmitCompositeConstructU32x2(EmitContext&) {
+void EmitCompositeConstructU32x2(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitCompositeConstructU32x3(EmitContext&) {
+void EmitCompositeConstructU32x3(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitCompositeConstructU32x4(EmitContext&) {
+void EmitCompositeConstructU32x4(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitCompositeExtractU32x2(EmitContext&) {
+void EmitCompositeExtractU32x2(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-Id EmitSPIRV::EmitCompositeExtractU32x3(EmitContext& ctx, Id vector, u32 index) {
+Id EmitCompositeExtractU32x3(EmitContext& ctx, Id vector, u32 index) {
return ctx.OpCompositeExtract(ctx.U32[1], vector, index);
}
-void EmitSPIRV::EmitCompositeExtractU32x4(EmitContext&) {
+void EmitCompositeExtractU32x4(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitCompositeConstructF16x2(EmitContext&) {
+void EmitCompositeConstructF16x2(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitCompositeConstructF16x3(EmitContext&) {
+void EmitCompositeConstructF16x3(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitCompositeConstructF16x4(EmitContext&) {
+void EmitCompositeConstructF16x4(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitCompositeExtractF16x2(EmitContext&) {
+void EmitCompositeExtractF16x2(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitCompositeExtractF16x3(EmitContext&) {
+void EmitCompositeExtractF16x3(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitCompositeExtractF16x4(EmitContext&) {
+void EmitCompositeExtractF16x4(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitCompositeConstructF32x2(EmitContext&) {
+void EmitCompositeConstructF32x2(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitCompositeConstructF32x3(EmitContext&) {
+void EmitCompositeConstructF32x3(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitCompositeConstructF32x4(EmitContext&) {
+void EmitCompositeConstructF32x4(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitCompositeExtractF32x2(EmitContext&) {
+void EmitCompositeExtractF32x2(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitCompositeExtractF32x3(EmitContext&) {
+void EmitCompositeExtractF32x3(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitCompositeExtractF32x4(EmitContext&) {
+void EmitCompositeExtractF32x4(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitCompositeConstructF64x2(EmitContext&) {
+void EmitCompositeConstructF64x2(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitCompositeConstructF64x3(EmitContext&) {
+void EmitCompositeConstructF64x3(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitCompositeConstructF64x4(EmitContext&) {
+void EmitCompositeConstructF64x4(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitCompositeExtractF64x2(EmitContext&) {
+void EmitCompositeExtractF64x2(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitCompositeExtractF64x3(EmitContext&) {
+void EmitCompositeExtractF64x3(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitCompositeExtractF64x4(EmitContext&) {
+void EmitCompositeExtractF64x4(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_context_get_set.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_context_get_set.cpp
index f4c9970eb..eb9c01c5a 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_context_get_set.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_context_get_set.cpp
@@ -6,31 +6,31 @@
namespace Shader::Backend::SPIRV {
-void EmitSPIRV::EmitGetRegister(EmitContext&) {
+void EmitGetRegister(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitSetRegister(EmitContext&) {
+void EmitSetRegister(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitGetPred(EmitContext&) {
+void EmitGetPred(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitSetPred(EmitContext&) {
+void EmitSetPred(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitSetGotoVariable(EmitContext&) {
+void EmitSetGotoVariable(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitGetGotoVariable(EmitContext&) {
+void EmitGetGotoVariable(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-Id EmitSPIRV::EmitGetCbuf(EmitContext& ctx, const IR::Value& binding, const IR::Value& offset) {
+Id EmitGetCbuf(EmitContext& ctx, const IR::Value& binding, const IR::Value& offset) {
if (!binding.IsImmediate()) {
throw NotImplementedException("Constant buffer indexing");
}
@@ -43,59 +43,59 @@ Id EmitSPIRV::EmitGetCbuf(EmitContext& ctx, const IR::Value& binding, const IR::
return ctx.OpLoad(ctx.U32[1], access_chain);
}
-void EmitSPIRV::EmitGetAttribute(EmitContext&) {
+void EmitGetAttribute(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitSetAttribute(EmitContext&) {
+void EmitSetAttribute(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitGetAttributeIndexed(EmitContext&) {
+void EmitGetAttributeIndexed(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitSetAttributeIndexed(EmitContext&) {
+void EmitSetAttributeIndexed(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitGetZFlag(EmitContext&) {
+void EmitGetZFlag(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitGetSFlag(EmitContext&) {
+void EmitGetSFlag(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitGetCFlag(EmitContext&) {
+void EmitGetCFlag(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitGetOFlag(EmitContext&) {
+void EmitGetOFlag(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitSetZFlag(EmitContext&) {
+void EmitSetZFlag(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitSetSFlag(EmitContext&) {
+void EmitSetSFlag(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitSetCFlag(EmitContext&) {
+void EmitSetCFlag(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitSetOFlag(EmitContext&) {
+void EmitSetOFlag(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-Id EmitSPIRV::EmitWorkgroupId(EmitContext& ctx) {
+Id EmitWorkgroupId(EmitContext& ctx) {
return ctx.OpLoad(ctx.U32[3], ctx.workgroup_id);
}
-Id EmitSPIRV::EmitLocalInvocationId(EmitContext& ctx) {
+Id EmitLocalInvocationId(EmitContext& ctx) {
return ctx.OpLoad(ctx.U32[3], ctx.local_invocation_id);
}
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_control_flow.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_control_flow.cpp
index 549c1907a..6c4199664 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_control_flow.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_control_flow.cpp
@@ -6,25 +6,25 @@
namespace Shader::Backend::SPIRV {
-void EmitSPIRV::EmitBranch(EmitContext& ctx, IR::Block* label) {
+void EmitBranch(EmitContext& ctx, IR::Block* label) {
ctx.OpBranch(label->Definition<Id>());
}
-void EmitSPIRV::EmitBranchConditional(EmitContext& ctx, Id condition, IR::Block* true_label,
+void EmitBranchConditional(EmitContext& ctx, Id condition, IR::Block* true_label,
IR::Block* false_label) {
ctx.OpBranchConditional(condition, true_label->Definition<Id>(), false_label->Definition<Id>());
}
-void EmitSPIRV::EmitLoopMerge(EmitContext& ctx, IR::Block* merge_label, IR::Block* continue_label) {
+void EmitLoopMerge(EmitContext& ctx, IR::Block* merge_label, IR::Block* continue_label) {
ctx.OpLoopMerge(merge_label->Definition<Id>(), continue_label->Definition<Id>(),
spv::LoopControlMask::MaskNone);
}
-void EmitSPIRV::EmitSelectionMerge(EmitContext& ctx, IR::Block* merge_label) {
+void EmitSelectionMerge(EmitContext& ctx, IR::Block* merge_label) {
ctx.OpSelectionMerge(merge_label->Definition<Id>(), spv::SelectionControlMask::MaskNone);
}
-void EmitSPIRV::EmitReturn(EmitContext& ctx) {
+void EmitReturn(EmitContext& ctx) {
ctx.OpReturn();
}
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_floating_point.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_floating_point.cpp
index c9bc121f8..d24fbb353 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_floating_point.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_floating_point.cpp
@@ -33,187 +33,187 @@ Id Decorate(EmitContext& ctx, IR::Inst* inst, Id op) {
} // Anonymous namespace
-void EmitSPIRV::EmitFPAbs16(EmitContext&) {
+void EmitFPAbs16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPAbs32(EmitContext&) {
+void EmitFPAbs32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPAbs64(EmitContext&) {
+void EmitFPAbs64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-Id EmitSPIRV::EmitFPAdd16(EmitContext& ctx, IR::Inst* inst, Id a, Id b) {
+Id EmitFPAdd16(EmitContext& ctx, IR::Inst* inst, Id a, Id b) {
return Decorate(ctx, inst, ctx.OpFAdd(ctx.F16[1], a, b));
}
-Id EmitSPIRV::EmitFPAdd32(EmitContext& ctx, IR::Inst* inst, Id a, Id b) {
+Id EmitFPAdd32(EmitContext& ctx, IR::Inst* inst, Id a, Id b) {
return Decorate(ctx, inst, ctx.OpFAdd(ctx.F32[1], a, b));
}
-Id EmitSPIRV::EmitFPAdd64(EmitContext& ctx, IR::Inst* inst, Id a, Id b) {
+Id EmitFPAdd64(EmitContext& ctx, IR::Inst* inst, Id a, Id b) {
return Decorate(ctx, inst, ctx.OpFAdd(ctx.F64[1], a, b));
}
-Id EmitSPIRV::EmitFPFma16(EmitContext& ctx, IR::Inst* inst, Id a, Id b, Id c) {
+Id EmitFPFma16(EmitContext& ctx, IR::Inst* inst, Id a, Id b, Id c) {
return Decorate(ctx, inst, ctx.OpFma(ctx.F16[1], a, b, c));
}
-Id EmitSPIRV::EmitFPFma32(EmitContext& ctx, IR::Inst* inst, Id a, Id b, Id c) {
+Id EmitFPFma32(EmitContext& ctx, IR::Inst* inst, Id a, Id b, Id c) {
return Decorate(ctx, inst, ctx.OpFma(ctx.F32[1], a, b, c));
}
-Id EmitSPIRV::EmitFPFma64(EmitContext& ctx, IR::Inst* inst, Id a, Id b, Id c) {
+Id EmitFPFma64(EmitContext& ctx, IR::Inst* inst, Id a, Id b, Id c) {
return Decorate(ctx, inst, ctx.OpFma(ctx.F64[1], a, b, c));
}
-void EmitSPIRV::EmitFPMax32(EmitContext&) {
+void EmitFPMax32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPMax64(EmitContext&) {
+void EmitFPMax64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPMin32(EmitContext&) {
+void EmitFPMin32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPMin64(EmitContext&) {
+void EmitFPMin64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-Id EmitSPIRV::EmitFPMul16(EmitContext& ctx, IR::Inst* inst, Id a, Id b) {
+Id EmitFPMul16(EmitContext& ctx, IR::Inst* inst, Id a, Id b) {
return Decorate(ctx, inst, ctx.OpFMul(ctx.F16[1], a, b));
}
-Id EmitSPIRV::EmitFPMul32(EmitContext& ctx, IR::Inst* inst, Id a, Id b) {
+Id EmitFPMul32(EmitContext& ctx, IR::Inst* inst, Id a, Id b) {
return Decorate(ctx, inst, ctx.OpFMul(ctx.F32[1], a, b));
}
-Id EmitSPIRV::EmitFPMul64(EmitContext& ctx, IR::Inst* inst, Id a, Id b) {
+Id EmitFPMul64(EmitContext& ctx, IR::Inst* inst, Id a, Id b) {
return Decorate(ctx, inst, ctx.OpFMul(ctx.F64[1], a, b));
}
-void EmitSPIRV::EmitFPNeg16(EmitContext&) {
+void EmitFPNeg16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPNeg32(EmitContext&) {
+void EmitFPNeg32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPNeg64(EmitContext&) {
+void EmitFPNeg64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPRecip32(EmitContext&) {
+void EmitFPRecip32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPRecip64(EmitContext&) {
+void EmitFPRecip64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPRecipSqrt32(EmitContext&) {
+void EmitFPRecipSqrt32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPRecipSqrt64(EmitContext&) {
+void EmitFPRecipSqrt64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPSqrt(EmitContext&) {
+void EmitFPSqrt(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPSin(EmitContext&) {
+void EmitFPSin(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPSinNotReduced(EmitContext&) {
+void EmitFPSinNotReduced(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPExp2(EmitContext&) {
+void EmitFPExp2(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPExp2NotReduced(EmitContext&) {
+void EmitFPExp2NotReduced(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPCos(EmitContext&) {
+void EmitFPCos(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPCosNotReduced(EmitContext&) {
+void EmitFPCosNotReduced(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPLog2(EmitContext&) {
+void EmitFPLog2(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPSaturate16(EmitContext&) {
+void EmitFPSaturate16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPSaturate32(EmitContext&) {
+void EmitFPSaturate32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPSaturate64(EmitContext&) {
+void EmitFPSaturate64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPRoundEven16(EmitContext&) {
+void EmitFPRoundEven16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPRoundEven32(EmitContext&) {
+void EmitFPRoundEven32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPRoundEven64(EmitContext&) {
+void EmitFPRoundEven64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPFloor16(EmitContext&) {
+void EmitFPFloor16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPFloor32(EmitContext&) {
+void EmitFPFloor32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPFloor64(EmitContext&) {
+void EmitFPFloor64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPCeil16(EmitContext&) {
+void EmitFPCeil16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPCeil32(EmitContext&) {
+void EmitFPCeil32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPCeil64(EmitContext&) {
+void EmitFPCeil64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPTrunc16(EmitContext&) {
+void EmitFPTrunc16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPTrunc32(EmitContext&) {
+void EmitFPTrunc32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitFPTrunc64(EmitContext&) {
+void EmitFPTrunc64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_integer.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_integer.cpp
index 32af94a73..a1d16b81e 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_integer.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_integer.cpp
@@ -6,126 +6,126 @@
namespace Shader::Backend::SPIRV {
-Id EmitSPIRV::EmitIAdd32(EmitContext& ctx, IR::Inst* inst, Id a, Id b) {
+Id EmitIAdd32(EmitContext& ctx, IR::Inst* inst, Id a, Id b) {
if (inst->HasAssociatedPseudoOperation()) {
throw NotImplementedException("Pseudo-operations on IAdd32");
}
return ctx.OpIAdd(ctx.U32[1], a, b);
}
-void EmitSPIRV::EmitIAdd64(EmitContext&) {
+void EmitIAdd64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-Id EmitSPIRV::EmitISub32(EmitContext& ctx, Id a, Id b) {
+Id EmitISub32(EmitContext& ctx, Id a, Id b) {
return ctx.OpISub(ctx.U32[1], a, b);
}
-void EmitSPIRV::EmitISub64(EmitContext&) {
+void EmitISub64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-Id EmitSPIRV::EmitIMul32(EmitContext& ctx, Id a, Id b) {
+Id EmitIMul32(EmitContext& ctx, Id a, Id b) {
return ctx.OpIMul(ctx.U32[1], a, b);
}
-void EmitSPIRV::EmitINeg32(EmitContext&) {
+void EmitINeg32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitIAbs32(EmitContext&) {
+void EmitIAbs32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-Id EmitSPIRV::EmitShiftLeftLogical32(EmitContext& ctx, Id base, Id shift) {
+Id EmitShiftLeftLogical32(EmitContext& ctx, Id base, Id shift) {
return ctx.OpShiftLeftLogical(ctx.U32[1], base, shift);
}
-void EmitSPIRV::EmitShiftRightLogical32(EmitContext&) {
+void EmitShiftRightLogical32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitShiftRightArithmetic32(EmitContext&) {
+void EmitShiftRightArithmetic32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitBitwiseAnd32(EmitContext&) {
+void EmitBitwiseAnd32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitBitwiseOr32(EmitContext&) {
+void EmitBitwiseOr32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitBitwiseXor32(EmitContext&) {
+void EmitBitwiseXor32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitBitFieldInsert(EmitContext&) {
+void EmitBitFieldInsert(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitBitFieldSExtract(EmitContext&) {
+void EmitBitFieldSExtract(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-Id EmitSPIRV::EmitBitFieldUExtract(EmitContext& ctx, Id base, Id offset, Id count) {
+Id EmitBitFieldUExtract(EmitContext& ctx, Id base, Id offset, Id count) {
return ctx.OpBitFieldUExtract(ctx.U32[1], base, offset, count);
}
-Id EmitSPIRV::EmitSLessThan(EmitContext& ctx, Id lhs, Id rhs) {
+Id EmitSLessThan(EmitContext& ctx, Id lhs, Id rhs) {
return ctx.OpSLessThan(ctx.U1, lhs, rhs);
}
-void EmitSPIRV::EmitULessThan(EmitContext&) {
+void EmitULessThan(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitIEqual(EmitContext&) {
+void EmitIEqual(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitSLessThanEqual(EmitContext&) {
+void EmitSLessThanEqual(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitULessThanEqual(EmitContext&) {
+void EmitULessThanEqual(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-Id EmitSPIRV::EmitSGreaterThan(EmitContext& ctx, Id lhs, Id rhs) {
+Id EmitSGreaterThan(EmitContext& ctx, Id lhs, Id rhs) {
return ctx.OpSGreaterThan(ctx.U1, lhs, rhs);
}
-void EmitSPIRV::EmitUGreaterThan(EmitContext&) {
+void EmitUGreaterThan(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitINotEqual(EmitContext&) {
+void EmitINotEqual(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitSGreaterThanEqual(EmitContext&) {
+void EmitSGreaterThanEqual(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-Id EmitSPIRV::EmitUGreaterThanEqual(EmitContext& ctx, Id lhs, Id rhs) {
+Id EmitUGreaterThanEqual(EmitContext& ctx, Id lhs, Id rhs) {
return ctx.OpUGreaterThanEqual(ctx.U1, lhs, rhs);
}
-void EmitSPIRV::EmitLogicalOr(EmitContext&) {
+void EmitLogicalOr(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitLogicalAnd(EmitContext&) {
+void EmitLogicalAnd(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitLogicalXor(EmitContext&) {
+void EmitLogicalXor(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitLogicalNot(EmitContext&) {
+void EmitLogicalNot(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_logical.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_logical.cpp
index 7b43c4ed8..ff2f4fb74 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_logical.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_logical.cpp
@@ -6,83 +6,83 @@
namespace Shader::Backend::SPIRV {
-void EmitSPIRV::EmitConvertS16F16(EmitContext&) {
+void EmitConvertS16F16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitConvertS16F32(EmitContext&) {
+void EmitConvertS16F32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitConvertS16F64(EmitContext&) {
+void EmitConvertS16F64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitConvertS32F16(EmitContext&) {
+void EmitConvertS32F16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitConvertS32F32(EmitContext&) {
+void EmitConvertS32F32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitConvertS32F64(EmitContext&) {
+void EmitConvertS32F64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitConvertS64F16(EmitContext&) {
+void EmitConvertS64F16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitConvertS64F32(EmitContext&) {
+void EmitConvertS64F32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitConvertS64F64(EmitContext&) {
+void EmitConvertS64F64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitConvertU16F16(EmitContext&) {
+void EmitConvertU16F16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitConvertU16F32(EmitContext&) {
+void EmitConvertU16F32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitConvertU16F64(EmitContext&) {
+void EmitConvertU16F64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitConvertU32F16(EmitContext&) {
+void EmitConvertU32F16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitConvertU32F32(EmitContext&) {
+void EmitConvertU32F32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitConvertU32F64(EmitContext&) {
+void EmitConvertU32F64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitConvertU64F16(EmitContext&) {
+void EmitConvertU64F16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitConvertU64F32(EmitContext&) {
+void EmitConvertU64F32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitConvertU64F64(EmitContext&) {
+void EmitConvertU64F64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitConvertU64U32(EmitContext&) {
+void EmitConvertU64U32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitConvertU32U64(EmitContext&) {
+void EmitConvertU32U64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_memory.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_memory.cpp
index 5769a3c95..77d698ffd 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_memory.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_memory.cpp
@@ -22,79 +22,79 @@ static Id StorageIndex(EmitContext& ctx, const IR::Value& offset, size_t element
return ctx.OpShiftRightLogical(ctx.U32[1], index, shift_id);
}
-void EmitSPIRV::EmitLoadGlobalU8(EmitContext&) {
+void EmitLoadGlobalU8(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitLoadGlobalS8(EmitContext&) {
+void EmitLoadGlobalS8(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitLoadGlobalU16(EmitContext&) {
+void EmitLoadGlobalU16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitLoadGlobalS16(EmitContext&) {
+void EmitLoadGlobalS16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitLoadGlobal32(EmitContext&) {
+void EmitLoadGlobal32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitLoadGlobal64(EmitContext&) {
+void EmitLoadGlobal64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitLoadGlobal128(EmitContext&) {
+void EmitLoadGlobal128(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitWriteGlobalU8(EmitContext&) {
+void EmitWriteGlobalU8(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitWriteGlobalS8(EmitContext&) {
+void EmitWriteGlobalS8(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitWriteGlobalU16(EmitContext&) {
+void EmitWriteGlobalU16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitWriteGlobalS16(EmitContext&) {
+void EmitWriteGlobalS16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitWriteGlobal32(EmitContext&) {
+void EmitWriteGlobal32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitWriteGlobal64(EmitContext&) {
+void EmitWriteGlobal64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitWriteGlobal128(EmitContext&) {
+void EmitWriteGlobal128(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitLoadStorageU8(EmitContext&) {
+void EmitLoadStorageU8(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitLoadStorageS8(EmitContext&) {
+void EmitLoadStorageS8(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitLoadStorageU16(EmitContext&) {
+void EmitLoadStorageU16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitLoadStorageS16(EmitContext&) {
+void EmitLoadStorageS16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-Id EmitSPIRV::EmitLoadStorage32(EmitContext& ctx, const IR::Value& binding,
+Id EmitLoadStorage32(EmitContext& ctx, const IR::Value& binding,
const IR::Value& offset) {
if (!binding.IsImmediate()) {
throw NotImplementedException("Dynamic storage buffer indexing");
@@ -105,31 +105,31 @@ Id EmitSPIRV::EmitLoadStorage32(EmitContext& ctx, const IR::Value& binding,
return ctx.OpLoad(ctx.U32[1], pointer);
}
-void EmitSPIRV::EmitLoadStorage64(EmitContext&) {
+void EmitLoadStorage64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitLoadStorage128(EmitContext&) {
+void EmitLoadStorage128(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitWriteStorageU8(EmitContext&) {
+void EmitWriteStorageU8(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitWriteStorageS8(EmitContext&) {
+void EmitWriteStorageS8(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitWriteStorageU16(EmitContext&) {
+void EmitWriteStorageU16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitWriteStorageS16(EmitContext&) {
+void EmitWriteStorageS16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitWriteStorage32(EmitContext& ctx, const IR::Value& binding,
+void EmitWriteStorage32(EmitContext& ctx, const IR::Value& binding,
const IR::Value& offset, Id value) {
if (!binding.IsImmediate()) {
throw NotImplementedException("Dynamic storage buffer indexing");
@@ -140,11 +140,11 @@ void EmitSPIRV::EmitWriteStorage32(EmitContext& ctx, const IR::Value& binding,
ctx.OpStore(pointer, value);
}
-void EmitSPIRV::EmitWriteStorage64(EmitContext&) {
+void EmitWriteStorage64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitWriteStorage128(EmitContext&) {
+void EmitWriteStorage128(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_select.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_select.cpp
index 40a856f72..8d5062724 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_select.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_select.cpp
@@ -6,19 +6,19 @@
namespace Shader::Backend::SPIRV {
-void EmitSPIRV::EmitSelect8(EmitContext&) {
+void EmitSelect8(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitSelect16(EmitContext&) {
+void EmitSelect16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitSelect32(EmitContext&) {
+void EmitSelect32(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-void EmitSPIRV::EmitSelect64(EmitContext&) {
+void EmitSelect64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
diff --git a/src/shader_recompiler/backend/spirv/emit_spirv_undefined.cpp b/src/shader_recompiler/backend/spirv/emit_spirv_undefined.cpp
index c1ed8f281..19b06dbe4 100644
--- a/src/shader_recompiler/backend/spirv/emit_spirv_undefined.cpp
+++ b/src/shader_recompiler/backend/spirv/emit_spirv_undefined.cpp
@@ -6,23 +6,23 @@
namespace Shader::Backend::SPIRV {
-Id EmitSPIRV::EmitUndefU1(EmitContext& ctx) {
+Id EmitUndefU1(EmitContext& ctx) {
return ctx.OpUndef(ctx.U1);
}
-Id EmitSPIRV::EmitUndefU8(EmitContext&) {
+Id EmitUndefU8(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-Id EmitSPIRV::EmitUndefU16(EmitContext&) {
+Id EmitUndefU16(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
-Id EmitSPIRV::EmitUndefU32(EmitContext& ctx) {
+Id EmitUndefU32(EmitContext& ctx) {
return ctx.OpUndef(ctx.U32[1]);
}
-Id EmitSPIRV::EmitUndefU64(EmitContext&) {
+Id EmitUndefU64(EmitContext&) {
throw NotImplementedException("SPIR-V Instruction");
}
diff --git a/src/shader_recompiler/environment.h b/src/shader_recompiler/environment.h
index f6230e817..0ba681fb9 100644
--- a/src/shader_recompiler/environment.h
+++ b/src/shader_recompiler/environment.h
@@ -1,5 +1,7 @@
#pragma once
+#include <array>
+
#include "common/common_types.h"
namespace Shader {
@@ -8,7 +10,9 @@ class Environment {
public:
virtual ~Environment() = default;
- [[nodiscard]] virtual u64 ReadInstruction(u32 address) const = 0;
+ [[nodiscard]] virtual u64 ReadInstruction(u32 address) = 0;
+
+ [[nodiscard]] virtual std::array<u32, 3> WorkgroupSize() = 0;
};
} // namespace Shader
diff --git a/src/shader_recompiler/file_environment.cpp b/src/shader_recompiler/file_environment.cpp
index b34bf462b..5127523f9 100644
--- a/src/shader_recompiler/file_environment.cpp
+++ b/src/shader_recompiler/file_environment.cpp
@@ -29,7 +29,7 @@ FileEnvironment::FileEnvironment(const char* path) {
FileEnvironment::~FileEnvironment() = default;
-u64 FileEnvironment::ReadInstruction(u32 offset) const {
+u64 FileEnvironment::ReadInstruction(u32 offset) {
if (offset % 8 != 0) {
throw InvalidArgument("offset={} is not aligned to 8", offset);
}
@@ -39,4 +39,8 @@ u64 FileEnvironment::ReadInstruction(u32 offset) const {
return data[offset / 8];
}
+std::array<u32, 3> FileEnvironment::WorkgroupSize() {
+ return {1, 1, 1};
+}
+
} // namespace Shader
diff --git a/src/shader_recompiler/file_environment.h b/src/shader_recompiler/file_environment.h
index c294bc6fa..b8c4bbadd 100644
--- a/src/shader_recompiler/file_environment.h
+++ b/src/shader_recompiler/file_environment.h
@@ -12,7 +12,9 @@ public:
explicit FileEnvironment(const char* path);
~FileEnvironment() override;
- u64 ReadInstruction(u32 offset) const override;
+ u64 ReadInstruction(u32 offset) override;
+
+ std::array<u32, 3> WorkgroupSize() override;
private:
std::vector<u64> data;
diff --git a/src/shader_recompiler/frontend/ir/basic_block.cpp b/src/shader_recompiler/frontend/ir/basic_block.cpp
index 5ae91dd7d..ec029dfd6 100644
--- a/src/shader_recompiler/frontend/ir/basic_block.cpp
+++ b/src/shader_recompiler/frontend/ir/basic_block.cpp
@@ -127,6 +127,8 @@ static std::string ArgToIndex(const std::map<const Block*, size_t>& block_to_ind
return fmt::format("#{}", arg.U32());
case Type::U64:
return fmt::format("#{}", arg.U64());
+ case Type::F32:
+ return fmt::format("#{}", arg.F32());
case Type::Reg:
return fmt::format("{}", arg.Reg());
case Type::Pred:
diff --git a/src/shader_recompiler/frontend/ir/post_order.cpp b/src/shader_recompiler/frontend/ir/post_order.cpp
index a48b8dec5..8709a2ea1 100644
--- a/src/shader_recompiler/frontend/ir/post_order.cpp
+++ b/src/shader_recompiler/frontend/ir/post_order.cpp
@@ -28,7 +28,7 @@ BlockList PostOrder(const BlockList& blocks) {
if (!visited.insert(branch).second) {
return false;
}
- // Calling push_back twice is faster than insert on msvc
+ // Calling push_back twice is faster than insert on MSVC
block_stack.push_back(block);
block_stack.push_back(branch);
return true;
diff --git a/src/shader_recompiler/frontend/maxwell/program.cpp b/src/shader_recompiler/frontend/maxwell/program.cpp
index 8331d576c..8c44ebb29 100644
--- a/src/shader_recompiler/frontend/maxwell/program.cpp
+++ b/src/shader_recompiler/frontend/maxwell/program.cpp
@@ -69,7 +69,7 @@ IR::Program TranslateProgram(ObjectPool<IR::Inst>& inst_pool, ObjectPool<IR::Blo
Optimization::VerificationPass(function);
}
Optimization::CollectShaderInfoPass(program);
- //*/
+ fmt::print(stdout, "{}\n", IR::DumpProgram(program));
return program;
}
diff --git a/src/shader_recompiler/frontend/maxwell/translate/impl/impl.cpp b/src/shader_recompiler/frontend/maxwell/translate/impl/impl.cpp
index 3c9eaddd9..079e3497f 100644
--- a/src/shader_recompiler/frontend/maxwell/translate/impl/impl.cpp
+++ b/src/shader_recompiler/frontend/maxwell/translate/impl/impl.cpp
@@ -24,6 +24,14 @@ void TranslatorVisitor::F(IR::Reg dest_reg, const IR::F32& value) {
X(dest_reg, ir.BitCast<IR::U32>(value));
}
+IR::U32 TranslatorVisitor::GetReg8(u64 insn) {
+ union {
+ u64 raw;
+ BitField<8, 8, IR::Reg> index;
+ } const reg{insn};
+ return X(reg.index);
+}
+
IR::U32 TranslatorVisitor::GetReg20(u64 insn) {
union {
u64 raw;
diff --git a/src/shader_recompiler/frontend/maxwell/translate/impl/impl.h b/src/shader_recompiler/frontend/maxwell/translate/impl/impl.h
index b701605d7..8bd468244 100644
--- a/src/shader_recompiler/frontend/maxwell/translate/impl/impl.h
+++ b/src/shader_recompiler/frontend/maxwell/translate/impl/impl.h
@@ -301,6 +301,7 @@ public:
void X(IR::Reg dest_reg, const IR::U32& value);
void F(IR::Reg dest_reg, const IR::F32& value);
+ [[nodiscard]] IR::U32 GetReg8(u64 insn);
[[nodiscard]] IR::U32 GetReg20(u64 insn);
[[nodiscard]] IR::U32 GetReg39(u64 insn);
[[nodiscard]] IR::F32 GetReg20F(u64 insn);
diff --git a/src/shader_recompiler/frontend/maxwell/translate/impl/move_register.cpp b/src/shader_recompiler/frontend/maxwell/translate/impl/move_register.cpp
index 1f83d1068..c3c4b9abd 100644
--- a/src/shader_recompiler/frontend/maxwell/translate/impl/move_register.cpp
+++ b/src/shader_recompiler/frontend/maxwell/translate/impl/move_register.cpp
@@ -10,36 +10,35 @@
namespace Shader::Maxwell {
namespace {
-union MOV {
- u64 raw;
- BitField<0, 8, IR::Reg> dest_reg;
- BitField<20, 8, IR::Reg> src_reg;
- BitField<39, 4, u64> mask;
-};
-
-void CheckMask(MOV mov) {
- if (mov.mask != 0xf) {
+void MOV(TranslatorVisitor& v, u64 insn, const IR::U32& src, bool is_mov32i = false) {
+ union {
+ u64 raw;
+ BitField<0, 8, IR::Reg> dest_reg;
+ BitField<39, 4, u64> mask;
+ BitField<12, 4, u64> mov32i_mask;
+ } const mov{insn};
+
+ if ((is_mov32i ? mov.mov32i_mask : mov.mask) != 0xf) {
throw NotImplementedException("Non-full move mask");
}
+ v.X(mov.dest_reg, src);
}
} // Anonymous namespace
void TranslatorVisitor::MOV_reg(u64 insn) {
- const MOV mov{insn};
- CheckMask(mov);
- X(mov.dest_reg, X(mov.src_reg));
+ MOV(*this, insn, GetReg8(insn));
}
void TranslatorVisitor::MOV_cbuf(u64 insn) {
- const MOV mov{insn};
- CheckMask(mov);
- X(mov.dest_reg, GetCbuf(insn));
+ MOV(*this, insn, GetCbuf(insn));
}
void TranslatorVisitor::MOV_imm(u64 insn) {
- const MOV mov{insn};
- CheckMask(mov);
- X(mov.dest_reg, GetImm20(insn));
+ MOV(*this, insn, GetImm20(insn));
+}
+
+void TranslatorVisitor::MOV32I(u64 insn) {
+ MOV(*this, insn, GetImm32(insn), true);
}
} // namespace Shader::Maxwell
diff --git a/src/shader_recompiler/frontend/maxwell/translate/impl/not_implemented.cpp b/src/shader_recompiler/frontend/maxwell/translate/impl/not_implemented.cpp
index 1bb160acb..6b2a1356b 100644
--- a/src/shader_recompiler/frontend/maxwell/translate/impl/not_implemented.cpp
+++ b/src/shader_recompiler/frontend/maxwell/translate/impl/not_implemented.cpp
@@ -617,10 +617,6 @@ void TranslatorVisitor::MEMBAR(u64) {
ThrowNotImplemented(Opcode::MEMBAR);
}
-void TranslatorVisitor::MOV32I(u64) {
- ThrowNotImplemented(Opcode::MOV32I);
-}
-
void TranslatorVisitor::NOP(u64) {
ThrowNotImplemented(Opcode::NOP);
}
diff --git a/src/shader_recompiler/main.cpp b/src/shader_recompiler/main.cpp
index 1610bb34e..050a37f18 100644
--- a/src/shader_recompiler/main.cpp
+++ b/src/shader_recompiler/main.cpp
@@ -76,5 +76,5 @@ int main() {
fmt::print(stdout, "{}\n", cfg.Dot());
IR::Program program{TranslateProgram(inst_pool, block_pool, env, cfg)};
fmt::print(stdout, "{}\n", IR::DumpProgram(program));
- Backend::SPIRV::EmitSPIRV spirv{program};
+ void(Backend::SPIRV::EmitSPIRV(env, program));
}
diff --git a/src/shader_recompiler/profile.h b/src/shader_recompiler/profile.h
new file mode 100644
index 000000000..c96d783b7
--- /dev/null
+++ b/src/shader_recompiler/profile.h
@@ -0,0 +1,13 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+namespace Shader {
+
+struct Profile {
+ bool unified_descriptor_binding;
+};
+
+} // namespace Shader
diff --git a/src/shader_recompiler/recompiler.cpp b/src/shader_recompiler/recompiler.cpp
new file mode 100644
index 000000000..b25081e39
--- /dev/null
+++ b/src/shader_recompiler/recompiler.cpp
@@ -0,0 +1,27 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <vector>
+
+#include "common/common_types.h"
+#include "shader_recompiler/backend/spirv/emit_spirv.h"
+#include "shader_recompiler/environment.h"
+#include "shader_recompiler/frontend/maxwell/control_flow.h"
+#include "shader_recompiler/frontend/maxwell/program.h"
+#include "shader_recompiler/object_pool.h"
+#include "shader_recompiler/recompiler.h"
+
+namespace Shader {
+
+std::pair<Info, std::vector<u32>> RecompileSPIRV(Environment& env, u32 start_address) {
+ ObjectPool<Maxwell::Flow::Block> flow_block_pool;
+ ObjectPool<IR::Inst> inst_pool;
+ ObjectPool<IR::Block> block_pool;
+
+ Maxwell::Flow::CFG cfg{env, flow_block_pool, start_address};
+ IR::Program program{Maxwell::TranslateProgram(inst_pool, block_pool, env, cfg)};
+ return {std::move(program.info), Backend::SPIRV::EmitSPIRV(env, program)};
+}
+
+} // namespace Shader
diff --git a/src/shader_recompiler/recompiler.h b/src/shader_recompiler/recompiler.h
new file mode 100644
index 000000000..4cb973878
--- /dev/null
+++ b/src/shader_recompiler/recompiler.h
@@ -0,0 +1,18 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <utility>
+#include <vector>
+
+#include "common/common_types.h"
+#include "shader_recompiler/environment.h"
+#include "shader_recompiler/shader_info.h"
+
+namespace Shader {
+
+[[nodiscard]] std::pair<Info, std::vector<u32>> RecompileSPIRV(Environment& env, u32 start_address);
+
+} // namespace Shader
diff --git a/src/video_core/CMakeLists.txt b/src/video_core/CMakeLists.txt
index c5ce71706..3323e6916 100644
--- a/src/video_core/CMakeLists.txt
+++ b/src/video_core/CMakeLists.txt
@@ -43,9 +43,6 @@ add_library(video_core STATIC
engines/maxwell_3d.h
engines/maxwell_dma.cpp
engines/maxwell_dma.h
- engines/shader_bytecode.h
- engines/shader_header.h
- engines/shader_type.h
framebuffer_config.h
macro/macro.cpp
macro/macro.h
@@ -123,6 +120,7 @@ add_library(video_core STATIC
renderer_vulkan/vk_master_semaphore.h
renderer_vulkan/vk_pipeline_cache.cpp
renderer_vulkan/vk_pipeline_cache.h
+ renderer_vulkan/vk_pipeline.h
renderer_vulkan/vk_query_cache.cpp
renderer_vulkan/vk_query_cache.h
renderer_vulkan/vk_rasterizer.cpp
@@ -201,7 +199,7 @@ add_library(video_core STATIC
create_target_directory_groups(video_core)
target_link_libraries(video_core PUBLIC common core)
-target_link_libraries(video_core PRIVATE glad xbyak)
+target_link_libraries(video_core PRIVATE glad shader_recompiler xbyak)
if (YUZU_USE_BUNDLED_FFMPEG AND NOT WIN32)
add_dependencies(video_core ffmpeg-build)
diff --git a/src/video_core/engines/kepler_compute.h b/src/video_core/engines/kepler_compute.h
index 0d7683c2d..f8b8d06ac 100644
--- a/src/video_core/engines/kepler_compute.h
+++ b/src/video_core/engines/kepler_compute.h
@@ -12,7 +12,6 @@
#include "common/common_types.h"
#include "video_core/engines/engine_interface.h"
#include "video_core/engines/engine_upload.h"
-#include "video_core/engines/shader_type.h"
#include "video_core/gpu.h"
#include "video_core/textures/texture.h"
diff --git a/src/video_core/engines/shader_bytecode.h b/src/video_core/engines/shader_bytecode.h
deleted file mode 100644
index 8b45f1b62..000000000
--- a/src/video_core/engines/shader_bytecode.h
+++ /dev/null
@@ -1,2298 +0,0 @@
-// Copyright 2018 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <array>
-#include <bitset>
-#include <optional>
-#include <tuple>
-#include <vector>
-
-#include "common/assert.h"
-#include "common/bit_field.h"
-#include "common/common_types.h"
-
-namespace Tegra::Shader {
-
-struct Register {
- /// Number of registers
- static constexpr std::size_t NumRegisters = 256;
-
- /// Register 255 is special cased to always be 0
- static constexpr std::size_t ZeroIndex = 255;
-
- enum class Size : u64 {
- Byte = 0,
- Short = 1,
- Word = 2,
- Long = 3,
- };
-
- constexpr Register() = default;
-
- constexpr Register(u64 value_) : value(value_) {}
-
- [[nodiscard]] constexpr operator u64() const {
- return value;
- }
-
- template <typename T>
- [[nodiscard]] constexpr u64 operator-(const T& oth) const {
- return value - oth;
- }
-
- template <typename T>
- [[nodiscard]] constexpr u64 operator&(const T& oth) const {
- return value & oth;
- }
-
- [[nodiscard]] constexpr u64 operator&(const Register& oth) const {
- return value & oth.value;
- }
-
- [[nodiscard]] constexpr u64 operator~() const {
- return ~value;
- }
-
- [[nodiscard]] u64 GetSwizzledIndex(u64 elem) const {
- elem = (value + elem) & 3;
- return (value & ~3) + elem;
- }
-
-private:
- u64 value{};
-};
-
-enum class AttributeSize : u64 {
- Word = 0,
- DoubleWord = 1,
- TripleWord = 2,
- QuadWord = 3,
-};
-
-union Attribute {
- Attribute() = default;
-
- constexpr explicit Attribute(u64 value_) : value(value_) {}
-
- enum class Index : u64 {
- LayerViewportPointSize = 6,
- Position = 7,
- Attribute_0 = 8,
- Attribute_31 = 39,
- FrontColor = 40,
- FrontSecondaryColor = 41,
- BackColor = 42,
- BackSecondaryColor = 43,
- ClipDistances0123 = 44,
- ClipDistances4567 = 45,
- PointCoord = 46,
- // This attribute contains a tuple of (~, ~, InstanceId, VertexId) when inside a vertex
- // shader, and a tuple of (TessCoord.x, TessCoord.y, TessCoord.z, ~) when inside a Tess Eval
- // shader.
- TessCoordInstanceIDVertexID = 47,
- TexCoord_0 = 48,
- TexCoord_7 = 55,
- // This attribute contains a tuple of (Unk, Unk, Unk, gl_FrontFacing) when inside a fragment
- // shader. It is unknown what the other values contain.
- FrontFacing = 63,
- };
-
- union {
- BitField<20, 10, u64> immediate;
- BitField<22, 2, u64> element;
- BitField<24, 6, Index> index;
- BitField<31, 1, u64> patch;
- BitField<47, 3, AttributeSize> size;
-
- [[nodiscard]] bool IsPhysical() const {
- return patch == 0 && element == 0 && static_cast<u64>(index.Value()) == 0;
- }
- } fmt20;
-
- union {
- BitField<30, 2, u64> element;
- BitField<32, 6, Index> index;
- } fmt28;
-
- BitField<39, 8, u64> reg;
- u64 value{};
-};
-
-union Sampler {
- Sampler() = default;
-
- constexpr explicit Sampler(u64 value_) : value(value_) {}
-
- enum class Index : u64 {
- Sampler_0 = 8,
- };
-
- BitField<36, 13, Index> index;
- u64 value{};
-};
-
-union Image {
- Image() = default;
-
- constexpr explicit Image(u64 value_) : value{value_} {}
-
- BitField<36, 13, u64> index;
- u64 value;
-};
-
-} // namespace Tegra::Shader
-
-namespace std {
-
-// TODO(bunnei): The below is forbidden by the C++ standard, but works fine. See #330.
-template <>
-struct make_unsigned<Tegra::Shader::Attribute> {
- using type = Tegra::Shader::Attribute;
-};
-
-template <>
-struct make_unsigned<Tegra::Shader::Register> {
- using type = Tegra::Shader::Register;
-};
-
-} // namespace std
-
-namespace Tegra::Shader {
-
-enum class Pred : u64 {
- UnusedIndex = 0x7,
- NeverExecute = 0xF,
-};
-
-enum class PredCondition : u64 {
- F = 0, // Always false
- LT = 1, // Ordered less than
- EQ = 2, // Ordered equal
- LE = 3, // Ordered less than or equal
- GT = 4, // Ordered greater than
- NE = 5, // Ordered not equal
- GE = 6, // Ordered greater than or equal
- NUM = 7, // Ordered
- NAN_ = 8, // Unordered
- LTU = 9, // Unordered less than
- EQU = 10, // Unordered equal
- LEU = 11, // Unordered less than or equal
- GTU = 12, // Unordered greater than
- NEU = 13, // Unordered not equal
- GEU = 14, // Unordered greater than or equal
- T = 15, // Always true
-};
-
-enum class PredOperation : u64 {
- And = 0,
- Or = 1,
- Xor = 2,
-};
-
-enum class LogicOperation : u64 {
- And = 0,
- Or = 1,
- Xor = 2,
- PassB = 3,
-};
-
-enum class SubOp : u64 {
- Cos = 0x0,
- Sin = 0x1,
- Ex2 = 0x2,
- Lg2 = 0x3,
- Rcp = 0x4,
- Rsq = 0x5,
- Sqrt = 0x8,
-};
-
-enum class F2iRoundingOp : u64 {
- RoundEven = 0,
- Floor = 1,
- Ceil = 2,
- Trunc = 3,
-};
-
-enum class F2fRoundingOp : u64 {
- None = 0,
- Pass = 3,
- Round = 8,
- Floor = 9,
- Ceil = 10,
- Trunc = 11,
-};
-
-enum class AtomicOp : u64 {
- Add = 0,
- Min = 1,
- Max = 2,
- Inc = 3,
- Dec = 4,
- And = 5,
- Or = 6,
- Xor = 7,
- Exch = 8,
- SafeAdd = 10,
-};
-
-enum class GlobalAtomicType : u64 {
- U32 = 0,
- S32 = 1,
- U64 = 2,
- F32_FTZ_RN = 3,
- F16x2_FTZ_RN = 4,
- S64 = 5,
-};
-
-enum class UniformType : u64 {
- UnsignedByte = 0,
- SignedByte = 1,
- UnsignedShort = 2,
- SignedShort = 3,
- Single = 4,
- Double = 5,
- Quad = 6,
- UnsignedQuad = 7,
-};
-
-enum class StoreType : u64 {
- Unsigned8 = 0,
- Signed8 = 1,
- Unsigned16 = 2,
- Signed16 = 3,
- Bits32 = 4,
- Bits64 = 5,
- Bits128 = 6,
-};
-
-enum class AtomicType : u64 {
- U32 = 0,
- S32 = 1,
- U64 = 2,
- S64 = 3,
-};
-
-enum class IMinMaxExchange : u64 {
- None = 0,
- XLo = 1,
- XMed = 2,
- XHi = 3,
-};
-
-enum class VideoType : u64 {
- Size16_Low = 0,
- Size16_High = 1,
- Size32 = 2,
- Invalid = 3,
-};
-
-enum class VmadShr : u64 {
- Shr7 = 1,
- Shr15 = 2,
-};
-
-enum class VmnmxType : u64 {
- Bits8,
- Bits16,
- Bits32,
-};
-
-enum class VmnmxOperation : u64 {
- Mrg_16H = 0,
- Mrg_16L = 1,
- Mrg_8B0 = 2,
- Mrg_8B2 = 3,
- Acc = 4,
- Min = 5,
- Max = 6,
- Nop = 7,
-};
-
-enum class XmadMode : u64 {
- None = 0,
- CLo = 1,
- CHi = 2,
- CSfu = 3,
- CBcc = 4,
-};
-
-enum class IAdd3Mode : u64 {
- None = 0,
- RightShift = 1,
- LeftShift = 2,
-};
-
-enum class IAdd3Height : u64 {
- None = 0,
- LowerHalfWord = 1,
- UpperHalfWord = 2,
-};
-
-enum class FlowCondition : u64 {
- Always = 0xF,
- Fcsm_Tr = 0x1C, // TODO(bunnei): What is this used for?
-};
-
-enum class ConditionCode : u64 {
- F = 0,
- LT = 1,
- EQ = 2,
- LE = 3,
- GT = 4,
- NE = 5,
- GE = 6,
- Num = 7,
- Nan = 8,
- LTU = 9,
- EQU = 10,
- LEU = 11,
- GTU = 12,
- NEU = 13,
- GEU = 14,
- T = 15,
- OFF = 16,
- LO = 17,
- SFF = 18,
- LS = 19,
- HI = 20,
- SFT = 21,
- HS = 22,
- OFT = 23,
- CSM_TA = 24,
- CSM_TR = 25,
- CSM_MX = 26,
- FCSM_TA = 27,
- FCSM_TR = 28,
- FCSM_MX = 29,
- RLE = 30,
- RGT = 31,
-};
-
-enum class PredicateResultMode : u64 {
- None = 0x0,
- NotZero = 0x3,
-};
-
-enum class TextureType : u64 {
- Texture1D = 0,
- Texture2D = 1,
- Texture3D = 2,
- TextureCube = 3,
-};
-
-enum class TextureQueryType : u64 {
- Dimension = 1,
- TextureType = 2,
- SamplePosition = 5,
- Filter = 16,
- LevelOfDetail = 18,
- Wrap = 20,
- BorderColor = 22,
-};
-
-enum class TextureProcessMode : u64 {
- None = 0,
- LZ = 1, // Load LOD of zero.
- LB = 2, // Load Bias.
- LL = 3, // Load LOD.
- LBA = 6, // Load Bias. The A is unknown, does not appear to differ with LB.
- LLA = 7 // Load LOD. The A is unknown, does not appear to differ with LL.
-};
-
-enum class TextureMiscMode : u64 {
- DC,
- AOFFI, // Uses Offset
- NDV,
- NODEP,
- MZ,
- PTP,
-};
-
-enum class SurfaceDataMode : u64 {
- P = 0,
- D_BA = 1,
-};
-
-enum class OutOfBoundsStore : u64 {
- Ignore = 0,
- Clamp = 1,
- Trap = 2,
-};
-
-enum class ImageType : u64 {
- Texture1D = 0,
- TextureBuffer = 1,
- Texture1DArray = 2,
- Texture2D = 3,
- Texture2DArray = 4,
- Texture3D = 5,
-};
-
-enum class IsberdMode : u64 {
- None = 0,
- Patch = 1,
- Prim = 2,
- Attr = 3,
-};
-
-enum class IsberdShift : u64 { None = 0, U16 = 1, B32 = 2 };
-
-enum class MembarType : u64 {
- CTA = 0,
- GL = 1,
- SYS = 2,
- VC = 3,
-};
-
-enum class MembarUnknown : u64 { Default = 0, IVALLD = 1, IVALLT = 2, IVALLTD = 3 };
-
-enum class HalfType : u64 {
- H0_H1 = 0,
- F32 = 1,
- H0_H0 = 2,
- H1_H1 = 3,
-};
-
-enum class HalfMerge : u64 {
- H0_H1 = 0,
- F32 = 1,
- Mrg_H0 = 2,
- Mrg_H1 = 3,
-};
-
-enum class HalfPrecision : u64 {
- None = 0,
- FTZ = 1,
- FMZ = 2,
-};
-
-enum class R2pMode : u64 {
- Pr = 0,
- Cc = 1,
-};
-
-enum class IpaInterpMode : u64 {
- Pass = 0,
- Multiply = 1,
- Constant = 2,
- Sc = 3,
-};
-
-enum class IpaSampleMode : u64 {
- Default = 0,
- Centroid = 1,
- Offset = 2,
-};
-
-enum class LmemLoadCacheManagement : u64 {
- Default = 0,
- LU = 1,
- CI = 2,
- CV = 3,
-};
-
-enum class StoreCacheManagement : u64 {
- Default = 0,
- CG = 1,
- CS = 2,
- WT = 3,
-};
-
-struct IpaMode {
- IpaInterpMode interpolation_mode;
- IpaSampleMode sampling_mode;
-
- [[nodiscard]] bool operator==(const IpaMode& a) const {
- return std::tie(interpolation_mode, sampling_mode) ==
- std::tie(a.interpolation_mode, a.sampling_mode);
- }
- [[nodiscard]] bool operator!=(const IpaMode& a) const {
- return !operator==(a);
- }
- [[nodiscard]] bool operator<(const IpaMode& a) const {
- return std::tie(interpolation_mode, sampling_mode) <
- std::tie(a.interpolation_mode, a.sampling_mode);
- }
-};
-
-enum class SystemVariable : u64 {
- LaneId = 0x00,
- VirtCfg = 0x02,
- VirtId = 0x03,
- Pm0 = 0x04,
- Pm1 = 0x05,
- Pm2 = 0x06,
- Pm3 = 0x07,
- Pm4 = 0x08,
- Pm5 = 0x09,
- Pm6 = 0x0a,
- Pm7 = 0x0b,
- OrderingTicket = 0x0f,
- PrimType = 0x10,
- InvocationId = 0x11,
- Ydirection = 0x12,
- ThreadKill = 0x13,
- ShaderType = 0x14,
- DirectBeWriteAddressLow = 0x15,
- DirectBeWriteAddressHigh = 0x16,
- DirectBeWriteEnabled = 0x17,
- MachineId0 = 0x18,
- MachineId1 = 0x19,
- MachineId2 = 0x1a,
- MachineId3 = 0x1b,
- Affinity = 0x1c,
- InvocationInfo = 0x1d,
- WscaleFactorXY = 0x1e,
- WscaleFactorZ = 0x1f,
- Tid = 0x20,
- TidX = 0x21,
- TidY = 0x22,
- TidZ = 0x23,
- CtaParam = 0x24,
- CtaIdX = 0x25,
- CtaIdY = 0x26,
- CtaIdZ = 0x27,
- NtId = 0x28,
- CirQueueIncrMinusOne = 0x29,
- Nlatc = 0x2a,
- SmSpaVersion = 0x2c,
- MultiPassShaderInfo = 0x2d,
- LwinHi = 0x2e,
- SwinHi = 0x2f,
- SwinLo = 0x30,
- SwinSz = 0x31,
- SmemSz = 0x32,
- SmemBanks = 0x33,
- LwinLo = 0x34,
- LwinSz = 0x35,
- LmemLosz = 0x36,
- LmemHioff = 0x37,
- EqMask = 0x38,
- LtMask = 0x39,
- LeMask = 0x3a,
- GtMask = 0x3b,
- GeMask = 0x3c,
- RegAlloc = 0x3d,
- CtxAddr = 0x3e, // .fmask = F_SM50
- BarrierAlloc = 0x3e, // .fmask = F_SM60
- GlobalErrorStatus = 0x40,
- WarpErrorStatus = 0x42,
- WarpErrorStatusClear = 0x43,
- PmHi0 = 0x48,
- PmHi1 = 0x49,
- PmHi2 = 0x4a,
- PmHi3 = 0x4b,
- PmHi4 = 0x4c,
- PmHi5 = 0x4d,
- PmHi6 = 0x4e,
- PmHi7 = 0x4f,
- ClockLo = 0x50,
- ClockHi = 0x51,
- GlobalTimerLo = 0x52,
- GlobalTimerHi = 0x53,
- HwTaskId = 0x60,
- CircularQueueEntryIndex = 0x61,
- CircularQueueEntryAddressLow = 0x62,
- CircularQueueEntryAddressHigh = 0x63,
-};
-
-enum class PhysicalAttributeDirection : u64 {
- Input = 0,
- Output = 1,
-};
-
-enum class VoteOperation : u64 {
- All = 0, // allThreadsNV
- Any = 1, // anyThreadNV
- Eq = 2, // allThreadsEqualNV
-};
-
-enum class ImageAtomicOperationType : u64 {
- U32 = 0,
- S32 = 1,
- U64 = 2,
- F32 = 3,
- S64 = 5,
- SD32 = 6,
- SD64 = 7,
-};
-
-enum class ImageAtomicOperation : u64 {
- Add = 0,
- Min = 1,
- Max = 2,
- Inc = 3,
- Dec = 4,
- And = 5,
- Or = 6,
- Xor = 7,
- Exch = 8,
-};
-
-enum class ShuffleOperation : u64 {
- Idx = 0, // shuffleNV
- Up = 1, // shuffleUpNV
- Down = 2, // shuffleDownNV
- Bfly = 3, // shuffleXorNV
-};
-
-enum class ShfType : u64 {
- Bits32 = 0,
- U64 = 2,
- S64 = 3,
-};
-
-enum class ShfXmode : u64 {
- None = 0,
- HI = 1,
- X = 2,
- XHI = 3,
-};
-
-union Instruction {
- constexpr Instruction& operator=(const Instruction& instr) {
- value = instr.value;
- return *this;
- }
-
- constexpr Instruction(u64 value_) : value{value_} {}
- constexpr Instruction(const Instruction& instr) : value(instr.value) {}
-
- [[nodiscard]] constexpr bool Bit(u64 offset) const {
- return ((value >> offset) & 1) != 0;
- }
-
- BitField<0, 8, Register> gpr0;
- BitField<8, 8, Register> gpr8;
- union {
- BitField<16, 4, Pred> full_pred;
- BitField<16, 3, u64> pred_index;
- } pred;
- BitField<19, 1, u64> negate_pred;
- BitField<20, 8, Register> gpr20;
- BitField<20, 4, SubOp> sub_op;
- BitField<28, 8, Register> gpr28;
- BitField<39, 8, Register> gpr39;
- BitField<48, 16, u64> opcode;
-
- union {
- BitField<8, 5, ConditionCode> cc;
- BitField<13, 1, u64> trigger;
- } nop;
-
- union {
- BitField<48, 2, VoteOperation> operation;
- BitField<45, 3, u64> dest_pred;
- BitField<39, 3, u64> value;
- BitField<42, 1, u64> negate_value;
- } vote;
-
- union {
- BitField<30, 2, ShuffleOperation> operation;
- BitField<48, 3, u64> pred48;
- BitField<28, 1, u64> is_index_imm;
- BitField<29, 1, u64> is_mask_imm;
- BitField<20, 5, u64> index_imm;
- BitField<34, 13, u64> mask_imm;
- } shfl;
-
- union {
- BitField<44, 1, u64> ftz;
- BitField<39, 2, u64> tab5cb8_2;
- BitField<38, 1, u64> ndv;
- BitField<47, 1, u64> cc;
- BitField<28, 8, u64> swizzle;
- } fswzadd;
-
- union {
- BitField<8, 8, Register> gpr;
- BitField<20, 24, s64> offset;
- } gmem;
-
- union {
- BitField<20, 16, u64> imm20_16;
- BitField<20, 19, u64> imm20_19;
- BitField<20, 32, s64> imm20_32;
- BitField<45, 1, u64> negate_b;
- BitField<46, 1, u64> abs_a;
- BitField<48, 1, u64> negate_a;
- BitField<49, 1, u64> abs_b;
- BitField<50, 1, u64> saturate_d;
- BitField<56, 1, u64> negate_imm;
-
- union {
- BitField<39, 3, u64> pred;
- BitField<42, 1, u64> negate_pred;
- } fmnmx;
-
- union {
- BitField<39, 1, u64> invert_a;
- BitField<40, 1, u64> invert_b;
- BitField<41, 2, LogicOperation> operation;
- BitField<44, 2, PredicateResultMode> pred_result_mode;
- BitField<48, 3, Pred> pred48;
- } lop;
-
- union {
- BitField<53, 2, LogicOperation> operation;
- BitField<55, 1, u64> invert_a;
- BitField<56, 1, u64> invert_b;
- } lop32i;
-
- union {
- BitField<28, 8, u64> imm_lut28;
- BitField<48, 8, u64> imm_lut48;
-
- [[nodiscard]] u32 GetImmLut28() const {
- return static_cast<u32>(imm_lut28);
- }
-
- [[nodiscard]] u32 GetImmLut48() const {
- return static_cast<u32>(imm_lut48);
- }
- } lop3;
-
- [[nodiscard]] u16 GetImm20_16() const {
- return static_cast<u16>(imm20_16);
- }
-
- [[nodiscard]] u32 GetImm20_19() const {
- u32 imm{static_cast<u32>(imm20_19)};
- imm <<= 12;
- imm |= negate_imm ? 0x80000000 : 0;
- return imm;
- }
-
- [[nodiscard]] u32 GetImm20_32() const {
- return static_cast<u32>(imm20_32);
- }
-
- [[nodiscard]] s32 GetSignedImm20_20() const {
- const auto immediate = static_cast<u32>(imm20_19 | (negate_imm << 19));
- // Sign extend the 20-bit value.
- const auto mask = 1U << (20 - 1);
- return static_cast<s32>((immediate ^ mask) - mask);
- }
- } alu;
-
- union {
- BitField<38, 1, u64> idx;
- BitField<51, 1, u64> saturate;
- BitField<52, 2, IpaSampleMode> sample_mode;
- BitField<54, 2, IpaInterpMode> interp_mode;
- } ipa;
-
- union {
- BitField<39, 2, u64> tab5cb8_2;
- BitField<41, 3, u64> postfactor;
- BitField<44, 2, u64> tab5c68_0;
- BitField<48, 1, u64> negate_b;
- } fmul;
-
- union {
- BitField<55, 1, u64> saturate;
- } fmul32;
-
- union {
- BitField<52, 1, u64> generates_cc;
- } op_32;
-
- union {
- BitField<48, 1, u64> is_signed;
- } shift;
-
- union {
- BitField<39, 1, u64> wrap;
- } shr;
-
- union {
- BitField<37, 2, ShfType> type;
- BitField<48, 2, ShfXmode> xmode;
- BitField<50, 1, u64> wrap;
- BitField<20, 6, u64> immediate;
- } shf;
-
- union {
- BitField<39, 5, u64> shift_amount;
- BitField<48, 1, u64> negate_b;
- BitField<49, 1, u64> negate_a;
- } alu_integer;
-
- union {
- BitField<43, 1, u64> x;
- } iadd;
-
- union {
- BitField<39, 1, u64> ftz;
- BitField<32, 1, u64> saturate;
- BitField<49, 2, HalfMerge> merge;
-
- BitField<44, 1, u64> abs_a;
- BitField<47, 2, HalfType> type_a;
-
- BitField<30, 1, u64> abs_b;
- BitField<28, 2, HalfType> type_b;
-
- BitField<35, 2, HalfType> type_c;
- } alu_half;
-
- union {
- BitField<39, 2, HalfPrecision> precision;
- BitField<39, 1, u64> ftz;
- BitField<52, 1, u64> saturate;
- BitField<49, 2, HalfMerge> merge;
-
- BitField<43, 1, u64> negate_a;
- BitField<44, 1, u64> abs_a;
- BitField<47, 2, HalfType> type_a;
- } alu_half_imm;
-
- union {
- BitField<29, 1, u64> first_negate;
- BitField<20, 9, u64> first;
-
- BitField<56, 1, u64> second_negate;
- BitField<30, 9, u64> second;
-
- [[nodiscard]] u32 PackImmediates() const {
- // Immediates are half floats shifted.
- constexpr u32 imm_shift = 6;
- return static_cast<u32>((first << imm_shift) | (second << (16 + imm_shift)));
- }
- } half_imm;
-
- union {
- union {
- BitField<37, 2, HalfPrecision> precision;
- BitField<32, 1, u64> saturate;
-
- BitField<31, 1, u64> negate_b;
- BitField<30, 1, u64> negate_c;
- BitField<35, 2, HalfType> type_c;
- } rr;
-
- BitField<57, 2, HalfPrecision> precision;
- BitField<52, 1, u64> saturate;
-
- BitField<49, 2, HalfMerge> merge;
-
- BitField<47, 2, HalfType> type_a;
-
- BitField<56, 1, u64> negate_b;
- BitField<28, 2, HalfType> type_b;
-
- BitField<51, 1, u64> negate_c;
- BitField<53, 2, HalfType> type_reg39;
- } hfma2;
-
- union {
- BitField<40, 1, u64> invert;
- } popc;
-
- union {
- BitField<41, 1, u64> sh;
- BitField<40, 1, u64> invert;
- BitField<48, 1, u64> is_signed;
- } flo;
-
- union {
- BitField<39, 3, u64> pred;
- BitField<42, 1, u64> neg_pred;
- } sel;
-
- union {
- BitField<39, 3, u64> pred;
- BitField<42, 1, u64> negate_pred;
- BitField<43, 2, IMinMaxExchange> exchange;
- BitField<48, 1, u64> is_signed;
- } imnmx;
-
- union {
- BitField<31, 2, IAdd3Height> height_c;
- BitField<33, 2, IAdd3Height> height_b;
- BitField<35, 2, IAdd3Height> height_a;
- BitField<37, 2, IAdd3Mode> mode;
- BitField<49, 1, u64> neg_c;
- BitField<50, 1, u64> neg_b;
- BitField<51, 1, u64> neg_a;
- } iadd3;
-
- union {
- BitField<54, 1, u64> saturate;
- BitField<56, 1, u64> negate_a;
- } iadd32i;
-
- union {
- BitField<53, 1, u64> negate_b;
- BitField<54, 1, u64> abs_a;
- BitField<56, 1, u64> negate_a;
- BitField<57, 1, u64> abs_b;
- } fadd32i;
-
- union {
- BitField<40, 1, u64> brev;
- BitField<47, 1, u64> rd_cc;
- BitField<48, 1, u64> is_signed;
- } bfe;
-
- union {
- BitField<48, 3, u64> pred48;
-
- union {
- BitField<20, 20, u64> entry_a;
- BitField<39, 5, u64> entry_b;
- BitField<45, 1, u64> neg;
- BitField<46, 1, u64> uses_cc;
- } imm;
-
- union {
- BitField<20, 14, u64> cb_index;
- BitField<34, 5, u64> cb_offset;
- BitField<56, 1, u64> neg;
- BitField<57, 1, u64> uses_cc;
- } hi;
-
- union {
- BitField<20, 14, u64> cb_index;
- BitField<34, 5, u64> cb_offset;
- BitField<39, 5, u64> entry_a;
- BitField<45, 1, u64> neg;
- BitField<46, 1, u64> uses_cc;
- } rz;
-
- union {
- BitField<39, 5, u64> entry_a;
- BitField<45, 1, u64> neg;
- BitField<46, 1, u64> uses_cc;
- } r1;
-
- union {
- BitField<28, 8, u64> entry_a;
- BitField<37, 1, u64> neg;
- BitField<38, 1, u64> uses_cc;
- } r2;
-
- } lea;
-
- union {
- BitField<0, 5, FlowCondition> cond;
- } flow;
-
- union {
- BitField<47, 1, u64> cc;
- BitField<48, 1, u64> negate_b;
- BitField<49, 1, u64> negate_c;
- BitField<51, 2, u64> tab5980_1;
- BitField<53, 2, u64> tab5980_0;
- } ffma;
-
- union {
- BitField<48, 3, UniformType> type;
- BitField<44, 2, u64> unknown;
- } ld_c;
-
- union {
- BitField<48, 3, StoreType> type;
- } ldst_sl;
-
- union {
- BitField<44, 2, u64> unknown;
- } ld_l;
-
- union {
- BitField<44, 2, StoreCacheManagement> cache_management;
- } st_l;
-
- union {
- BitField<48, 3, UniformType> type;
- BitField<46, 2, u64> cache_mode;
- } ldg;
-
- union {
- BitField<48, 3, UniformType> type;
- BitField<46, 2, u64> cache_mode;
- } stg;
-
- union {
- BitField<23, 3, AtomicOp> operation;
- BitField<48, 1, u64> extended;
- BitField<20, 3, GlobalAtomicType> type;
- } red;
-
- union {
- BitField<52, 4, AtomicOp> operation;
- BitField<49, 3, GlobalAtomicType> type;
- BitField<28, 20, s64> offset;
- } atom;
-
- union {
- BitField<52, 4, AtomicOp> operation;
- BitField<28, 2, AtomicType> type;
- BitField<30, 22, s64> offset;
-
- [[nodiscard]] s32 GetImmediateOffset() const {
- return static_cast<s32>(offset << 2);
- }
- } atoms;
-
- union {
- BitField<32, 1, PhysicalAttributeDirection> direction;
- BitField<47, 3, AttributeSize> size;
- BitField<20, 11, u64> address;
- } al2p;
-
- union {
- BitField<53, 3, UniformType> type;
- BitField<52, 1, u64> extended;
- } generic;
-
- union {
- BitField<0, 3, u64> pred0;
- BitField<3, 3, u64> pred3;
- BitField<6, 1, u64> neg_b;
- BitField<7, 1, u64> abs_a;
- BitField<39, 3, u64> pred39;
- BitField<42, 1, u64> neg_pred;
- BitField<43, 1, u64> neg_a;
- BitField<44, 1, u64> abs_b;
- BitField<45, 2, PredOperation> op;
- BitField<47, 1, u64> ftz;
- BitField<48, 4, PredCondition> cond;
- } fsetp;
-
- union {
- BitField<0, 3, u64> pred0;
- BitField<3, 3, u64> pred3;
- BitField<39, 3, u64> pred39;
- BitField<42, 1, u64> neg_pred;
- BitField<45, 2, PredOperation> op;
- BitField<48, 1, u64> is_signed;
- BitField<49, 3, PredCondition> cond;
- } isetp;
-
- union {
- BitField<48, 1, u64> is_signed;
- BitField<49, 3, PredCondition> cond;
- } icmp;
-
- union {
- BitField<0, 3, u64> pred0;
- BitField<3, 3, u64> pred3;
- BitField<12, 3, u64> pred12;
- BitField<15, 1, u64> neg_pred12;
- BitField<24, 2, PredOperation> cond;
- BitField<29, 3, u64> pred29;
- BitField<32, 1, u64> neg_pred29;
- BitField<39, 3, u64> pred39;
- BitField<42, 1, u64> neg_pred39;
- BitField<45, 2, PredOperation> op;
- } psetp;
-
- union {
- BitField<43, 4, PredCondition> cond;
- BitField<45, 2, PredOperation> op;
- BitField<3, 3, u64> pred3;
- BitField<0, 3, u64> pred0;
- BitField<39, 3, u64> pred39;
- } vsetp;
-
- union {
- BitField<12, 3, u64> pred12;
- BitField<15, 1, u64> neg_pred12;
- BitField<24, 2, PredOperation> cond;
- BitField<29, 3, u64> pred29;
- BitField<32, 1, u64> neg_pred29;
- BitField<39, 3, u64> pred39;
- BitField<42, 1, u64> neg_pred39;
- BitField<44, 1, u64> bf;
- BitField<45, 2, PredOperation> op;
- } pset;
-
- union {
- BitField<0, 3, u64> pred0;
- BitField<3, 3, u64> pred3;
- BitField<8, 5, ConditionCode> cc; // flag in cc
- BitField<39, 3, u64> pred39;
- BitField<42, 1, u64> neg_pred39;
- BitField<45, 4, PredOperation> op; // op with pred39
- } csetp;
-
- union {
- BitField<6, 1, u64> ftz;
- BitField<45, 2, PredOperation> op;
- BitField<3, 3, u64> pred3;
- BitField<0, 3, u64> pred0;
- BitField<43, 1, u64> negate_a;
- BitField<44, 1, u64> abs_a;
- BitField<47, 2, HalfType> type_a;
- union {
- BitField<35, 4, PredCondition> cond;
- BitField<49, 1, u64> h_and;
- BitField<31, 1, u64> negate_b;
- BitField<30, 1, u64> abs_b;
- BitField<28, 2, HalfType> type_b;
- } reg;
- union {
- BitField<56, 1, u64> negate_b;
- BitField<54, 1, u64> abs_b;
- } cbuf;
- union {
- BitField<49, 4, PredCondition> cond;
- BitField<53, 1, u64> h_and;
- } cbuf_and_imm;
- BitField<42, 1, u64> neg_pred;
- BitField<39, 3, u64> pred39;
- } hsetp2;
-
- union {
- BitField<40, 1, R2pMode> mode;
- BitField<41, 2, u64> byte;
- BitField<20, 7, u64> immediate_mask;
- } p2r_r2p;
-
- union {
- BitField<39, 3, u64> pred39;
- BitField<42, 1, u64> neg_pred;
- BitField<43, 1, u64> neg_a;
- BitField<44, 1, u64> abs_b;
- BitField<45, 2, PredOperation> op;
- BitField<48, 4, PredCondition> cond;
- BitField<52, 1, u64> bf;
- BitField<53, 1, u64> neg_b;
- BitField<54, 1, u64> abs_a;
- BitField<55, 1, u64> ftz;
- } fset;
-
- union {
- BitField<47, 1, u64> ftz;
- BitField<48, 4, PredCondition> cond;
- } fcmp;
-
- union {
- BitField<49, 1, u64> bf;
- BitField<35, 3, PredCondition> cond;
- BitField<50, 1, u64> ftz;
- BitField<45, 2, PredOperation> op;
- BitField<43, 1, u64> negate_a;
- BitField<44, 1, u64> abs_a;
- BitField<47, 2, HalfType> type_a;
- BitField<31, 1, u64> negate_b;
- BitField<30, 1, u64> abs_b;
- BitField<28, 2, HalfType> type_b;
- BitField<42, 1, u64> neg_pred;
- BitField<39, 3, u64> pred39;
- } hset2;
-
- union {
- BitField<39, 3, u64> pred39;
- BitField<42, 1, u64> neg_pred;
- BitField<44, 1, u64> bf;
- BitField<45, 2, PredOperation> op;
- BitField<48, 1, u64> is_signed;
- BitField<49, 3, PredCondition> cond;
- } iset;
-
- union {
- BitField<45, 1, u64> negate_a;
- BitField<49, 1, u64> abs_a;
- BitField<10, 2, Register::Size> src_size;
- BitField<13, 1, u64> is_input_signed;
- BitField<8, 2, Register::Size> dst_size;
- BitField<12, 1, u64> is_output_signed;
-
- union {
- BitField<39, 2, u64> tab5cb8_2;
- } i2f;
-
- union {
- BitField<39, 2, F2iRoundingOp> rounding;
- } f2i;
-
- union {
- BitField<39, 4, u64> rounding;
- // H0, H1 extract for F16 missing
- BitField<41, 1, u64> selector; // Guessed as some games set it, TODO: reverse this value
- [[nodiscard]] F2fRoundingOp GetRoundingMode() const {
- constexpr u64 rounding_mask = 0x0B;
- return static_cast<F2fRoundingOp>(rounding.Value() & rounding_mask);
- }
- } f2f;
-
- union {
- BitField<41, 2, u64> selector;
- } int_src;
-
- union {
- BitField<41, 1, u64> selector;
- } float_src;
- } conversion;
-
- union {
- BitField<28, 1, u64> array;
- BitField<29, 2, TextureType> texture_type;
- BitField<31, 4, u64> component_mask;
- BitField<49, 1, u64> nodep_flag;
- BitField<50, 1, u64> dc_flag;
- BitField<54, 1, u64> aoffi_flag;
- BitField<55, 3, TextureProcessMode> process_mode;
-
- [[nodiscard]] bool IsComponentEnabled(std::size_t component) const {
- return ((1ULL << component) & component_mask) != 0;
- }
-
- [[nodiscard]] TextureProcessMode GetTextureProcessMode() const {
- return process_mode;
- }
-
- [[nodiscard]] bool UsesMiscMode(TextureMiscMode mode) const {
- switch (mode) {
- case TextureMiscMode::DC:
- return dc_flag != 0;
- case TextureMiscMode::NODEP:
- return nodep_flag != 0;
- case TextureMiscMode::AOFFI:
- return aoffi_flag != 0;
- default:
- break;
- }
- return false;
- }
- } tex;
-
- union {
- BitField<28, 1, u64> array;
- BitField<29, 2, TextureType> texture_type;
- BitField<31, 4, u64> component_mask;
- BitField<49, 1, u64> nodep_flag;
- BitField<50, 1, u64> dc_flag;
- BitField<36, 1, u64> aoffi_flag;
- BitField<37, 3, TextureProcessMode> process_mode;
-
- [[nodiscard]] bool IsComponentEnabled(std::size_t component) const {
- return ((1ULL << component) & component_mask) != 0;
- }
-
- [[nodiscard]] TextureProcessMode GetTextureProcessMode() const {
- return process_mode;
- }
-
- [[nodiscard]] bool UsesMiscMode(TextureMiscMode mode) const {
- switch (mode) {
- case TextureMiscMode::DC:
- return dc_flag != 0;
- case TextureMiscMode::NODEP:
- return nodep_flag != 0;
- case TextureMiscMode::AOFFI:
- return aoffi_flag != 0;
- default:
- break;
- }
- return false;
- }
- } tex_b;
-
- union {
- BitField<22, 6, TextureQueryType> query_type;
- BitField<31, 4, u64> component_mask;
- BitField<49, 1, u64> nodep_flag;
-
- [[nodiscard]] bool UsesMiscMode(TextureMiscMode mode) const {
- switch (mode) {
- case TextureMiscMode::NODEP:
- return nodep_flag != 0;
- default:
- break;
- }
- return false;
- }
-
- [[nodiscard]] bool IsComponentEnabled(std::size_t component) const {
- return ((1ULL << component) & component_mask) != 0;
- }
- } txq;
-
- union {
- BitField<28, 1, u64> array;
- BitField<29, 2, TextureType> texture_type;
- BitField<31, 4, u64> component_mask;
- BitField<35, 1, u64> ndv_flag;
- BitField<49, 1, u64> nodep_flag;
-
- [[nodiscard]] bool IsComponentEnabled(std::size_t component) const {
- return ((1ULL << component) & component_mask) != 0;
- }
-
- [[nodiscard]] bool UsesMiscMode(TextureMiscMode mode) const {
- switch (mode) {
- case TextureMiscMode::NDV:
- return (ndv_flag != 0);
- case TextureMiscMode::NODEP:
- return (nodep_flag != 0);
- default:
- break;
- }
- return false;
- }
- } tmml;
-
- union {
- BitField<28, 1, u64> array;
- BitField<29, 2, TextureType> texture_type;
- BitField<35, 1, u64> ndv_flag;
- BitField<49, 1, u64> nodep_flag;
- BitField<50, 1, u64> dc_flag;
- BitField<54, 2, u64> offset_mode;
- BitField<56, 2, u64> component;
-
- [[nodiscard]] bool UsesMiscMode(TextureMiscMode mode) const {
- switch (mode) {
- case TextureMiscMode::NDV:
- return ndv_flag != 0;
- case TextureMiscMode::NODEP:
- return nodep_flag != 0;
- case TextureMiscMode::DC:
- return dc_flag != 0;
- case TextureMiscMode::AOFFI:
- return offset_mode == 1;
- case TextureMiscMode::PTP:
- return offset_mode == 2;
- default:
- break;
- }
- return false;
- }
- } tld4;
-
- union {
- BitField<35, 1, u64> ndv_flag;
- BitField<49, 1, u64> nodep_flag;
- BitField<50, 1, u64> dc_flag;
- BitField<33, 2, u64> offset_mode;
- BitField<37, 2, u64> component;
-
- [[nodiscard]] bool UsesMiscMode(TextureMiscMode mode) const {
- switch (mode) {
- case TextureMiscMode::NDV:
- return ndv_flag != 0;
- case TextureMiscMode::NODEP:
- return nodep_flag != 0;
- case TextureMiscMode::DC:
- return dc_flag != 0;
- case TextureMiscMode::AOFFI:
- return offset_mode == 1;
- case TextureMiscMode::PTP:
- return offset_mode == 2;
- default:
- break;
- }
- return false;
- }
- } tld4_b;
-
- union {
- BitField<49, 1, u64> nodep_flag;
- BitField<50, 1, u64> dc_flag;
- BitField<51, 1, u64> aoffi_flag;
- BitField<52, 2, u64> component;
- BitField<55, 1, u64> fp16_flag;
-
- [[nodiscard]] bool UsesMiscMode(TextureMiscMode mode) const {
- switch (mode) {
- case TextureMiscMode::DC:
- return dc_flag != 0;
- case TextureMiscMode::NODEP:
- return nodep_flag != 0;
- case TextureMiscMode::AOFFI:
- return aoffi_flag != 0;
- default:
- break;
- }
- return false;
- }
- } tld4s;
-
- union {
- BitField<0, 8, Register> gpr0;
- BitField<28, 8, Register> gpr28;
- BitField<49, 1, u64> nodep_flag;
- BitField<50, 3, u64> component_mask_selector;
- BitField<53, 4, u64> texture_info;
- BitField<59, 1, u64> fp32_flag;
-
- [[nodiscard]] TextureType GetTextureType() const {
- // The TEXS instruction has a weird encoding for the texture type.
- if (texture_info == 0) {
- return TextureType::Texture1D;
- }
- if (texture_info >= 1 && texture_info <= 9) {
- return TextureType::Texture2D;
- }
- if (texture_info >= 10 && texture_info <= 11) {
- return TextureType::Texture3D;
- }
- if (texture_info >= 12 && texture_info <= 13) {
- return TextureType::TextureCube;
- }
-
- LOG_CRITICAL(HW_GPU, "Unhandled texture_info: {}", texture_info.Value());
- UNREACHABLE();
- return TextureType::Texture1D;
- }
-
- [[nodiscard]] TextureProcessMode GetTextureProcessMode() const {
- switch (texture_info) {
- case 0:
- case 2:
- case 6:
- case 8:
- case 9:
- case 11:
- return TextureProcessMode::LZ;
- case 3:
- case 5:
- case 13:
- return TextureProcessMode::LL;
- default:
- break;
- }
- return TextureProcessMode::None;
- }
-
- [[nodiscard]] bool UsesMiscMode(TextureMiscMode mode) const {
- switch (mode) {
- case TextureMiscMode::DC:
- return (texture_info >= 4 && texture_info <= 6) || texture_info == 9;
- case TextureMiscMode::NODEP:
- return nodep_flag != 0;
- default:
- break;
- }
- return false;
- }
-
- [[nodiscard]] bool IsArrayTexture() const {
- // TEXS only supports Texture2D arrays.
- return texture_info >= 7 && texture_info <= 9;
- }
-
- [[nodiscard]] bool HasTwoDestinations() const {
- return gpr28.Value() != Register::ZeroIndex;
- }
-
- [[nodiscard]] bool IsComponentEnabled(std::size_t component) const {
- static constexpr std::array<std::array<u32, 8>, 4> mask_lut{{
- {},
- {0x1, 0x2, 0x4, 0x8, 0x3, 0x9, 0xa, 0xc},
- {0x1, 0x2, 0x4, 0x8, 0x3, 0x9, 0xa, 0xc},
- {0x7, 0xb, 0xd, 0xe, 0xf},
- }};
-
- std::size_t index{gpr0.Value() != Register::ZeroIndex ? 1U : 0U};
- index |= gpr28.Value() != Register::ZeroIndex ? 2 : 0;
-
- u32 mask = mask_lut[index][component_mask_selector];
- // A mask of 0 means this instruction uses an unimplemented mask.
- ASSERT(mask != 0);
- return ((1ull << component) & mask) != 0;
- }
- } texs;
-
- union {
- BitField<28, 1, u64> is_array;
- BitField<29, 2, TextureType> texture_type;
- BitField<35, 1, u64> aoffi;
- BitField<49, 1, u64> nodep_flag;
- BitField<50, 1, u64> ms; // Multisample?
- BitField<54, 1, u64> cl;
- BitField<55, 1, u64> process_mode;
-
- [[nodiscard]] TextureProcessMode GetTextureProcessMode() const {
- return process_mode == 0 ? TextureProcessMode::LZ : TextureProcessMode::LL;
- }
- } tld;
-
- union {
- BitField<49, 1, u64> nodep_flag;
- BitField<53, 4, u64> texture_info;
- BitField<59, 1, u64> fp32_flag;
-
- [[nodiscard]] TextureType GetTextureType() const {
- // The TLDS instruction has a weird encoding for the texture type.
- if (texture_info <= 1) {
- return TextureType::Texture1D;
- }
- if (texture_info == 2 || texture_info == 8 || texture_info == 12 ||
- (texture_info >= 4 && texture_info <= 6)) {
- return TextureType::Texture2D;
- }
- if (texture_info == 7) {
- return TextureType::Texture3D;
- }
-
- LOG_CRITICAL(HW_GPU, "Unhandled texture_info: {}", texture_info.Value());
- UNREACHABLE();
- return TextureType::Texture1D;
- }
-
- [[nodiscard]] TextureProcessMode GetTextureProcessMode() const {
- if (texture_info == 1 || texture_info == 5 || texture_info == 12) {
- return TextureProcessMode::LL;
- }
- return TextureProcessMode::LZ;
- }
-
- [[nodiscard]] bool UsesMiscMode(TextureMiscMode mode) const {
- switch (mode) {
- case TextureMiscMode::AOFFI:
- return texture_info == 12 || texture_info == 4;
- case TextureMiscMode::MZ:
- return texture_info == 5;
- case TextureMiscMode::NODEP:
- return nodep_flag != 0;
- default:
- break;
- }
- return false;
- }
-
- [[nodiscard]] bool IsArrayTexture() const {
- // TEXS only supports Texture2D arrays.
- return texture_info == 8;
- }
- } tlds;
-
- union {
- BitField<28, 1, u64> is_array;
- BitField<29, 2, TextureType> texture_type;
- BitField<35, 1, u64> aoffi_flag;
- BitField<49, 1, u64> nodep_flag;
-
- [[nodiscard]] bool UsesMiscMode(TextureMiscMode mode) const {
- switch (mode) {
- case TextureMiscMode::AOFFI:
- return aoffi_flag != 0;
- case TextureMiscMode::NODEP:
- return nodep_flag != 0;
- default:
- break;
- }
- return false;
- }
-
- } txd;
-
- union {
- BitField<24, 2, StoreCacheManagement> cache_management;
- BitField<33, 3, ImageType> image_type;
- BitField<49, 2, OutOfBoundsStore> out_of_bounds_store;
- BitField<51, 1, u64> is_immediate;
- BitField<52, 1, SurfaceDataMode> mode;
-
- BitField<20, 3, StoreType> store_data_layout;
- BitField<20, 4, u64> component_mask_selector;
-
- [[nodiscard]] bool IsComponentEnabled(std::size_t component) const {
- ASSERT(mode == SurfaceDataMode::P);
- constexpr u8 R = 0b0001;
- constexpr u8 G = 0b0010;
- constexpr u8 B = 0b0100;
- constexpr u8 A = 0b1000;
- constexpr std::array<u8, 16> mask = {
- 0, (R), (G), (R | G), (B), (R | B),
- (G | B), (R | G | B), (A), (R | A), (G | A), (R | G | A),
- (B | A), (R | B | A), (G | B | A), (R | G | B | A)};
- return std::bitset<4>{mask.at(component_mask_selector)}.test(component);
- }
-
- [[nodiscard]] StoreType GetStoreDataLayout() const {
- ASSERT(mode == SurfaceDataMode::D_BA);
- return store_data_layout;
- }
- } suldst;
-
- union {
- BitField<28, 1, u64> is_ba;
- BitField<51, 3, ImageAtomicOperationType> operation_type;
- BitField<33, 3, ImageType> image_type;
- BitField<29, 4, ImageAtomicOperation> operation;
- BitField<49, 2, OutOfBoundsStore> out_of_bounds_store;
- } suatom_d;
-
- union {
- BitField<20, 24, u64> target;
- BitField<5, 1, u64> constant_buffer;
-
- [[nodiscard]] s32 GetBranchTarget() const {
- // Sign extend the branch target offset
- const auto mask = 1U << (24 - 1);
- const auto target_value = static_cast<u32>(target);
- constexpr auto instruction_size = static_cast<s32>(sizeof(Instruction));
-
- // The branch offset is relative to the next instruction and is stored in bytes, so
- // divide it by the size of an instruction and add 1 to it.
- return static_cast<s32>((target_value ^ mask) - mask) / instruction_size + 1;
- }
- } bra;
-
- union {
- BitField<20, 24, u64> target;
- BitField<5, 1, u64> constant_buffer;
-
- [[nodiscard]] s32 GetBranchExtend() const {
- // Sign extend the branch target offset
- const auto mask = 1U << (24 - 1);
- const auto target_value = static_cast<u32>(target);
- constexpr auto instruction_size = static_cast<s32>(sizeof(Instruction));
-
- // The branch offset is relative to the next instruction and is stored in bytes, so
- // divide it by the size of an instruction and add 1 to it.
- return static_cast<s32>((target_value ^ mask) - mask) / instruction_size + 1;
- }
- } brx;
-
- union {
- BitField<39, 1, u64> emit; // EmitVertex
- BitField<40, 1, u64> cut; // EndPrimitive
- } out;
-
- union {
- BitField<31, 1, u64> skew;
- BitField<32, 1, u64> o;
- BitField<33, 2, IsberdMode> mode;
- BitField<47, 2, IsberdShift> shift;
- } isberd;
-
- union {
- BitField<8, 2, MembarType> type;
- BitField<0, 2, MembarUnknown> unknown;
- } membar;
-
- union {
- BitField<48, 1, u64> signed_a;
- BitField<38, 1, u64> is_byte_chunk_a;
- BitField<36, 2, VideoType> type_a;
- BitField<36, 2, u64> byte_height_a;
-
- BitField<49, 1, u64> signed_b;
- BitField<50, 1, u64> use_register_b;
- BitField<30, 1, u64> is_byte_chunk_b;
- BitField<28, 2, VideoType> type_b;
- BitField<28, 2, u64> byte_height_b;
- } video;
-
- union {
- BitField<51, 2, VmadShr> shr;
- BitField<55, 1, u64> saturate; // Saturates the result (a * b + c)
- BitField<47, 1, u64> cc;
- } vmad;
-
- union {
- BitField<54, 1, u64> is_dest_signed;
- BitField<48, 1, u64> is_src_a_signed;
- BitField<49, 1, u64> is_src_b_signed;
- BitField<37, 2, u64> src_format_a;
- BitField<29, 2, u64> src_format_b;
- BitField<56, 1, u64> mx;
- BitField<55, 1, u64> sat;
- BitField<36, 2, u64> selector_a;
- BitField<28, 2, u64> selector_b;
- BitField<50, 1, u64> is_op_b_register;
- BitField<51, 3, VmnmxOperation> operation;
-
- [[nodiscard]] VmnmxType SourceFormatA() const {
- switch (src_format_a) {
- case 0b11:
- return VmnmxType::Bits32;
- case 0b10:
- return VmnmxType::Bits16;
- default:
- return VmnmxType::Bits8;
- }
- }
-
- [[nodiscard]] VmnmxType SourceFormatB() const {
- switch (src_format_b) {
- case 0b11:
- return VmnmxType::Bits32;
- case 0b10:
- return VmnmxType::Bits16;
- default:
- return VmnmxType::Bits8;
- }
- }
- } vmnmx;
-
- union {
- BitField<20, 16, u64> imm20_16;
- BitField<35, 1, u64> high_b_rr; // used on RR
- BitField<36, 1, u64> product_shift_left;
- BitField<37, 1, u64> merge_37;
- BitField<48, 1, u64> sign_a;
- BitField<49, 1, u64> sign_b;
- BitField<50, 2, XmadMode> mode_cbf; // used by CR, RC
- BitField<50, 3, XmadMode> mode;
- BitField<52, 1, u64> high_b;
- BitField<53, 1, u64> high_a;
- BitField<55, 1, u64> product_shift_left_second; // used on CR
- BitField<56, 1, u64> merge_56;
- } xmad;
-
- union {
- BitField<20, 14, u64> shifted_offset;
- BitField<34, 5, u64> index;
-
- [[nodiscard]] u64 GetOffset() const {
- return shifted_offset * 4;
- }
- } cbuf34;
-
- union {
- BitField<20, 16, s64> offset;
- BitField<36, 5, u64> index;
-
- [[nodiscard]] s64 GetOffset() const {
- return offset;
- }
- } cbuf36;
-
- // Unsure about the size of this one.
- // It's always used with a gpr0, so any size should be fine.
- BitField<20, 8, SystemVariable> sys20;
-
- BitField<47, 1, u64> generates_cc;
- BitField<61, 1, u64> is_b_imm;
- BitField<60, 1, u64> is_b_gpr;
- BitField<59, 1, u64> is_c_gpr;
- BitField<20, 24, s64> smem_imm;
- BitField<0, 5, ConditionCode> flow_condition_code;
-
- Attribute attribute;
- Sampler sampler;
- Image image;
-
- u64 value;
-};
-static_assert(sizeof(Instruction) == 0x8, "Incorrect structure size");
-static_assert(std::is_standard_layout_v<Instruction>, "Instruction is not standard layout");
-
-class OpCode {
-public:
- enum class Id {
- KIL,
- SSY,
- SYNC,
- BRK,
- DEPBAR,
- VOTE,
- VOTE_VTG,
- SHFL,
- FSWZADD,
- BFE_C,
- BFE_R,
- BFE_IMM,
- BFI_RC,
- BFI_IMM_R,
- BRA,
- BRX,
- PBK,
- LD_A,
- LD_L,
- LD_S,
- LD_C,
- LD, // Load from generic memory
- LDG, // Load from global memory
- ST_A,
- ST_L,
- ST_S,
- ST, // Store in generic memory
- STG, // Store in global memory
- RED, // Reduction operation
- ATOM, // Atomic operation on global memory
- ATOMS, // Atomic operation on shared memory
- AL2P, // Transforms attribute memory into physical memory
- TEX,
- TEX_B, // Texture Load Bindless
- TXQ, // Texture Query
- TXQ_B, // Texture Query Bindless
- TEXS, // Texture Fetch with scalar/non-vec4 source/destinations
- TLD, // Texture Load
- TLDS, // Texture Load with scalar/non-vec4 source/destinations
- TLD4, // Texture Gather 4
- TLD4_B, // Texture Gather 4 Bindless
- TLD4S, // Texture Load 4 with scalar / non - vec4 source / destinations
- TMML_B, // Texture Mip Map Level
- TMML, // Texture Mip Map Level
- TXD, // Texture Gradient/Load with Derivates
- TXD_B, // Texture Gradient/Load with Derivates Bindless
- SUST, // Surface Store
- SULD, // Surface Load
- SUATOM, // Surface Atomic Operation
- EXIT,
- NOP,
- IPA,
- OUT_R, // Emit vertex/primitive
- ISBERD,
- BAR,
- MEMBAR,
- VMAD,
- VSETP,
- VMNMX,
- FFMA_IMM, // Fused Multiply and Add
- FFMA_CR,
- FFMA_RC,
- FFMA_RR,
- FADD_C,
- FADD_R,
- FADD_IMM,
- FADD32I,
- FMUL_C,
- FMUL_R,
- FMUL_IMM,
- FMUL32_IMM,
- IADD_C,
- IADD_R,
- IADD_IMM,
- IADD3_C, // Add 3 Integers
- IADD3_R,
- IADD3_IMM,
- IADD32I,
- ISCADD_C, // Scale and Add
- ISCADD_R,
- ISCADD_IMM,
- FLO_R,
- FLO_C,
- FLO_IMM,
- LEA_R1,
- LEA_R2,
- LEA_RZ,
- LEA_IMM,
- LEA_HI,
- HADD2_C,
- HADD2_R,
- HADD2_IMM,
- HMUL2_C,
- HMUL2_R,
- HMUL2_IMM,
- HFMA2_CR,
- HFMA2_RC,
- HFMA2_RR,
- HFMA2_IMM_R,
- HSETP2_C,
- HSETP2_R,
- HSETP2_IMM,
- HSET2_C,
- HSET2_R,
- HSET2_IMM,
- POPC_C,
- POPC_R,
- POPC_IMM,
- SEL_C,
- SEL_R,
- SEL_IMM,
- ICMP_RC,
- ICMP_R,
- ICMP_CR,
- ICMP_IMM,
- FCMP_RR,
- FCMP_RC,
- FCMP_IMMR,
- MUFU, // Multi-Function Operator
- RRO_C, // Range Reduction Operator
- RRO_R,
- RRO_IMM,
- F2F_C,
- F2F_R,
- F2F_IMM,
- F2I_C,
- F2I_R,
- F2I_IMM,
- I2F_C,
- I2F_R,
- I2F_IMM,
- I2I_C,
- I2I_R,
- I2I_IMM,
- LOP_C,
- LOP_R,
- LOP_IMM,
- LOP32I,
- LOP3_C,
- LOP3_R,
- LOP3_IMM,
- MOV_C,
- MOV_R,
- MOV_IMM,
- S2R,
- MOV32_IMM,
- SHL_C,
- SHL_R,
- SHL_IMM,
- SHR_C,
- SHR_R,
- SHR_IMM,
- SHF_RIGHT_R,
- SHF_RIGHT_IMM,
- SHF_LEFT_R,
- SHF_LEFT_IMM,
- FMNMX_C,
- FMNMX_R,
- FMNMX_IMM,
- IMNMX_C,
- IMNMX_R,
- IMNMX_IMM,
- FSETP_C, // Set Predicate
- FSETP_R,
- FSETP_IMM,
- FSET_C,
- FSET_R,
- FSET_IMM,
- ISETP_C,
- ISETP_IMM,
- ISETP_R,
- ISET_R,
- ISET_C,
- ISET_IMM,
- PSETP,
- PSET,
- CSETP,
- R2P_IMM,
- P2R_IMM,
- XMAD_IMM,
- XMAD_CR,
- XMAD_RC,
- XMAD_RR,
- };
-
- enum class Type {
- Trivial,
- Arithmetic,
- ArithmeticImmediate,
- ArithmeticInteger,
- ArithmeticIntegerImmediate,
- ArithmeticHalf,
- ArithmeticHalfImmediate,
- Bfe,
- Bfi,
- Shift,
- Ffma,
- Hfma2,
- Flow,
- Synch,
- Warp,
- Memory,
- Texture,
- Image,
- FloatSet,
- FloatSetPredicate,
- IntegerSet,
- IntegerSetPredicate,
- HalfSet,
- HalfSetPredicate,
- PredicateSetPredicate,
- PredicateSetRegister,
- RegisterSetPredicate,
- Conversion,
- Video,
- Xmad,
- Unknown,
- };
-
- /// Returns whether an opcode has an execution predicate field or not (ie, whether it can be
- /// conditionally executed).
- [[nodiscard]] static bool IsPredicatedInstruction(Id opcode) {
- // TODO(Subv): Add the rest of unpredicated instructions.
- return opcode != Id::SSY && opcode != Id::PBK;
- }
-
- class Matcher {
- public:
- constexpr Matcher(const char* const name_, u16 mask_, u16 expected_, Id id_, Type type_)
- : name{name_}, mask{mask_}, expected{expected_}, id{id_}, type{type_} {}
-
- [[nodiscard]] constexpr const char* GetName() const {
- return name;
- }
-
- [[nodiscard]] constexpr u16 GetMask() const {
- return mask;
- }
-
- [[nodiscard]] constexpr Id GetId() const {
- return id;
- }
-
- [[nodiscard]] constexpr Type GetType() const {
- return type;
- }
-
- /**
- * Tests to see if the given instruction is the instruction this matcher represents.
- * @param instruction The instruction to test
- * @returns true if the given instruction matches.
- */
- [[nodiscard]] constexpr bool Matches(u16 instruction) const {
- return (instruction & mask) == expected;
- }
-
- private:
- const char* name;
- u16 mask;
- u16 expected;
- Id id;
- Type type;
- };
-
- using DecodeResult = std::optional<std::reference_wrapper<const Matcher>>;
- [[nodiscard]] static DecodeResult Decode(Instruction instr) {
- static const auto table{GetDecodeTable()};
-
- const auto matches_instruction = [instr](const auto& matcher) {
- return matcher.Matches(static_cast<u16>(instr.opcode));
- };
-
- auto iter = std::find_if(table.begin(), table.end(), matches_instruction);
- return iter != table.end() ? std::optional<std::reference_wrapper<const Matcher>>(*iter)
- : std::nullopt;
- }
-
-private:
- struct Detail {
- private:
- static constexpr std::size_t opcode_bitsize = 16;
-
- /**
- * Generates the mask and the expected value after masking from a given bitstring.
- * A '0' in a bitstring indicates that a zero must be present at that bit position.
- * A '1' in a bitstring indicates that a one must be present at that bit position.
- */
- [[nodiscard]] static constexpr auto GetMaskAndExpect(const char* const bitstring) {
- u16 mask = 0, expect = 0;
- for (std::size_t i = 0; i < opcode_bitsize; i++) {
- const std::size_t bit_position = opcode_bitsize - i - 1;
- switch (bitstring[i]) {
- case '0':
- mask |= static_cast<u16>(1U << bit_position);
- break;
- case '1':
- expect |= static_cast<u16>(1U << bit_position);
- mask |= static_cast<u16>(1U << bit_position);
- break;
- default:
- // Ignore
- break;
- }
- }
- return std::make_pair(mask, expect);
- }
-
- public:
- /// Creates a matcher that can match and parse instructions based on bitstring.
- [[nodiscard]] static constexpr auto GetMatcher(const char* const bitstring, Id op,
- Type type, const char* const name) {
- const auto [mask, expected] = GetMaskAndExpect(bitstring);
- return Matcher(name, mask, expected, op, type);
- }
- };
-
- [[nodiscard]] static std::vector<Matcher> GetDecodeTable() {
- std::vector<Matcher> table = {
-#define INST(bitstring, op, type, name) Detail::GetMatcher(bitstring, op, type, name)
- INST("111000110011----", Id::KIL, Type::Flow, "KIL"),
- INST("111000101001----", Id::SSY, Type::Flow, "SSY"),
- INST("111000101010----", Id::PBK, Type::Flow, "PBK"),
- INST("111000100100----", Id::BRA, Type::Flow, "BRA"),
- INST("111000100101----", Id::BRX, Type::Flow, "BRX"),
- INST("1111000011111---", Id::SYNC, Type::Flow, "SYNC"),
- INST("111000110100----", Id::BRK, Type::Flow, "BRK"),
- INST("111000110000----", Id::EXIT, Type::Flow, "EXIT"),
- INST("1111000011110---", Id::DEPBAR, Type::Synch, "DEPBAR"),
- INST("0101000011011---", Id::VOTE, Type::Warp, "VOTE"),
- INST("0101000011100---", Id::VOTE_VTG, Type::Warp, "VOTE_VTG"),
- INST("1110111100010---", Id::SHFL, Type::Warp, "SHFL"),
- INST("0101000011111---", Id::FSWZADD, Type::Warp, "FSWZADD"),
- INST("1110111111011---", Id::LD_A, Type::Memory, "LD_A"),
- INST("1110111101001---", Id::LD_S, Type::Memory, "LD_S"),
- INST("1110111101000---", Id::LD_L, Type::Memory, "LD_L"),
- INST("1110111110010---", Id::LD_C, Type::Memory, "LD_C"),
- INST("100-------------", Id::LD, Type::Memory, "LD"),
- INST("1110111011010---", Id::LDG, Type::Memory, "LDG"),
- INST("1110111111110---", Id::ST_A, Type::Memory, "ST_A"),
- INST("1110111101011---", Id::ST_S, Type::Memory, "ST_S"),
- INST("1110111101010---", Id::ST_L, Type::Memory, "ST_L"),
- INST("101-------------", Id::ST, Type::Memory, "ST"),
- INST("1110111011011---", Id::STG, Type::Memory, "STG"),
- INST("1110101111111---", Id::RED, Type::Memory, "RED"),
- INST("11101101--------", Id::ATOM, Type::Memory, "ATOM"),
- INST("11101100--------", Id::ATOMS, Type::Memory, "ATOMS"),
- INST("1110111110100---", Id::AL2P, Type::Memory, "AL2P"),
- INST("110000----111---", Id::TEX, Type::Texture, "TEX"),
- INST("1101111010111---", Id::TEX_B, Type::Texture, "TEX_B"),
- INST("1101111101001---", Id::TXQ, Type::Texture, "TXQ"),
- INST("1101111101010---", Id::TXQ_B, Type::Texture, "TXQ_B"),
- INST("1101-00---------", Id::TEXS, Type::Texture, "TEXS"),
- INST("11011100--11----", Id::TLD, Type::Texture, "TLD"),
- INST("1101-01---------", Id::TLDS, Type::Texture, "TLDS"),
- INST("110010----111---", Id::TLD4, Type::Texture, "TLD4"),
- INST("1101111011111---", Id::TLD4_B, Type::Texture, "TLD4_B"),
- INST("11011111-0------", Id::TLD4S, Type::Texture, "TLD4S"),
- INST("110111110110----", Id::TMML_B, Type::Texture, "TMML_B"),
- INST("1101111101011---", Id::TMML, Type::Texture, "TMML"),
- INST("11011110011110--", Id::TXD_B, Type::Texture, "TXD_B"),
- INST("11011110001110--", Id::TXD, Type::Texture, "TXD"),
- INST("11101011001-----", Id::SUST, Type::Image, "SUST"),
- INST("11101011000-----", Id::SULD, Type::Image, "SULD"),
- INST("1110101000------", Id::SUATOM, Type::Image, "SUATOM_D"),
- INST("0101000010110---", Id::NOP, Type::Trivial, "NOP"),
- INST("11100000--------", Id::IPA, Type::Trivial, "IPA"),
- INST("1111101111100---", Id::OUT_R, Type::Trivial, "OUT_R"),
- INST("1110111111010---", Id::ISBERD, Type::Trivial, "ISBERD"),
- INST("1111000010101---", Id::BAR, Type::Trivial, "BAR"),
- INST("1110111110011---", Id::MEMBAR, Type::Trivial, "MEMBAR"),
- INST("01011111--------", Id::VMAD, Type::Video, "VMAD"),
- INST("0101000011110---", Id::VSETP, Type::Video, "VSETP"),
- INST("0011101---------", Id::VMNMX, Type::Video, "VMNMX"),
- INST("0011001-1-------", Id::FFMA_IMM, Type::Ffma, "FFMA_IMM"),
- INST("010010011-------", Id::FFMA_CR, Type::Ffma, "FFMA_CR"),
- INST("010100011-------", Id::FFMA_RC, Type::Ffma, "FFMA_RC"),
- INST("010110011-------", Id::FFMA_RR, Type::Ffma, "FFMA_RR"),
- INST("0100110001011---", Id::FADD_C, Type::Arithmetic, "FADD_C"),
- INST("0101110001011---", Id::FADD_R, Type::Arithmetic, "FADD_R"),
- INST("0011100-01011---", Id::FADD_IMM, Type::Arithmetic, "FADD_IMM"),
- INST("000010----------", Id::FADD32I, Type::ArithmeticImmediate, "FADD32I"),
- INST("0100110001101---", Id::FMUL_C, Type::Arithmetic, "FMUL_C"),
- INST("0101110001101---", Id::FMUL_R, Type::Arithmetic, "FMUL_R"),
- INST("0011100-01101---", Id::FMUL_IMM, Type::Arithmetic, "FMUL_IMM"),
- INST("00011110--------", Id::FMUL32_IMM, Type::ArithmeticImmediate, "FMUL32_IMM"),
- INST("0100110000010---", Id::IADD_C, Type::ArithmeticInteger, "IADD_C"),
- INST("0101110000010---", Id::IADD_R, Type::ArithmeticInteger, "IADD_R"),
- INST("0011100-00010---", Id::IADD_IMM, Type::ArithmeticInteger, "IADD_IMM"),
- INST("010011001100----", Id::IADD3_C, Type::ArithmeticInteger, "IADD3_C"),
- INST("010111001100----", Id::IADD3_R, Type::ArithmeticInteger, "IADD3_R"),
- INST("0011100-1100----", Id::IADD3_IMM, Type::ArithmeticInteger, "IADD3_IMM"),
- INST("0001110---------", Id::IADD32I, Type::ArithmeticIntegerImmediate, "IADD32I"),
- INST("0100110000011---", Id::ISCADD_C, Type::ArithmeticInteger, "ISCADD_C"),
- INST("0101110000011---", Id::ISCADD_R, Type::ArithmeticInteger, "ISCADD_R"),
- INST("0011100-00011---", Id::ISCADD_IMM, Type::ArithmeticInteger, "ISCADD_IMM"),
- INST("0100110000001---", Id::POPC_C, Type::ArithmeticInteger, "POPC_C"),
- INST("0101110000001---", Id::POPC_R, Type::ArithmeticInteger, "POPC_R"),
- INST("0011100-00001---", Id::POPC_IMM, Type::ArithmeticInteger, "POPC_IMM"),
- INST("0100110010100---", Id::SEL_C, Type::ArithmeticInteger, "SEL_C"),
- INST("0101110010100---", Id::SEL_R, Type::ArithmeticInteger, "SEL_R"),
- INST("0011100-10100---", Id::SEL_IMM, Type::ArithmeticInteger, "SEL_IMM"),
- INST("010100110100----", Id::ICMP_RC, Type::ArithmeticInteger, "ICMP_RC"),
- INST("010110110100----", Id::ICMP_R, Type::ArithmeticInteger, "ICMP_R"),
- INST("010010110100----", Id::ICMP_CR, Type::ArithmeticInteger, "ICMP_CR"),
- INST("0011011-0100----", Id::ICMP_IMM, Type::ArithmeticInteger, "ICMP_IMM"),
- INST("0101110000110---", Id::FLO_R, Type::ArithmeticInteger, "FLO_R"),
- INST("0100110000110---", Id::FLO_C, Type::ArithmeticInteger, "FLO_C"),
- INST("0011100-00110---", Id::FLO_IMM, Type::ArithmeticInteger, "FLO_IMM"),
- INST("0101101111011---", Id::LEA_R2, Type::ArithmeticInteger, "LEA_R2"),
- INST("0101101111010---", Id::LEA_R1, Type::ArithmeticInteger, "LEA_R1"),
- INST("001101101101----", Id::LEA_IMM, Type::ArithmeticInteger, "LEA_IMM"),
- INST("010010111101----", Id::LEA_RZ, Type::ArithmeticInteger, "LEA_RZ"),
- INST("00011000--------", Id::LEA_HI, Type::ArithmeticInteger, "LEA_HI"),
- INST("0111101-1-------", Id::HADD2_C, Type::ArithmeticHalf, "HADD2_C"),
- INST("0101110100010---", Id::HADD2_R, Type::ArithmeticHalf, "HADD2_R"),
- INST("0111101-0-------", Id::HADD2_IMM, Type::ArithmeticHalfImmediate, "HADD2_IMM"),
- INST("0111100-1-------", Id::HMUL2_C, Type::ArithmeticHalf, "HMUL2_C"),
- INST("0101110100001---", Id::HMUL2_R, Type::ArithmeticHalf, "HMUL2_R"),
- INST("0111100-0-------", Id::HMUL2_IMM, Type::ArithmeticHalfImmediate, "HMUL2_IMM"),
- INST("01110---1-------", Id::HFMA2_CR, Type::Hfma2, "HFMA2_CR"),
- INST("01100---1-------", Id::HFMA2_RC, Type::Hfma2, "HFMA2_RC"),
- INST("0101110100000---", Id::HFMA2_RR, Type::Hfma2, "HFMA2_RR"),
- INST("01110---0-------", Id::HFMA2_IMM_R, Type::Hfma2, "HFMA2_R_IMM"),
- INST("0111111-1-------", Id::HSETP2_C, Type::HalfSetPredicate, "HSETP2_C"),
- INST("0101110100100---", Id::HSETP2_R, Type::HalfSetPredicate, "HSETP2_R"),
- INST("0111111-0-------", Id::HSETP2_IMM, Type::HalfSetPredicate, "HSETP2_IMM"),
- INST("0111110-1-------", Id::HSET2_C, Type::HalfSet, "HSET2_C"),
- INST("0101110100011---", Id::HSET2_R, Type::HalfSet, "HSET2_R"),
- INST("0111110-0-------", Id::HSET2_IMM, Type::HalfSet, "HSET2_IMM"),
- INST("010110111010----", Id::FCMP_RR, Type::Arithmetic, "FCMP_RR"),
- INST("010010111010----", Id::FCMP_RC, Type::Arithmetic, "FCMP_RC"),
- INST("0011011-1010----", Id::FCMP_IMMR, Type::Arithmetic, "FCMP_IMMR"),
- INST("0101000010000---", Id::MUFU, Type::Arithmetic, "MUFU"),
- INST("0100110010010---", Id::RRO_C, Type::Arithmetic, "RRO_C"),
- INST("0101110010010---", Id::RRO_R, Type::Arithmetic, "RRO_R"),
- INST("0011100-10010---", Id::RRO_IMM, Type::Arithmetic, "RRO_IMM"),
- INST("0100110010101---", Id::F2F_C, Type::Conversion, "F2F_C"),
- INST("0101110010101---", Id::F2F_R, Type::Conversion, "F2F_R"),
- INST("0011100-10101---", Id::F2F_IMM, Type::Conversion, "F2F_IMM"),
- INST("0100110010110---", Id::F2I_C, Type::Conversion, "F2I_C"),
- INST("0101110010110---", Id::F2I_R, Type::Conversion, "F2I_R"),
- INST("0011100-10110---", Id::F2I_IMM, Type::Conversion, "F2I_IMM"),
- INST("0100110010011---", Id::MOV_C, Type::Arithmetic, "MOV_C"),
- INST("0101110010011---", Id::MOV_R, Type::Arithmetic, "MOV_R"),
- INST("0011100-10011---", Id::MOV_IMM, Type::Arithmetic, "MOV_IMM"),
- INST("1111000011001---", Id::S2R, Type::Trivial, "S2R"),
- INST("000000010000----", Id::MOV32_IMM, Type::ArithmeticImmediate, "MOV32_IMM"),
- INST("0100110001100---", Id::FMNMX_C, Type::Arithmetic, "FMNMX_C"),
- INST("0101110001100---", Id::FMNMX_R, Type::Arithmetic, "FMNMX_R"),
- INST("0011100-01100---", Id::FMNMX_IMM, Type::Arithmetic, "FMNMX_IMM"),
- INST("0100110000100---", Id::IMNMX_C, Type::ArithmeticInteger, "IMNMX_C"),
- INST("0101110000100---", Id::IMNMX_R, Type::ArithmeticInteger, "IMNMX_R"),
- INST("0011100-00100---", Id::IMNMX_IMM, Type::ArithmeticInteger, "IMNMX_IMM"),
- INST("0100110000000---", Id::BFE_C, Type::Bfe, "BFE_C"),
- INST("0101110000000---", Id::BFE_R, Type::Bfe, "BFE_R"),
- INST("0011100-00000---", Id::BFE_IMM, Type::Bfe, "BFE_IMM"),
- INST("0101001111110---", Id::BFI_RC, Type::Bfi, "BFI_RC"),
- INST("0011011-11110---", Id::BFI_IMM_R, Type::Bfi, "BFI_IMM_R"),
- INST("0100110001000---", Id::LOP_C, Type::ArithmeticInteger, "LOP_C"),
- INST("0101110001000---", Id::LOP_R, Type::ArithmeticInteger, "LOP_R"),
- INST("0011100-01000---", Id::LOP_IMM, Type::ArithmeticInteger, "LOP_IMM"),
- INST("000001----------", Id::LOP32I, Type::ArithmeticIntegerImmediate, "LOP32I"),
- INST("0000001---------", Id::LOP3_C, Type::ArithmeticInteger, "LOP3_C"),
- INST("0101101111100---", Id::LOP3_R, Type::ArithmeticInteger, "LOP3_R"),
- INST("0011110---------", Id::LOP3_IMM, Type::ArithmeticInteger, "LOP3_IMM"),
- INST("0100110001001---", Id::SHL_C, Type::Shift, "SHL_C"),
- INST("0101110001001---", Id::SHL_R, Type::Shift, "SHL_R"),
- INST("0011100-01001---", Id::SHL_IMM, Type::Shift, "SHL_IMM"),
- INST("0100110000101---", Id::SHR_C, Type::Shift, "SHR_C"),
- INST("0101110000101---", Id::SHR_R, Type::Shift, "SHR_R"),
- INST("0011100-00101---", Id::SHR_IMM, Type::Shift, "SHR_IMM"),
- INST("0101110011111---", Id::SHF_RIGHT_R, Type::Shift, "SHF_RIGHT_R"),
- INST("0011100-11111---", Id::SHF_RIGHT_IMM, Type::Shift, "SHF_RIGHT_IMM"),
- INST("0101101111111---", Id::SHF_LEFT_R, Type::Shift, "SHF_LEFT_R"),
- INST("0011011-11111---", Id::SHF_LEFT_IMM, Type::Shift, "SHF_LEFT_IMM"),
- INST("0100110011100---", Id::I2I_C, Type::Conversion, "I2I_C"),
- INST("0101110011100---", Id::I2I_R, Type::Conversion, "I2I_R"),
- INST("0011100-11100---", Id::I2I_IMM, Type::Conversion, "I2I_IMM"),
- INST("0100110010111---", Id::I2F_C, Type::Conversion, "I2F_C"),
- INST("0101110010111---", Id::I2F_R, Type::Conversion, "I2F_R"),
- INST("0011100-10111---", Id::I2F_IMM, Type::Conversion, "I2F_IMM"),
- INST("01011000--------", Id::FSET_R, Type::FloatSet, "FSET_R"),
- INST("0100100---------", Id::FSET_C, Type::FloatSet, "FSET_C"),
- INST("0011000---------", Id::FSET_IMM, Type::FloatSet, "FSET_IMM"),
- INST("010010111011----", Id::FSETP_C, Type::FloatSetPredicate, "FSETP_C"),
- INST("010110111011----", Id::FSETP_R, Type::FloatSetPredicate, "FSETP_R"),
- INST("0011011-1011----", Id::FSETP_IMM, Type::FloatSetPredicate, "FSETP_IMM"),
- INST("010010110110----", Id::ISETP_C, Type::IntegerSetPredicate, "ISETP_C"),
- INST("010110110110----", Id::ISETP_R, Type::IntegerSetPredicate, "ISETP_R"),
- INST("0011011-0110----", Id::ISETP_IMM, Type::IntegerSetPredicate, "ISETP_IMM"),
- INST("010110110101----", Id::ISET_R, Type::IntegerSet, "ISET_R"),
- INST("010010110101----", Id::ISET_C, Type::IntegerSet, "ISET_C"),
- INST("0011011-0101----", Id::ISET_IMM, Type::IntegerSet, "ISET_IMM"),
- INST("0101000010001---", Id::PSET, Type::PredicateSetRegister, "PSET"),
- INST("0101000010010---", Id::PSETP, Type::PredicateSetPredicate, "PSETP"),
- INST("010100001010----", Id::CSETP, Type::PredicateSetPredicate, "CSETP"),
- INST("0011100-11110---", Id::R2P_IMM, Type::RegisterSetPredicate, "R2P_IMM"),
- INST("0011100-11101---", Id::P2R_IMM, Type::RegisterSetPredicate, "P2R_IMM"),
- INST("0011011-00------", Id::XMAD_IMM, Type::Xmad, "XMAD_IMM"),
- INST("0100111---------", Id::XMAD_CR, Type::Xmad, "XMAD_CR"),
- INST("010100010-------", Id::XMAD_RC, Type::Xmad, "XMAD_RC"),
- INST("0101101100------", Id::XMAD_RR, Type::Xmad, "XMAD_RR"),
- };
-#undef INST
- std::stable_sort(table.begin(), table.end(), [](const auto& a, const auto& b) {
- // If a matcher has more bits in its mask it is more specific, so it
- // should come first.
- return std::bitset<16>(a.GetMask()).count() > std::bitset<16>(b.GetMask()).count();
- });
-
- return table;
- }
-};
-
-} // namespace Tegra::Shader
diff --git a/src/video_core/engines/shader_header.h b/src/video_core/engines/shader_header.h
deleted file mode 100644
index e0d7b89c5..000000000
--- a/src/video_core/engines/shader_header.h
+++ /dev/null
@@ -1,158 +0,0 @@
-// Copyright 2018 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <array>
-#include <optional>
-
-#include "common/bit_field.h"
-#include "common/common_funcs.h"
-#include "common/common_types.h"
-
-namespace Tegra::Shader {
-
-enum class OutputTopology : u32 {
- PointList = 1,
- LineStrip = 6,
- TriangleStrip = 7,
-};
-
-enum class PixelImap : u8 {
- Unused = 0,
- Constant = 1,
- Perspective = 2,
- ScreenLinear = 3,
-};
-
-// Documentation in:
-// http://download.nvidia.com/open-gpu-doc/Shader-Program-Header/1/Shader-Program-Header.html
-struct Header {
- union {
- BitField<0, 5, u32> sph_type;
- BitField<5, 5, u32> version;
- BitField<10, 4, u32> shader_type;
- BitField<14, 1, u32> mrt_enable;
- BitField<15, 1, u32> kills_pixels;
- BitField<16, 1, u32> does_global_store;
- BitField<17, 4, u32> sass_version;
- BitField<21, 5, u32> reserved;
- BitField<26, 1, u32> does_load_or_store;
- BitField<27, 1, u32> does_fp64;
- BitField<28, 4, u32> stream_out_mask;
- } common0;
-
- union {
- BitField<0, 24, u32> shader_local_memory_low_size;
- BitField<24, 8, u32> per_patch_attribute_count;
- } common1;
-
- union {
- BitField<0, 24, u32> shader_local_memory_high_size;
- BitField<24, 8, u32> threads_per_input_primitive;
- } common2;
-
- union {
- BitField<0, 24, u32> shader_local_memory_crs_size;
- BitField<24, 4, OutputTopology> output_topology;
- BitField<28, 4, u32> reserved;
- } common3;
-
- union {
- BitField<0, 12, u32> max_output_vertices;
- BitField<12, 8, u32> store_req_start; // NOTE: not used by geometry shaders.
- BitField<20, 4, u32> reserved;
- BitField<24, 8, u32> store_req_end; // NOTE: not used by geometry shaders.
- } common4;
-
- union {
- struct {
- INSERT_PADDING_BYTES_NOINIT(3); // ImapSystemValuesA
- INSERT_PADDING_BYTES_NOINIT(1); // ImapSystemValuesB
- INSERT_PADDING_BYTES_NOINIT(16); // ImapGenericVector[32]
- INSERT_PADDING_BYTES_NOINIT(2); // ImapColor
- union {
- BitField<0, 8, u16> clip_distances;
- BitField<8, 1, u16> point_sprite_s;
- BitField<9, 1, u16> point_sprite_t;
- BitField<10, 1, u16> fog_coordinate;
- BitField<12, 1, u16> tessellation_eval_point_u;
- BitField<13, 1, u16> tessellation_eval_point_v;
- BitField<14, 1, u16> instance_id;
- BitField<15, 1, u16> vertex_id;
- };
- INSERT_PADDING_BYTES_NOINIT(5); // ImapFixedFncTexture[10]
- INSERT_PADDING_BYTES_NOINIT(1); // ImapReserved
- INSERT_PADDING_BYTES_NOINIT(3); // OmapSystemValuesA
- INSERT_PADDING_BYTES_NOINIT(1); // OmapSystemValuesB
- INSERT_PADDING_BYTES_NOINIT(16); // OmapGenericVector[32]
- INSERT_PADDING_BYTES_NOINIT(2); // OmapColor
- INSERT_PADDING_BYTES_NOINIT(2); // OmapSystemValuesC
- INSERT_PADDING_BYTES_NOINIT(5); // OmapFixedFncTexture[10]
- INSERT_PADDING_BYTES_NOINIT(1); // OmapReserved
- } vtg;
-
- struct {
- INSERT_PADDING_BYTES_NOINIT(3); // ImapSystemValuesA
- INSERT_PADDING_BYTES_NOINIT(1); // ImapSystemValuesB
-
- union {
- BitField<0, 2, PixelImap> x;
- BitField<2, 2, PixelImap> y;
- BitField<4, 2, PixelImap> z;
- BitField<6, 2, PixelImap> w;
- u8 raw;
- } imap_generic_vector[32];
-
- INSERT_PADDING_BYTES_NOINIT(2); // ImapColor
- INSERT_PADDING_BYTES_NOINIT(2); // ImapSystemValuesC
- INSERT_PADDING_BYTES_NOINIT(10); // ImapFixedFncTexture[10]
- INSERT_PADDING_BYTES_NOINIT(2); // ImapReserved
-
- struct {
- u32 target;
- union {
- BitField<0, 1, u32> sample_mask;
- BitField<1, 1, u32> depth;
- BitField<2, 30, u32> reserved;
- };
- } omap;
-
- bool IsColorComponentOutputEnabled(u32 render_target, u32 component) const {
- const u32 bit = render_target * 4 + component;
- return omap.target & (1 << bit);
- }
-
- PixelImap GetPixelImap(u32 attribute) const {
- const auto get_index = [this, attribute](u32 index) {
- return static_cast<PixelImap>(
- (imap_generic_vector[attribute].raw >> (index * 2)) & 3);
- };
-
- std::optional<PixelImap> result;
- for (u32 component = 0; component < 4; ++component) {
- const PixelImap index = get_index(component);
- if (index == PixelImap::Unused) {
- continue;
- }
- if (result && result != index) {
- LOG_CRITICAL(HW_GPU, "Generic attribute conflict in interpolation mode");
- }
- result = index;
- }
- return result.value_or(PixelImap::Unused);
- }
- } ps;
-
- std::array<u32, 0xF> raw;
- };
-
- u64 GetLocalMemorySize() const {
- return (common1.shader_local_memory_low_size |
- (common2.shader_local_memory_high_size << 24));
- }
-};
-static_assert(sizeof(Header) == 0x50, "Incorrect structure size");
-
-} // namespace Tegra::Shader
diff --git a/src/video_core/renderer_vulkan/vk_compute_pipeline.cpp b/src/video_core/renderer_vulkan/vk_compute_pipeline.cpp
index 7a3660496..588ce6139 100644
--- a/src/video_core/renderer_vulkan/vk_compute_pipeline.cpp
+++ b/src/video_core/renderer_vulkan/vk_compute_pipeline.cpp
@@ -4,6 +4,9 @@
#include <vector>
+#include <boost/container/small_vector.hpp>
+
+#include "video_core/renderer_vulkan/vk_buffer_cache.h"
#include "video_core/renderer_vulkan/vk_compute_pipeline.h"
#include "video_core/renderer_vulkan/vk_descriptor_pool.h"
#include "video_core/renderer_vulkan/vk_pipeline_cache.h"
@@ -13,9 +16,142 @@
#include "video_core/vulkan_common/vulkan_wrapper.h"
namespace Vulkan {
+namespace {
+vk::DescriptorSetLayout CreateDescriptorSetLayout(const Device& device, const Shader::Info& info) {
+ boost::container::small_vector<VkDescriptorSetLayoutBinding, 24> bindings;
+ u32 binding{};
+ for ([[maybe_unused]] const auto& desc : info.constant_buffer_descriptors) {
+ bindings.push_back({
+ .binding = binding,
+ .descriptorType = VK_DESCRIPTOR_TYPE_UNIFORM_BUFFER,
+ .descriptorCount = 1,
+ .stageFlags = VK_SHADER_STAGE_COMPUTE_BIT,
+ .pImmutableSamplers = nullptr,
+ });
+ ++binding;
+ }
+ for ([[maybe_unused]] const auto& desc : info.storage_buffers_descriptors) {
+ bindings.push_back({
+ .binding = binding,
+ .descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
+ .descriptorCount = 1,
+ .stageFlags = VK_SHADER_STAGE_COMPUTE_BIT,
+ .pImmutableSamplers = nullptr,
+ });
+ ++binding;
+ }
+ return device.GetLogical().CreateDescriptorSetLayout({
+ .sType = VK_STRUCTURE_TYPE_DESCRIPTOR_SET_LAYOUT_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .bindingCount = static_cast<u32>(bindings.size()),
+ .pBindings = bindings.data(),
+ });
+}
+
+vk::DescriptorUpdateTemplateKHR CreateDescriptorUpdateTemplate(
+ const Device& device, const Shader::Info& info, VkDescriptorSetLayout descriptor_set_layout,
+ VkPipelineLayout pipeline_layout) {
+ boost::container::small_vector<VkDescriptorUpdateTemplateEntry, 24> entries;
+ size_t offset{};
+ u32 binding{};
+ for ([[maybe_unused]] const auto& desc : info.constant_buffer_descriptors) {
+ entries.push_back({
+ .dstBinding = binding,
+ .dstArrayElement = 0,
+ .descriptorCount = 1,
+ .descriptorType = VK_DESCRIPTOR_TYPE_UNIFORM_BUFFER,
+ .offset = offset,
+ .stride = sizeof(DescriptorUpdateEntry),
+ });
+ ++binding;
+ offset += sizeof(DescriptorUpdateEntry);
+ }
+ for ([[maybe_unused]] const auto& desc : info.storage_buffers_descriptors) {
+ entries.push_back({
+ .dstBinding = binding,
+ .dstArrayElement = 0,
+ .descriptorCount = 1,
+ .descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
+ .offset = offset,
+ .stride = sizeof(DescriptorUpdateEntry),
+ });
+ ++binding;
+ offset += sizeof(DescriptorUpdateEntry);
+ }
+ return device.GetLogical().CreateDescriptorUpdateTemplateKHR({
+ .sType = VK_STRUCTURE_TYPE_DESCRIPTOR_UPDATE_TEMPLATE_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .descriptorUpdateEntryCount = static_cast<u32>(entries.size()),
+ .pDescriptorUpdateEntries = entries.data(),
+ .templateType = VK_DESCRIPTOR_UPDATE_TEMPLATE_TYPE_DESCRIPTOR_SET,
+ .descriptorSetLayout = descriptor_set_layout,
+ .pipelineBindPoint = VK_PIPELINE_BIND_POINT_COMPUTE,
+ .pipelineLayout = pipeline_layout,
+ .set = 0,
+ });
+}
+} // Anonymous namespace
+
+ComputePipeline::ComputePipeline(const Device& device, VKDescriptorPool& descriptor_pool,
+ VKUpdateDescriptorQueue& update_descriptor_queue_,
+ const Shader::Info& info_, vk::ShaderModule spv_module_)
+ : update_descriptor_queue{&update_descriptor_queue_}, info{info_},
+ spv_module(std::move(spv_module_)),
+ descriptor_set_layout(CreateDescriptorSetLayout(device, info)),
+ descriptor_allocator(descriptor_pool, *descriptor_set_layout),
+ pipeline_layout{device.GetLogical().CreatePipelineLayout({
+ .sType = VK_STRUCTURE_TYPE_PIPELINE_LAYOUT_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .setLayoutCount = 1,
+ .pSetLayouts = descriptor_set_layout.address(),
+ .pushConstantRangeCount = 0,
+ .pPushConstantRanges = nullptr,
+ })},
+ descriptor_update_template{
+ CreateDescriptorUpdateTemplate(device, info, *descriptor_set_layout, *pipeline_layout)},
+ pipeline{device.GetLogical().CreateComputePipeline({
+ .sType = VK_STRUCTURE_TYPE_COMPUTE_PIPELINE_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .stage{
+ .sType = VK_STRUCTURE_TYPE_PIPELINE_SHADER_STAGE_CREATE_INFO,
+ .pNext = nullptr,
+ .flags = 0,
+ .stage = VK_SHADER_STAGE_COMPUTE_BIT,
+ .module = *spv_module,
+ .pName = "main",
+ .pSpecializationInfo = nullptr,
+ },
+ .layout = *pipeline_layout,
+ .basePipelineHandle = 0,
+ .basePipelineIndex = 0,
+ })} {}
+
+void ComputePipeline::ConfigureBufferCache(BufferCache& buffer_cache) {
+ u32 enabled_uniforms{};
+ for (const auto& desc : info.constant_buffer_descriptors) {
+ enabled_uniforms |= ((1ULL << desc.count) - 1) << desc.index;
+ }
+ buffer_cache.SetEnabledComputeUniformBuffers(enabled_uniforms);
-ComputePipeline::ComputePipeline() = default;
+ buffer_cache.UnbindComputeStorageBuffers();
+ size_t index{};
+ for (const auto& desc : info.storage_buffers_descriptors) {
+ ASSERT(desc.count == 1);
+ buffer_cache.BindComputeStorageBuffer(index, desc.cbuf_index, desc.cbuf_offset, true);
+ ++index;
+ }
+ buffer_cache.UpdateComputeBuffers();
+ buffer_cache.BindHostComputeBuffers();
+}
-ComputePipeline::~ComputePipeline() = default;
+VkDescriptorSet ComputePipeline::UpdateDescriptorSet() {
+ const VkDescriptorSet descriptor_set{descriptor_allocator.Commit()};
+ update_descriptor_queue->Send(*descriptor_update_template, descriptor_set);
+ return descriptor_set;
+}
} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_compute_pipeline.h b/src/video_core/renderer_vulkan/vk_compute_pipeline.h
index 433d8bb3d..dc045d524 100644
--- a/src/video_core/renderer_vulkan/vk_compute_pipeline.h
+++ b/src/video_core/renderer_vulkan/vk_compute_pipeline.h
@@ -5,19 +5,52 @@
#pragma once
#include "common/common_types.h"
+#include "shader_recompiler/shader_info.h"
+#include "video_core/renderer_vulkan/vk_buffer_cache.h"
#include "video_core/renderer_vulkan/vk_descriptor_pool.h"
+#include "video_core/renderer_vulkan/vk_pipeline.h"
+#include "video_core/renderer_vulkan/vk_update_descriptor.h"
#include "video_core/vulkan_common/vulkan_wrapper.h"
namespace Vulkan {
class Device;
-class VKScheduler;
-class VKUpdateDescriptorQueue;
-class ComputePipeline {
+class ComputePipeline : public Pipeline {
public:
- explicit ComputePipeline();
- ~ComputePipeline();
+ explicit ComputePipeline() = default;
+ explicit ComputePipeline(const Device& device, VKDescriptorPool& descriptor_pool,
+ VKUpdateDescriptorQueue& update_descriptor_queue,
+ const Shader::Info& info, vk::ShaderModule spv_module);
+
+ ComputePipeline& operator=(ComputePipeline&&) noexcept = default;
+ ComputePipeline(ComputePipeline&&) noexcept = default;
+
+ ComputePipeline& operator=(const ComputePipeline&) = delete;
+ ComputePipeline(const ComputePipeline&) = delete;
+
+ void ConfigureBufferCache(BufferCache& buffer_cache);
+
+ [[nodiscard]] VkDescriptorSet UpdateDescriptorSet();
+
+ [[nodiscard]] VkPipeline Handle() const noexcept {
+ return *pipeline;
+ }
+
+ [[nodiscard]] VkPipelineLayout PipelineLayout() const noexcept {
+ return *pipeline_layout;
+ }
+
+private:
+ VKUpdateDescriptorQueue* update_descriptor_queue;
+ Shader::Info info;
+
+ vk::ShaderModule spv_module;
+ vk::DescriptorSetLayout descriptor_set_layout;
+ DescriptorAllocator descriptor_allocator;
+ vk::PipelineLayout pipeline_layout;
+ vk::DescriptorUpdateTemplateKHR descriptor_update_template;
+ vk::Pipeline pipeline;
};
} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_descriptor_pool.cpp b/src/video_core/renderer_vulkan/vk_descriptor_pool.cpp
index ef9fb5910..3bea1ff44 100644
--- a/src/video_core/renderer_vulkan/vk_descriptor_pool.cpp
+++ b/src/video_core/renderer_vulkan/vk_descriptor_pool.cpp
@@ -19,9 +19,7 @@ constexpr std::size_t SETS_GROW_RATE = 0x20;
DescriptorAllocator::DescriptorAllocator(VKDescriptorPool& descriptor_pool_,
VkDescriptorSetLayout layout_)
: ResourcePool(descriptor_pool_.master_semaphore, SETS_GROW_RATE),
- descriptor_pool{descriptor_pool_}, layout{layout_} {}
-
-DescriptorAllocator::~DescriptorAllocator() = default;
+ descriptor_pool{&descriptor_pool_}, layout{layout_} {}
VkDescriptorSet DescriptorAllocator::Commit() {
const std::size_t index = CommitResource();
@@ -29,7 +27,7 @@ VkDescriptorSet DescriptorAllocator::Commit() {
}
void DescriptorAllocator::Allocate(std::size_t begin, std::size_t end) {
- descriptors_allocations.push_back(descriptor_pool.AllocateDescriptors(layout, end - begin));
+ descriptors_allocations.push_back(descriptor_pool->AllocateDescriptors(layout, end - begin));
}
VKDescriptorPool::VKDescriptorPool(const Device& device_, VKScheduler& scheduler)
diff --git a/src/video_core/renderer_vulkan/vk_descriptor_pool.h b/src/video_core/renderer_vulkan/vk_descriptor_pool.h
index f892be7be..2501f9967 100644
--- a/src/video_core/renderer_vulkan/vk_descriptor_pool.h
+++ b/src/video_core/renderer_vulkan/vk_descriptor_pool.h
@@ -17,8 +17,12 @@ class VKScheduler;
class DescriptorAllocator final : public ResourcePool {
public:
+ explicit DescriptorAllocator() = default;
explicit DescriptorAllocator(VKDescriptorPool& descriptor_pool, VkDescriptorSetLayout layout);
- ~DescriptorAllocator() override;
+ ~DescriptorAllocator() override = default;
+
+ DescriptorAllocator& operator=(DescriptorAllocator&&) noexcept = default;
+ DescriptorAllocator(DescriptorAllocator&&) noexcept = default;
DescriptorAllocator& operator=(const DescriptorAllocator&) = delete;
DescriptorAllocator(const DescriptorAllocator&) = delete;
@@ -29,8 +33,8 @@ protected:
void Allocate(std::size_t begin, std::size_t end) override;
private:
- VKDescriptorPool& descriptor_pool;
- const VkDescriptorSetLayout layout;
+ VKDescriptorPool* descriptor_pool{};
+ VkDescriptorSetLayout layout{};
std::vector<vk::DescriptorSets> descriptors_allocations;
};
diff --git a/src/video_core/renderer_vulkan/vk_pipeline.h b/src/video_core/renderer_vulkan/vk_pipeline.h
new file mode 100644
index 000000000..b06288403
--- /dev/null
+++ b/src/video_core/renderer_vulkan/vk_pipeline.h
@@ -0,0 +1,36 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <cstddef>
+
+#include "video_core/vulkan_common/vulkan_wrapper.h"
+
+namespace Vulkan {
+
+class Pipeline {
+public:
+ /// Add a reference count to the pipeline
+ void AddRef() noexcept {
+ ++ref_count;
+ }
+
+ [[nodiscard]] bool RemoveRef() noexcept {
+ --ref_count;
+ return ref_count == 0;
+ }
+
+ [[nodiscard]] u64 UsageTick() const noexcept {
+ return usage_tick;
+ }
+
+protected:
+ u64 usage_tick{};
+
+private:
+ size_t ref_count{};
+};
+
+} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp b/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp
index 7d0ba1180..4bf3e4819 100644
--- a/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp
+++ b/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp
@@ -12,6 +12,8 @@
#include "common/microprofile.h"
#include "core/core.h"
#include "core/memory.h"
+#include "shader_recompiler/environment.h"
+#include "shader_recompiler/recompiler.h"
#include "video_core/engines/kepler_compute.h"
#include "video_core/engines/maxwell_3d.h"
#include "video_core/memory_manager.h"
@@ -22,43 +24,105 @@
#include "video_core/renderer_vulkan/vk_pipeline_cache.h"
#include "video_core/renderer_vulkan/vk_rasterizer.h"
#include "video_core/renderer_vulkan/vk_scheduler.h"
+#include "video_core/renderer_vulkan/vk_shader_util.h"
#include "video_core/renderer_vulkan/vk_update_descriptor.h"
#include "video_core/shader_cache.h"
#include "video_core/shader_notify.h"
#include "video_core/vulkan_common/vulkan_device.h"
#include "video_core/vulkan_common/vulkan_wrapper.h"
+#pragma optimize("", off)
+
namespace Vulkan {
MICROPROFILE_DECLARE(Vulkan_PipelineCache);
using Tegra::Engines::ShaderType;
namespace {
-size_t StageFromProgram(size_t program) {
- return program == 0 ? 0 : program - 1;
-}
+class Environment final : public Shader::Environment {
+public:
+ explicit Environment(Tegra::Engines::KeplerCompute& kepler_compute_,
+ Tegra::MemoryManager& gpu_memory_, GPUVAddr program_base_)
+ : kepler_compute{kepler_compute_}, gpu_memory{gpu_memory_}, program_base{program_base_} {}
+
+ ~Environment() override = default;
+
+ [[nodiscard]] std::optional<u128> Analyze(u32 start_address) {
+ const std::optional<u64> size{TryFindSize(start_address)};
+ if (!size) {
+ return std::nullopt;
+ }
+ cached_lowest = start_address;
+ cached_highest = start_address + static_cast<u32>(*size);
+ return Common::CityHash128(reinterpret_cast<const char*>(code.data()), code.size());
+ }
-ShaderType StageFromProgram(Maxwell::ShaderProgram program) {
- return static_cast<ShaderType>(StageFromProgram(static_cast<size_t>(program)));
-}
+ [[nodiscard]] size_t ShaderSize() const noexcept {
+ return read_highest - read_lowest + INST_SIZE;
+ }
-ShaderType GetShaderType(Maxwell::ShaderProgram program) {
- switch (program) {
- case Maxwell::ShaderProgram::VertexB:
- return ShaderType::Vertex;
- case Maxwell::ShaderProgram::TesselationControl:
- return ShaderType::TesselationControl;
- case Maxwell::ShaderProgram::TesselationEval:
- return ShaderType::TesselationEval;
- case Maxwell::ShaderProgram::Geometry:
- return ShaderType::Geometry;
- case Maxwell::ShaderProgram::Fragment:
- return ShaderType::Fragment;
- default:
- UNIMPLEMENTED_MSG("program={}", program);
- return ShaderType::Vertex;
+ [[nodiscard]] u128 ComputeHash() const {
+ const size_t size{ShaderSize()};
+ auto data = std::make_unique<u64[]>(size);
+ gpu_memory.ReadBlock(program_base + read_lowest, data.get(), size);
+ return Common::CityHash128(reinterpret_cast<const char*>(data.get()), size);
}
-}
+
+ u64 ReadInstruction(u32 address) override {
+ read_lowest = std::min(read_lowest, address);
+ read_highest = std::max(read_highest, address);
+
+ if (address >= cached_lowest && address < cached_highest) {
+ return code[address / INST_SIZE];
+ }
+ return gpu_memory.Read<u64>(program_base + address);
+ }
+
+ std::array<u32, 3> WorkgroupSize() override {
+ const auto& qmd{kepler_compute.launch_description};
+ return {qmd.block_dim_x, qmd.block_dim_y, qmd.block_dim_z};
+ }
+
+private:
+ static constexpr size_t INST_SIZE = sizeof(u64);
+ static constexpr size_t BLOCK_SIZE = 0x1000;
+ static constexpr size_t MAXIMUM_SIZE = 0x100000;
+
+ static constexpr u64 SELF_BRANCH_A = 0xE2400FFFFF87000FULL;
+ static constexpr u64 SELF_BRANCH_B = 0xE2400FFFFF07000FULL;
+
+ std::optional<u64> TryFindSize(u32 start_address) {
+ GPUVAddr guest_addr = program_base + start_address;
+ size_t offset = 0;
+ size_t size = BLOCK_SIZE;
+ while (size <= MAXIMUM_SIZE) {
+ code.resize(size / INST_SIZE);
+ u64* const data = code.data() + offset / INST_SIZE;
+ gpu_memory.ReadBlock(guest_addr, data, BLOCK_SIZE);
+ for (size_t i = 0; i < BLOCK_SIZE; i += INST_SIZE) {
+ const u64 inst = data[i / INST_SIZE];
+ if (inst == SELF_BRANCH_A || inst == SELF_BRANCH_B) {
+ return offset + i;
+ }
+ }
+ guest_addr += BLOCK_SIZE;
+ size += BLOCK_SIZE;
+ offset += BLOCK_SIZE;
+ }
+ return std::nullopt;
+ }
+
+ Tegra::Engines::KeplerCompute& kepler_compute;
+ Tegra::MemoryManager& gpu_memory;
+ GPUVAddr program_base;
+
+ u32 read_lowest = 0;
+ u32 read_highest = 0;
+
+ std::vector<u64> code;
+ u32 cached_lowest = std::numeric_limits<u32>::max();
+ u32 cached_highest = 0;
+};
} // Anonymous namespace
size_t ComputePipelineCacheKey::Hash() const noexcept {
@@ -70,35 +134,91 @@ bool ComputePipelineCacheKey::operator==(const ComputePipelineCacheKey& rhs) con
return std::memcmp(&rhs, this, sizeof *this) == 0;
}
-Shader::Shader() = default;
-
-Shader::~Shader() = default;
-
PipelineCache::PipelineCache(RasterizerVulkan& rasterizer_, Tegra::GPU& gpu_,
Tegra::Engines::Maxwell3D& maxwell3d_,
Tegra::Engines::KeplerCompute& kepler_compute_,
Tegra::MemoryManager& gpu_memory_, const Device& device_,
VKScheduler& scheduler_, VKDescriptorPool& descriptor_pool_,
VKUpdateDescriptorQueue& update_descriptor_queue_)
- : VideoCommon::ShaderCache<Shader>{rasterizer_}, gpu{gpu_}, maxwell3d{maxwell3d_},
+ : VideoCommon::ShaderCache<ShaderInfo>{rasterizer_}, gpu{gpu_}, maxwell3d{maxwell3d_},
kepler_compute{kepler_compute_}, gpu_memory{gpu_memory_}, device{device_},
scheduler{scheduler_}, descriptor_pool{descriptor_pool_}, update_descriptor_queue{
update_descriptor_queue_} {}
PipelineCache::~PipelineCache() = default;
-ComputePipeline& PipelineCache::GetComputePipeline(const ComputePipelineCacheKey& key) {
+ComputePipeline* PipelineCache::CurrentComputePipeline() {
MICROPROFILE_SCOPE(Vulkan_PipelineCache);
- const auto [pair, is_cache_miss] = compute_cache.try_emplace(key);
- auto& entry = pair->second;
- if (!is_cache_miss) {
- return *entry;
+ const GPUVAddr program_base{kepler_compute.regs.code_loc.Address()};
+ const auto& qmd{kepler_compute.launch_description};
+ const GPUVAddr shader_addr{program_base + qmd.program_start};
+ const std::optional<VAddr> cpu_shader_addr{gpu_memory.GpuToCpuAddress(shader_addr)};
+ if (!cpu_shader_addr) {
+ return nullptr;
+ }
+ ShaderInfo* const shader{TryGet(*cpu_shader_addr)};
+ if (!shader) {
+ return CreateComputePipelineWithoutShader(*cpu_shader_addr);
+ }
+ const ComputePipelineCacheKey key{MakeComputePipelineKey(shader->unique_hash)};
+ const auto [pair, is_new]{compute_cache.try_emplace(key)};
+ auto& pipeline{pair->second};
+ if (!is_new) {
+ return &pipeline;
+ }
+ pipeline = CreateComputePipeline(shader);
+ shader->compute_users.push_back(key);
+ return &pipeline;
+}
+
+ComputePipeline PipelineCache::CreateComputePipeline(ShaderInfo* shader_info) {
+ const GPUVAddr program_base{kepler_compute.regs.code_loc.Address()};
+ const auto& qmd{kepler_compute.launch_description};
+ Environment env{kepler_compute, gpu_memory, program_base};
+ if (const std::optional<u128> cached_hash{env.Analyze(qmd.program_start)}) {
+ // TODO: Load from cache
}
- LOG_INFO(Render_Vulkan, "Compile 0x{:016X}", key.Hash());
- throw "Bad";
+ const auto [info, code]{Shader::RecompileSPIRV(env, qmd.program_start)};
+ shader_info->unique_hash = env.ComputeHash();
+ shader_info->size_bytes = env.ShaderSize();
+ return ComputePipeline{device, descriptor_pool, update_descriptor_queue, info,
+ BuildShader(device, code)};
}
-void PipelineCache::OnShaderRemoval(Shader*) {}
+ComputePipeline* PipelineCache::CreateComputePipelineWithoutShader(VAddr shader_cpu_addr) {
+ ShaderInfo shader;
+ ComputePipeline pipeline{CreateComputePipeline(&shader)};
+ const ComputePipelineCacheKey key{MakeComputePipelineKey(shader.unique_hash)};
+ shader.compute_users.push_back(key);
+ pipeline.AddRef();
+
+ const size_t size_bytes{shader.size_bytes};
+ Register(std::make_unique<ShaderInfo>(std::move(shader)), shader_cpu_addr, size_bytes);
+ return &compute_cache.emplace(key, std::move(pipeline)).first->second;
+}
+
+ComputePipelineCacheKey PipelineCache::MakeComputePipelineKey(u128 unique_hash) const {
+ const auto& qmd{kepler_compute.launch_description};
+ return {
+ .unique_hash = unique_hash,
+ .shared_memory_size = qmd.shared_alloc,
+ .workgroup_size{qmd.block_dim_x, qmd.block_dim_y, qmd.block_dim_z},
+ };
+}
+
+void PipelineCache::OnShaderRemoval(ShaderInfo* shader) {
+ for (const ComputePipelineCacheKey& key : shader->compute_users) {
+ const auto it = compute_cache.find(key);
+ ASSERT(it != compute_cache.end());
+
+ Pipeline& pipeline = it->second;
+ if (pipeline.RemoveRef()) {
+ // Wait for the pipeline to be free of GPU usage before destroying it
+ scheduler.Wait(pipeline.UsageTick());
+ compute_cache.erase(it);
+ }
+ }
+}
} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_pipeline_cache.h b/src/video_core/renderer_vulkan/vk_pipeline_cache.h
index e3e63340d..eb35abc27 100644
--- a/src/video_core/renderer_vulkan/vk_pipeline_cache.h
+++ b/src/video_core/renderer_vulkan/vk_pipeline_cache.h
@@ -36,7 +36,7 @@ class VKUpdateDescriptorQueue;
using Maxwell = Tegra::Engines::Maxwell3D::Regs;
struct ComputePipelineCacheKey {
- GPUVAddr shader;
+ u128 unique_hash;
u32 shared_memory_size;
std::array<u32, 3> workgroup_size;
@@ -67,13 +67,13 @@ struct hash<Vulkan::ComputePipelineCacheKey> {
namespace Vulkan {
-class Shader {
-public:
- explicit Shader();
- ~Shader();
+struct ShaderInfo {
+ u128 unique_hash{};
+ size_t size_bytes{};
+ std::vector<ComputePipelineCacheKey> compute_users;
};
-class PipelineCache final : public VideoCommon::ShaderCache<Shader> {
+class PipelineCache final : public VideoCommon::ShaderCache<ShaderInfo> {
public:
explicit PipelineCache(RasterizerVulkan& rasterizer, Tegra::GPU& gpu,
Tegra::Engines::Maxwell3D& maxwell3d,
@@ -83,12 +83,18 @@ public:
VKUpdateDescriptorQueue& update_descriptor_queue);
~PipelineCache() override;
- ComputePipeline& GetComputePipeline(const ComputePipelineCacheKey& key);
+ [[nodiscard]] ComputePipeline* CurrentComputePipeline();
protected:
- void OnShaderRemoval(Shader* shader) final;
+ void OnShaderRemoval(ShaderInfo* shader) override;
private:
+ ComputePipeline CreateComputePipeline(ShaderInfo* shader);
+
+ ComputePipeline* CreateComputePipelineWithoutShader(VAddr shader_cpu_addr);
+
+ ComputePipelineCacheKey MakeComputePipelineKey(u128 unique_hash) const;
+
Tegra::GPU& gpu;
Tegra::Engines::Maxwell3D& maxwell3d;
Tegra::Engines::KeplerCompute& kepler_compute;
@@ -99,13 +105,7 @@ private:
VKDescriptorPool& descriptor_pool;
VKUpdateDescriptorQueue& update_descriptor_queue;
- std::unique_ptr<Shader> null_shader;
- std::unique_ptr<Shader> null_kernel;
-
- std::array<Shader*, Maxwell::MaxShaderProgram> last_shaders{};
-
- std::mutex pipeline_cache;
- std::unordered_map<ComputePipelineCacheKey, std::unique_ptr<ComputePipeline>> compute_cache;
+ std::unordered_map<ComputePipelineCacheKey, ComputePipeline> compute_cache;
};
} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.cpp b/src/video_core/renderer_vulkan/vk_rasterizer.cpp
index f152297d9..b757454c4 100644
--- a/src/video_core/renderer_vulkan/vk_rasterizer.cpp
+++ b/src/video_core/renderer_vulkan/vk_rasterizer.cpp
@@ -36,6 +36,8 @@
#include "video_core/vulkan_common/vulkan_device.h"
#include "video_core/vulkan_common/vulkan_wrapper.h"
+#pragma optimize("", off)
+
namespace Vulkan {
using Maxwell = Tegra::Engines::Maxwell3D::Regs;
@@ -237,7 +239,26 @@ void RasterizerVulkan::Clear() {
}
void RasterizerVulkan::DispatchCompute() {
- UNREACHABLE_MSG("Not implemented");
+ ComputePipeline* const pipeline{pipeline_cache.CurrentComputePipeline()};
+ if (!pipeline) {
+ return;
+ }
+ std::scoped_lock lock{buffer_cache.mutex};
+ update_descriptor_queue.Acquire();
+ pipeline->ConfigureBufferCache(buffer_cache);
+ const VkDescriptorSet descriptor_set{pipeline->UpdateDescriptorSet()};
+
+ const auto& qmd{kepler_compute.launch_description};
+ const std::array<u32, 3> dim{qmd.grid_dim_x, qmd.grid_dim_y, qmd.grid_dim_z};
+ const VkPipeline pipeline_handle{pipeline->Handle()};
+ const VkPipelineLayout pipeline_layout{pipeline->PipelineLayout()};
+ scheduler.Record(
+ [pipeline_handle, pipeline_layout, dim, descriptor_set](vk::CommandBuffer cmdbuf) {
+ cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_COMPUTE, pipeline_handle);
+ cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_COMPUTE, pipeline_layout, 0,
+ descriptor_set, nullptr);
+ cmdbuf.Dispatch(dim[0], dim[1], dim[2]);
+ });
}
void RasterizerVulkan::ResetCounter(VideoCore::QueryType type) {
diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.h b/src/video_core/renderer_vulkan/vk_rasterizer.h
index 31017dc2b..3fd03b915 100644
--- a/src/video_core/renderer_vulkan/vk_rasterizer.h
+++ b/src/video_core/renderer_vulkan/vk_rasterizer.h
@@ -21,7 +21,6 @@
#include "video_core/renderer_vulkan/vk_buffer_cache.h"
#include "video_core/renderer_vulkan/vk_descriptor_pool.h"
#include "video_core/renderer_vulkan/vk_fence_manager.h"
-#include "video_core/renderer_vulkan/vk_graphics_pipeline.h"
#include "video_core/renderer_vulkan/vk_pipeline_cache.h"
#include "video_core/renderer_vulkan/vk_query_cache.h"
#include "video_core/renderer_vulkan/vk_scheduler.h"
@@ -150,8 +149,6 @@ private:
BlitImageHelper blit_image;
ASTCDecoderPass astc_decoder_pass;
- GraphicsPipelineCacheKey graphics_key;
-
TextureCacheRuntime texture_cache_runtime;
TextureCache texture_cache;
BufferCacheRuntime buffer_cache_runtime;
diff --git a/src/video_core/renderer_vulkan/vk_resource_pool.cpp b/src/video_core/renderer_vulkan/vk_resource_pool.cpp
index a8bf7bda8..2dd514968 100644
--- a/src/video_core/renderer_vulkan/vk_resource_pool.cpp
+++ b/src/video_core/renderer_vulkan/vk_resource_pool.cpp
@@ -10,18 +10,16 @@
namespace Vulkan {
ResourcePool::ResourcePool(MasterSemaphore& master_semaphore_, size_t grow_step_)
- : master_semaphore{master_semaphore_}, grow_step{grow_step_} {}
-
-ResourcePool::~ResourcePool() = default;
+ : master_semaphore{&master_semaphore_}, grow_step{grow_step_} {}
size_t ResourcePool::CommitResource() {
// Refresh semaphore to query updated results
- master_semaphore.Refresh();
- const u64 gpu_tick = master_semaphore.KnownGpuTick();
+ master_semaphore->Refresh();
+ const u64 gpu_tick = master_semaphore->KnownGpuTick();
const auto search = [this, gpu_tick](size_t begin, size_t end) -> std::optional<size_t> {
for (size_t iterator = begin; iterator < end; ++iterator) {
if (gpu_tick >= ticks[iterator]) {
- ticks[iterator] = master_semaphore.CurrentTick();
+ ticks[iterator] = master_semaphore->CurrentTick();
return iterator;
}
}
@@ -36,7 +34,7 @@ size_t ResourcePool::CommitResource() {
// Both searches failed, the pool is full; handle it.
const size_t free_resource = ManageOverflow();
- ticks[free_resource] = master_semaphore.CurrentTick();
+ ticks[free_resource] = master_semaphore->CurrentTick();
found = free_resource;
}
}
diff --git a/src/video_core/renderer_vulkan/vk_resource_pool.h b/src/video_core/renderer_vulkan/vk_resource_pool.h
index 9d0bb3b4d..f0b80ad59 100644
--- a/src/video_core/renderer_vulkan/vk_resource_pool.h
+++ b/src/video_core/renderer_vulkan/vk_resource_pool.h
@@ -18,8 +18,16 @@ class MasterSemaphore;
*/
class ResourcePool {
public:
+ explicit ResourcePool() = default;
explicit ResourcePool(MasterSemaphore& master_semaphore, size_t grow_step);
- virtual ~ResourcePool();
+
+ virtual ~ResourcePool() = default;
+
+ ResourcePool& operator=(ResourcePool&&) noexcept = default;
+ ResourcePool(ResourcePool&&) noexcept = default;
+
+ ResourcePool& operator=(const ResourcePool&) = default;
+ ResourcePool(const ResourcePool&) = default;
protected:
size_t CommitResource();
@@ -34,7 +42,7 @@ private:
/// Allocates a new page of resources.
void Grow();
- MasterSemaphore& master_semaphore;
+ MasterSemaphore* master_semaphore{};
size_t grow_step = 0; ///< Number of new resources created after an overflow
size_t hint_iterator = 0; ///< Hint to where the next free resources is likely to be found
std::vector<u64> ticks; ///< Ticks for each resource