mirror of
https://github.com/shadps4-emu/shadPS4.git
synced 2025-07-22 10:04:39 +00:00
Merge 026e29f7e6
into 8bc30270c8
This commit is contained in:
commit
844dfad867
@ -271,6 +271,9 @@ void SetupCapabilities(const Info& info, const Profile& profile, EmitContext& ct
|
||||
if (info.has_image_query) {
|
||||
ctx.AddCapability(spv::Capability::ImageQuery);
|
||||
}
|
||||
if (info.has_layer_output) {
|
||||
ctx.AddCapability(spv::Capability::ShaderLayer);
|
||||
}
|
||||
if ((info.uses_image_atomic_float_min_max && profile.supports_image_fp32_atomic_min_max) ||
|
||||
(info.uses_buffer_atomic_float_min_max && profile.supports_buffer_fp32_atomic_min_max)) {
|
||||
ctx.AddExtension("SPV_EXT_shader_atomic_float_min_max");
|
||||
|
@ -16,39 +16,6 @@
|
||||
namespace Shader::Backend::SPIRV {
|
||||
namespace {
|
||||
|
||||
Id VsOutputAttrPointer(EmitContext& ctx, VsOutput output) {
|
||||
switch (output) {
|
||||
case VsOutput::ClipDist0:
|
||||
case VsOutput::ClipDist1:
|
||||
case VsOutput::ClipDist2:
|
||||
case VsOutput::ClipDist3:
|
||||
case VsOutput::ClipDist4:
|
||||
case VsOutput::ClipDist5:
|
||||
case VsOutput::ClipDist6:
|
||||
case VsOutput::ClipDist7: {
|
||||
const u32 index = u32(output) - u32(VsOutput::ClipDist0);
|
||||
const Id clip_num{ctx.ConstU32(index)};
|
||||
ASSERT_MSG(Sirit::ValidId(ctx.clip_distances), "Clip distance used but not defined");
|
||||
return ctx.OpAccessChain(ctx.output_f32, ctx.clip_distances, clip_num);
|
||||
}
|
||||
case VsOutput::CullDist0:
|
||||
case VsOutput::CullDist1:
|
||||
case VsOutput::CullDist2:
|
||||
case VsOutput::CullDist3:
|
||||
case VsOutput::CullDist4:
|
||||
case VsOutput::CullDist5:
|
||||
case VsOutput::CullDist6:
|
||||
case VsOutput::CullDist7: {
|
||||
const u32 index = u32(output) - u32(VsOutput::CullDist0);
|
||||
const Id cull_num{ctx.ConstU32(index)};
|
||||
ASSERT_MSG(Sirit::ValidId(ctx.cull_distances), "Cull distance used but not defined");
|
||||
return ctx.OpAccessChain(ctx.output_f32, ctx.cull_distances, cull_num);
|
||||
}
|
||||
default:
|
||||
UNREACHABLE();
|
||||
}
|
||||
}
|
||||
|
||||
Id OutputAttrPointer(EmitContext& ctx, IR::Attribute attr, u32 element) {
|
||||
if (IR::IsParam(attr)) {
|
||||
const u32 attr_index{u32(attr) - u32(IR::Attribute::Param0)};
|
||||
@ -76,19 +43,18 @@ Id OutputAttrPointer(EmitContext& ctx, IR::Attribute attr, u32 element) {
|
||||
}
|
||||
}
|
||||
switch (attr) {
|
||||
case IR::Attribute::Position0: {
|
||||
case IR::Attribute::Position0:
|
||||
return ctx.OpAccessChain(ctx.output_f32, ctx.output_position, ctx.ConstU32(element));
|
||||
}
|
||||
case IR::Attribute::Position1:
|
||||
case IR::Attribute::Position2:
|
||||
case IR::Attribute::Position3: {
|
||||
const u32 index = u32(attr) - u32(IR::Attribute::Position1);
|
||||
return VsOutputAttrPointer(ctx, ctx.runtime_info.vs_info.outputs[index][element]);
|
||||
}
|
||||
case IR::Attribute::ClipDistance:
|
||||
return ctx.OpAccessChain(ctx.output_f32, ctx.clip_distances, ctx.ConstU32(element));
|
||||
case IR::Attribute::CullDistance:
|
||||
return ctx.OpAccessChain(ctx.output_f32, ctx.cull_distances, ctx.ConstU32(element));
|
||||
case IR::Attribute::RenderTargetId:
|
||||
return ctx.output_layer;
|
||||
case IR::Attribute::Depth:
|
||||
return ctx.frag_depth;
|
||||
default:
|
||||
throw NotImplementedException("Write attribute {}", attr);
|
||||
UNREACHABLE_MSG("Write attribute {}", attr);
|
||||
}
|
||||
}
|
||||
|
||||
@ -109,13 +75,15 @@ std::pair<Id, bool> OutputAttrComponentType(EmitContext& ctx, IR::Attribute attr
|
||||
}
|
||||
switch (attr) {
|
||||
case IR::Attribute::Position0:
|
||||
case IR::Attribute::Position1:
|
||||
case IR::Attribute::Position2:
|
||||
case IR::Attribute::Position3:
|
||||
case IR::Attribute::ClipDistance:
|
||||
case IR::Attribute::CullDistance:
|
||||
case IR::Attribute::Depth:
|
||||
return {ctx.F32[1], false};
|
||||
case IR::Attribute::RenderTargetId:
|
||||
case IR::Attribute::ViewportId:
|
||||
return {ctx.S32[1], true};
|
||||
default:
|
||||
throw NotImplementedException("Write attribute {}", attr);
|
||||
UNREACHABLE_MSG("Write attribute {}", attr);
|
||||
}
|
||||
}
|
||||
} // Anonymous namespace
|
||||
@ -350,14 +318,10 @@ Id EmitGetAttributeU32(EmitContext& ctx, IR::Attribute attr, u32 comp) {
|
||||
}
|
||||
|
||||
void EmitSetAttribute(EmitContext& ctx, IR::Attribute attr, Id value, u32 element) {
|
||||
if (attr == IR::Attribute::Position1) {
|
||||
LOG_WARNING(Render_Vulkan, "Ignoring pos1 export");
|
||||
return;
|
||||
}
|
||||
const Id pointer{OutputAttrPointer(ctx, attr, element)};
|
||||
const auto component_type{OutputAttrComponentType(ctx, attr)};
|
||||
if (component_type.second) {
|
||||
ctx.OpStore(pointer, ctx.OpBitcast(component_type.first, value));
|
||||
const auto [component_type, is_integer]{OutputAttrComponentType(ctx, attr)};
|
||||
if (is_integer) {
|
||||
ctx.OpStore(pointer, ctx.OpBitcast(component_type, value));
|
||||
} else {
|
||||
ctx.OpStore(pointer, value);
|
||||
}
|
||||
|
@ -556,23 +556,25 @@ void EmitContext::DefineInputs() {
|
||||
}
|
||||
}
|
||||
|
||||
void EmitContext::DefineVertexBlock() {
|
||||
output_position = DefineVariable(F32[4], spv::BuiltIn::Position, spv::StorageClass::Output);
|
||||
if (info.stores.GetAny(IR::Attribute::ClipDistance)) {
|
||||
clip_distances = DefineVariable(TypeArray(F32[1], ConstU32(8U)), spv::BuiltIn::ClipDistance,
|
||||
spv::StorageClass::Output);
|
||||
}
|
||||
if (info.stores.GetAny(IR::Attribute::CullDistance)) {
|
||||
cull_distances = DefineVariable(TypeArray(F32[1], ConstU32(8U)), spv::BuiltIn::CullDistance,
|
||||
spv::StorageClass::Output);
|
||||
}
|
||||
if (info.stores.GetAny(IR::Attribute::RenderTargetId)) {
|
||||
output_layer = DefineVariable(S32[1], spv::BuiltIn::Layer, spv::StorageClass::Output);
|
||||
}
|
||||
}
|
||||
|
||||
void EmitContext::DefineOutputs() {
|
||||
switch (l_stage) {
|
||||
case LogicalStage::Vertex: {
|
||||
// No point in defining builtin outputs (i.e. position) unless next stage is fragment?
|
||||
// Might cause problems linking with tcs
|
||||
|
||||
output_position = DefineVariable(F32[4], spv::BuiltIn::Position, spv::StorageClass::Output);
|
||||
const bool has_extra_pos_stores = info.stores.Get(IR::Attribute::Position1) ||
|
||||
info.stores.Get(IR::Attribute::Position2) ||
|
||||
info.stores.Get(IR::Attribute::Position3);
|
||||
if (has_extra_pos_stores) {
|
||||
const Id type{TypeArray(F32[1], ConstU32(8U))};
|
||||
clip_distances =
|
||||
DefineVariable(type, spv::BuiltIn::ClipDistance, spv::StorageClass::Output);
|
||||
cull_distances =
|
||||
DefineVariable(type, spv::BuiltIn::CullDistance, spv::StorageClass::Output);
|
||||
}
|
||||
DefineVertexBlock();
|
||||
if (stage == Shader::Stage::Local && runtime_info.ls_info.links_with_tcs) {
|
||||
const u32 num_attrs = Common::AlignUp(runtime_info.ls_info.ls_stride, 16) >> 4;
|
||||
if (num_attrs > 0) {
|
||||
@ -632,17 +634,7 @@ void EmitContext::DefineOutputs() {
|
||||
break;
|
||||
}
|
||||
case LogicalStage::TessellationEval: {
|
||||
output_position = DefineVariable(F32[4], spv::BuiltIn::Position, spv::StorageClass::Output);
|
||||
const bool has_extra_pos_stores = info.stores.Get(IR::Attribute::Position1) ||
|
||||
info.stores.Get(IR::Attribute::Position2) ||
|
||||
info.stores.Get(IR::Attribute::Position3);
|
||||
if (has_extra_pos_stores) {
|
||||
const Id type{TypeArray(F32[1], ConstU32(8U))};
|
||||
clip_distances =
|
||||
DefineVariable(type, spv::BuiltIn::ClipDistance, spv::StorageClass::Output);
|
||||
cull_distances =
|
||||
DefineVariable(type, spv::BuiltIn::CullDistance, spv::StorageClass::Output);
|
||||
}
|
||||
DefineVertexBlock();
|
||||
for (u32 i = 0; i < IR::NumParams; i++) {
|
||||
const IR::Attribute param{IR::Attribute::Param0 + i};
|
||||
if (!info.stores.GetAny(param)) {
|
||||
@ -682,8 +674,7 @@ void EmitContext::DefineOutputs() {
|
||||
break;
|
||||
}
|
||||
case LogicalStage::Geometry: {
|
||||
output_position = DefineVariable(F32[4], spv::BuiltIn::Position, spv::StorageClass::Output);
|
||||
|
||||
DefineVertexBlock();
|
||||
for (u32 attr_id = 0; attr_id < info.gs_copy_data.num_attrs; attr_id++) {
|
||||
const Id id{DefineOutput(F32[4], attr_id)};
|
||||
Name(id, fmt::format("out_attr{}", attr_id));
|
||||
|
@ -245,6 +245,7 @@ public:
|
||||
boost::container::small_vector<Id, 16> interfaces;
|
||||
|
||||
Id output_position{};
|
||||
Id output_layer{};
|
||||
Id primitive_id{};
|
||||
Id vertex_index{};
|
||||
Id instance_id{};
|
||||
@ -383,6 +384,7 @@ private:
|
||||
void DefineArithmeticTypes();
|
||||
void DefineInterfaces();
|
||||
void DefineInputs();
|
||||
void DefineVertexBlock();
|
||||
void DefineOutputs();
|
||||
void DefinePushDataBlock();
|
||||
void DefineBuffers();
|
||||
|
@ -2,6 +2,7 @@
|
||||
// SPDX-License-Identifier: GPL-2.0-or-later
|
||||
|
||||
#include "shader_recompiler/frontend/translate/translate.h"
|
||||
#include "shader_recompiler/ir/position.h"
|
||||
#include "shader_recompiler/ir/reinterpret.h"
|
||||
#include "shader_recompiler/runtime_info.h"
|
||||
|
||||
@ -126,8 +127,10 @@ void Translator::ExportCompressed(IR::Attribute attribute, u32 idx, const IR::U3
|
||||
|
||||
void Translator::ExportUncompressed(IR::Attribute attribute, u32 comp, const IR::F32& value) {
|
||||
if (IsMrt(attribute)) {
|
||||
ExportMrtUncompressed(attribute, comp, value);
|
||||
return;
|
||||
return ExportMrtUncompressed(attribute, comp, value);
|
||||
}
|
||||
if (IsPosition(attribute)) {
|
||||
return IR::ExportPosition(ir, runtime_info.vs_info, attribute, comp, value);
|
||||
}
|
||||
ir.SetAttribute(attribute, value, comp);
|
||||
}
|
||||
|
@ -317,6 +317,7 @@ private:
|
||||
const PsColorBuffer& color_buffer);
|
||||
void ExportMrtCompressed(IR::Attribute attribute, u32 idx, const IR::U32& value);
|
||||
void ExportMrtUncompressed(IR::Attribute attribute, u32 comp, const IR::F32& value);
|
||||
void ExportPosition(IR::Attribute attribute, u32 comp, const IR::F32& value);
|
||||
void ExportCompressed(IR::Attribute attribute, u32 idx, const IR::U32& value);
|
||||
void ExportUncompressed(IR::Attribute attribute, u32 comp, const IR::F32& value);
|
||||
|
||||
|
@ -215,6 +215,7 @@ struct Info {
|
||||
bool has_image_query{};
|
||||
bool has_perspective_interp{};
|
||||
bool has_linear_interp{};
|
||||
bool has_layer_output{};
|
||||
bool uses_buffer_atomic_float_min_max{};
|
||||
bool uses_image_atomic_float_min_max{};
|
||||
bool uses_lane_id{};
|
||||
|
@ -4,6 +4,7 @@
|
||||
#include "common/assert.h"
|
||||
#include "shader_recompiler/ir/ir_emitter.h"
|
||||
#include "shader_recompiler/ir/opcodes.h"
|
||||
#include "shader_recompiler/ir/position.h"
|
||||
#include "shader_recompiler/ir/program.h"
|
||||
#include "shader_recompiler/ir/reg.h"
|
||||
#include "shader_recompiler/recompiler.h"
|
||||
@ -145,11 +146,12 @@ void RingAccessElimination(const IR::Program& program, const RuntimeInfo& runtim
|
||||
ASSERT(it != info.gs_copy_data.attr_map.cend());
|
||||
const auto& [attr, comp] = it->second;
|
||||
|
||||
inst.ReplaceOpcode(IR::Opcode::SetAttribute);
|
||||
inst.ClearArgs();
|
||||
inst.SetArg(0, IR::Value{attr});
|
||||
inst.SetArg(1, data);
|
||||
inst.SetArg(2, ir.Imm32(comp));
|
||||
inst.Invalidate();
|
||||
if (IsPosition(attr)) {
|
||||
ExportPosition(ir, runtime_info.gs_info, attr, comp, data);
|
||||
} else {
|
||||
ir.SetAttribute(attr, data, comp);
|
||||
}
|
||||
break;
|
||||
}
|
||||
default:
|
||||
|
@ -157,6 +157,10 @@ void CollectShaderInfoPass(IR::Program& program, const Profile& profile) {
|
||||
}
|
||||
}
|
||||
|
||||
if (info.stores.GetAny(IR::Attribute::RenderTargetId)) {
|
||||
info.has_layer_output = true;
|
||||
}
|
||||
|
||||
// In case Flatbuf has not already been bound by IR and is needed
|
||||
// to query buffer sizes, bind it now.
|
||||
if (!profile.supports_robust_buffer_access && !info.uses_dma) {
|
||||
|
53
src/shader_recompiler/ir/position.h
Normal file
53
src/shader_recompiler/ir/position.h
Normal file
@ -0,0 +1,53 @@
|
||||
// SPDX-FileCopyrightText: Copyright 2025 shadPS4 Emulator Project
|
||||
// SPDX-License-Identifier: GPL-2.0-or-later
|
||||
|
||||
#pragma once
|
||||
|
||||
#include "shader_recompiler/ir/ir_emitter.h"
|
||||
#include "shader_recompiler/runtime_info.h"
|
||||
|
||||
namespace Shader::IR {
|
||||
|
||||
/// Maps special position export to builtin attribute stores
|
||||
inline void ExportPosition(IREmitter& ir, const auto& stage, Attribute attribute, u32 comp,
|
||||
const IR::F32& value) {
|
||||
if (attribute == Attribute::Position0) {
|
||||
ir.SetAttribute(attribute, value, comp);
|
||||
return;
|
||||
}
|
||||
const u32 index = u32(attribute) - u32(Attribute::Position1);
|
||||
const auto output = stage.outputs[index][comp];
|
||||
switch (output) {
|
||||
case Output::ClipDist0:
|
||||
case Output::ClipDist1:
|
||||
case Output::ClipDist2:
|
||||
case Output::ClipDist3:
|
||||
case Output::ClipDist4:
|
||||
case Output::ClipDist5:
|
||||
case Output::ClipDist6:
|
||||
case Output::ClipDist7: {
|
||||
const u32 index = u32(output) - u32(Output::ClipDist0);
|
||||
ir.SetAttribute(IR::Attribute::ClipDistance, value, index);
|
||||
break;
|
||||
}
|
||||
case Output::CullDist0:
|
||||
case Output::CullDist1:
|
||||
case Output::CullDist2:
|
||||
case Output::CullDist3:
|
||||
case Output::CullDist4:
|
||||
case Output::CullDist5:
|
||||
case Output::CullDist6:
|
||||
case Output::CullDist7: {
|
||||
const u32 index = u32(output) - u32(Output::CullDist0);
|
||||
ir.SetAttribute(IR::Attribute::CullDistance, value, index);
|
||||
break;
|
||||
}
|
||||
case Output::GsMrtIndex:
|
||||
ir.SetAttribute(IR::Attribute::RenderTargetId, value);
|
||||
break;
|
||||
default:
|
||||
UNREACHABLE_MSG("Unhandled output {} on attribute {}", u32(output), u32(attribute));
|
||||
}
|
||||
}
|
||||
|
||||
} // namespace Shader::IR
|
@ -53,7 +53,7 @@ struct ExportRuntimeInfo {
|
||||
auto operator<=>(const ExportRuntimeInfo&) const noexcept = default;
|
||||
};
|
||||
|
||||
enum class VsOutput : u8 {
|
||||
enum class Output : u8 {
|
||||
None,
|
||||
PointSprite,
|
||||
EdgeFlag,
|
||||
@ -78,11 +78,11 @@ enum class VsOutput : u8 {
|
||||
ClipDist6,
|
||||
ClipDist7,
|
||||
};
|
||||
using VsOutputMap = std::array<VsOutput, 4>;
|
||||
using OutputMap = std::array<Output, 4>;
|
||||
|
||||
struct VertexRuntimeInfo {
|
||||
u32 num_outputs;
|
||||
std::array<VsOutputMap, 3> outputs;
|
||||
std::array<OutputMap, 3> outputs;
|
||||
bool emulate_depth_negative_one_to_one{};
|
||||
bool clip_disable{};
|
||||
// Domain
|
||||
@ -143,6 +143,8 @@ struct HullRuntimeInfo {
|
||||
static constexpr auto GsMaxOutputStreams = 4u;
|
||||
using GsOutputPrimTypes = std::array<AmdGpu::GsOutputPrimitiveType, GsMaxOutputStreams>;
|
||||
struct GeometryRuntimeInfo {
|
||||
u32 num_outputs;
|
||||
std::array<OutputMap, 3> outputs;
|
||||
u32 num_invocations{};
|
||||
u32 output_vertices{};
|
||||
u32 in_vertex_data_size{};
|
||||
|
@ -397,6 +397,7 @@ bool Instance::CreateDevice() {
|
||||
.hostQueryReset = vk12_features.hostQueryReset,
|
||||
.timelineSemaphore = vk12_features.timelineSemaphore,
|
||||
.bufferDeviceAddress = vk12_features.bufferDeviceAddress,
|
||||
.shaderOutputLayer = vk12_features.shaderOutputLayer,
|
||||
},
|
||||
vk::PhysicalDeviceVulkan13Features{
|
||||
.robustImageAccess = vk13_features.robustImageAccess,
|
||||
|
@ -23,8 +23,8 @@ extern std::unique_ptr<Vulkan::Presenter> presenter;
|
||||
namespace Vulkan {
|
||||
|
||||
using Shader::LogicalStage;
|
||||
using Shader::Output;
|
||||
using Shader::Stage;
|
||||
using Shader::VsOutput;
|
||||
|
||||
constexpr static auto SpirvVersion1_6 = 0x00010600U;
|
||||
|
||||
@ -35,49 +35,55 @@ constexpr static std::array DescriptorHeapSizes = {
|
||||
vk::DescriptorPoolSize{vk::DescriptorType::eSampler, 1024},
|
||||
};
|
||||
|
||||
void GatherVertexOutputs(Shader::VertexRuntimeInfo& info,
|
||||
const AmdGpu::Liverpool::VsOutputControl& ctl) {
|
||||
const auto add_output = [&](VsOutput x, VsOutput y, VsOutput z, VsOutput w) {
|
||||
if (x != VsOutput::None || y != VsOutput::None || z != VsOutput::None ||
|
||||
w != VsOutput::None) {
|
||||
info.outputs[info.num_outputs++] = Shader::VsOutputMap{x, y, z, w};
|
||||
}
|
||||
};
|
||||
// VS_OUT_MISC_VEC
|
||||
add_output(ctl.use_vtx_point_size ? VsOutput::PointSprite : VsOutput::None,
|
||||
ctl.use_vtx_edge_flag
|
||||
? VsOutput::EdgeFlag
|
||||
: (ctl.use_vtx_gs_cut_flag ? VsOutput::GsCutFlag : VsOutput::None),
|
||||
ctl.use_vtx_kill_flag
|
||||
? VsOutput::KillFlag
|
||||
: (ctl.use_vtx_render_target_idx ? VsOutput::GsMrtIndex : VsOutput::None),
|
||||
ctl.use_vtx_viewport_idx ? VsOutput::GsVpIndex : VsOutput::None);
|
||||
// VS_OUT_CCDIST0
|
||||
add_output(ctl.IsClipDistEnabled(0)
|
||||
? VsOutput::ClipDist0
|
||||
: (ctl.IsCullDistEnabled(0) ? VsOutput::CullDist0 : VsOutput::None),
|
||||
ctl.IsClipDistEnabled(1)
|
||||
? VsOutput::ClipDist1
|
||||
: (ctl.IsCullDistEnabled(1) ? VsOutput::CullDist1 : VsOutput::None),
|
||||
ctl.IsClipDistEnabled(2)
|
||||
? VsOutput::ClipDist2
|
||||
: (ctl.IsCullDistEnabled(2) ? VsOutput::CullDist2 : VsOutput::None),
|
||||
ctl.IsClipDistEnabled(3)
|
||||
? VsOutput::ClipDist3
|
||||
: (ctl.IsCullDistEnabled(3) ? VsOutput::CullDist3 : VsOutput::None));
|
||||
// VS_OUT_CCDIST1
|
||||
add_output(ctl.IsClipDistEnabled(4)
|
||||
? VsOutput::ClipDist4
|
||||
: (ctl.IsCullDistEnabled(4) ? VsOutput::CullDist4 : VsOutput::None),
|
||||
ctl.IsClipDistEnabled(5)
|
||||
? VsOutput::ClipDist5
|
||||
: (ctl.IsCullDistEnabled(5) ? VsOutput::CullDist5 : VsOutput::None),
|
||||
ctl.IsClipDistEnabled(6)
|
||||
? VsOutput::ClipDist6
|
||||
: (ctl.IsCullDistEnabled(6) ? VsOutput::CullDist6 : VsOutput::None),
|
||||
ctl.IsClipDistEnabled(7)
|
||||
? VsOutput::ClipDist7
|
||||
: (ctl.IsCullDistEnabled(7) ? VsOutput::CullDist7 : VsOutput::None));
|
||||
u32 MapOutputs(std::span<Shader::OutputMap, 3> outputs,
|
||||
const AmdGpu::Liverpool::VsOutputControl& ctl) {
|
||||
u32 num_outputs = 0;
|
||||
|
||||
if (ctl.vs_out_misc_enable) {
|
||||
auto& misc_vec = outputs[num_outputs++];
|
||||
misc_vec[0] = ctl.use_vtx_point_size ? Output::PointSprite : Output::None;
|
||||
misc_vec[1] = ctl.use_vtx_edge_flag
|
||||
? Output::EdgeFlag
|
||||
: (ctl.use_vtx_gs_cut_flag ? Output::GsCutFlag : Output::None);
|
||||
misc_vec[2] = ctl.use_vtx_kill_flag
|
||||
? Output::KillFlag
|
||||
: (ctl.use_vtx_render_target_idx ? Output::GsMrtIndex : Output::None);
|
||||
misc_vec[3] = ctl.use_vtx_viewport_idx ? Output::GsVpIndex : Output::None;
|
||||
}
|
||||
|
||||
if (ctl.vs_out_ccdist0_enable) {
|
||||
auto& ccdist0 = outputs[num_outputs++];
|
||||
ccdist0[0] = ctl.IsClipDistEnabled(0)
|
||||
? Output::ClipDist0
|
||||
: (ctl.IsCullDistEnabled(0) ? Output::CullDist0 : Output::None);
|
||||
ccdist0[1] = ctl.IsClipDistEnabled(1)
|
||||
? Output::ClipDist1
|
||||
: (ctl.IsCullDistEnabled(1) ? Output::CullDist1 : Output::None);
|
||||
ccdist0[2] = ctl.IsClipDistEnabled(2)
|
||||
? Output::ClipDist2
|
||||
: (ctl.IsCullDistEnabled(2) ? Output::CullDist2 : Output::None);
|
||||
ccdist0[3] = ctl.IsClipDistEnabled(3)
|
||||
? Output::ClipDist3
|
||||
: (ctl.IsCullDistEnabled(3) ? Output::CullDist3 : Output::None);
|
||||
}
|
||||
|
||||
if (ctl.vs_out_ccdist1_enable) {
|
||||
auto& ccdist1 = outputs[num_outputs++];
|
||||
ccdist1[0] = ctl.IsClipDistEnabled(4)
|
||||
? Output::ClipDist4
|
||||
: (ctl.IsCullDistEnabled(4) ? Output::CullDist4 : Output::None);
|
||||
ccdist1[1] = ctl.IsClipDistEnabled(5)
|
||||
? Output::ClipDist5
|
||||
: (ctl.IsCullDistEnabled(5) ? Output::CullDist5 : Output::None);
|
||||
ccdist1[2] = ctl.IsClipDistEnabled(6)
|
||||
? Output::ClipDist6
|
||||
: (ctl.IsCullDistEnabled(6) ? Output::CullDist6 : Output::None);
|
||||
ccdist1[3] = ctl.IsClipDistEnabled(7)
|
||||
? Output::ClipDist7
|
||||
: (ctl.IsCullDistEnabled(7) ? Output::CullDist7 : Output::None);
|
||||
}
|
||||
|
||||
return num_outputs;
|
||||
}
|
||||
|
||||
const Shader::RuntimeInfo& PipelineCache::BuildRuntimeInfo(Stage stage, LogicalStage l_stage) {
|
||||
@ -121,7 +127,7 @@ const Shader::RuntimeInfo& PipelineCache::BuildRuntimeInfo(Stage stage, LogicalS
|
||||
}
|
||||
case Stage::Vertex: {
|
||||
BuildCommon(regs.vs_program);
|
||||
GatherVertexOutputs(info.vs_info, regs.vs_output_control);
|
||||
info.vs_info.num_outputs = MapOutputs(info.vs_info.outputs, regs.vs_output_control);
|
||||
info.vs_info.emulate_depth_negative_one_to_one =
|
||||
!instance.IsDepthClipControlSupported() &&
|
||||
regs.clipper_control.clip_space == Liverpool::ClipSpace::MinusWToW;
|
||||
@ -136,6 +142,7 @@ const Shader::RuntimeInfo& PipelineCache::BuildRuntimeInfo(Stage stage, LogicalS
|
||||
case Stage::Geometry: {
|
||||
BuildCommon(regs.gs_program);
|
||||
auto& gs_info = info.gs_info;
|
||||
gs_info.num_outputs = MapOutputs(gs_info.outputs, regs.vs_output_control);
|
||||
gs_info.output_vertices = regs.vgt_gs_max_vert_out;
|
||||
gs_info.num_invocations =
|
||||
regs.vgt_gs_instance_cnt.IsEnabled() ? regs.vgt_gs_instance_cnt.count : 1;
|
||||
|
@ -162,6 +162,7 @@ RenderState Rasterizer::PrepareRenderState(u32 mrt_mask) {
|
||||
const auto mip = image_view.info.range.base.level;
|
||||
state.width = std::min<u32>(state.width, std::max(image.info.size.width >> mip, 1u));
|
||||
state.height = std::min<u32>(state.height, std::max(image.info.size.height >> mip, 1u));
|
||||
state.num_layers = std::min<u32>(state.num_layers, image_view.info.range.extent.layers);
|
||||
state.color_attachments[state.num_color_attachments++] = {
|
||||
.imageView = *image_view.image_view,
|
||||
.imageLayout = vk::ImageLayout::eUndefined,
|
||||
@ -195,6 +196,7 @@ RenderState Rasterizer::PrepareRenderState(u32 mrt_mask) {
|
||||
state.height = std::min<u32>(state.height, image.info.size.height);
|
||||
state.has_depth = regs.depth_buffer.DepthValid();
|
||||
state.has_stencil = regs.depth_buffer.StencilValid();
|
||||
state.num_layers = std::min<u32>(state.num_layers, image_view.info.range.extent.layers);
|
||||
if (state.has_depth) {
|
||||
state.depth_attachment = {
|
||||
.imageView = *image_view.image_view,
|
||||
@ -218,6 +220,10 @@ RenderState Rasterizer::PrepareRenderState(u32 mrt_mask) {
|
||||
texture_cache.TouchMeta(htile_address, slice, false);
|
||||
}
|
||||
|
||||
if (state.num_layers == std::numeric_limits<u32>::max()) {
|
||||
state.num_layers = 1;
|
||||
}
|
||||
|
||||
return state;
|
||||
}
|
||||
|
||||
|
@ -4,6 +4,7 @@
|
||||
#include <mutex>
|
||||
#include "common/assert.h"
|
||||
#include "common/debug.h"
|
||||
#include "common/logging/log.h"
|
||||
#include "imgui/renderer/texture_manager.h"
|
||||
#include "video_core/renderer_vulkan/vk_instance.h"
|
||||
#include "video_core/renderer_vulkan/vk_scheduler.h"
|
||||
@ -34,9 +35,9 @@ void Scheduler::BeginRendering(const RenderState& new_state) {
|
||||
is_rendering = true;
|
||||
render_state = new_state;
|
||||
|
||||
const auto width =
|
||||
const u32 width =
|
||||
render_state.width != std::numeric_limits<u32>::max() ? render_state.width : 1;
|
||||
const auto height =
|
||||
const u32 height =
|
||||
render_state.height != std::numeric_limits<u32>::max() ? render_state.height : 1;
|
||||
|
||||
const vk::RenderingInfo rendering_info = {
|
||||
@ -45,7 +46,7 @@ void Scheduler::BeginRendering(const RenderState& new_state) {
|
||||
.offset = {0, 0},
|
||||
.extent = {width, height},
|
||||
},
|
||||
.layerCount = 1,
|
||||
.layerCount = render_state.num_layers,
|
||||
.colorAttachmentCount = render_state.num_color_attachments,
|
||||
.pColorAttachments = render_state.num_color_attachments > 0
|
||||
? render_state.color_attachments.data()
|
||||
|
@ -24,6 +24,7 @@ struct RenderState {
|
||||
vk::RenderingAttachmentInfo depth_attachment{};
|
||||
vk::RenderingAttachmentInfo stencil_attachment{};
|
||||
u32 num_color_attachments{};
|
||||
u32 num_layers = std::numeric_limits<u32>::max();
|
||||
bool has_depth{};
|
||||
bool has_stencil{};
|
||||
u32 width = std::numeric_limits<u32>::max();
|
||||
|
Loading…
Reference in New Issue
Block a user