early-access version 3089

This commit is contained in:
pineappleEA
2022-11-05 19:21:57 +01:00
parent b601909c6d
commit e03f443390
46 changed files with 491 additions and 63 deletions

View File

@@ -2970,7 +2970,7 @@ public:
CullFace gl_cull_face; ///< 0x1920
Viewport::PixelCenter viewport_pixel_center; ///< 0x1924
INSERT_PADDING_BYTES_NOINIT(0x4);
u32 viewport_scale_offset_enbled; ///< 0x192C
u32 viewport_scale_offset_enabled; ///< 0x192C
INSERT_PADDING_BYTES_NOINIT(0xC);
ViewportClipControl viewport_clip_control; ///< 0x193C
UserClip::Op user_clip_op; ///< 0x1940
@@ -3482,7 +3482,7 @@ ASSERT_REG_POSITION(gl_cull_test_enabled, 0x1918);
ASSERT_REG_POSITION(gl_front_face, 0x191C);
ASSERT_REG_POSITION(gl_cull_face, 0x1920);
ASSERT_REG_POSITION(viewport_pixel_center, 0x1924);
ASSERT_REG_POSITION(viewport_scale_offset_enbled, 0x192C);
ASSERT_REG_POSITION(viewport_scale_offset_enabled, 0x192C);
ASSERT_REG_POSITION(viewport_clip_control, 0x193C);
ASSERT_REG_POSITION(user_clip_op, 0x1940);
ASSERT_REG_POSITION(render_enable_override, 0x1944);

View File

@@ -29,17 +29,17 @@ constexpr std::array PROGRAM_LUT{
[[nodiscard]] GLenum GetTextureBufferFormat(GLenum gl_format) {
switch (gl_format) {
case GL_RGBA8_SNORM:
return GL_RGBA8;
return GL_RGBA8I;
case GL_R8_SNORM:
return GL_R8;
return GL_R8I;
case GL_RGBA16_SNORM:
return GL_RGBA16;
return GL_RGBA16I;
case GL_R16_SNORM:
return GL_R16;
return GL_R16I;
case GL_RG16_SNORM:
return GL_RG16;
return GL_RG16I;
case GL_RG8_SNORM:
return GL_RG8;
return GL_RG8I;
default:
return gl_format;
}
@@ -96,9 +96,6 @@ GLuint Buffer::View(u32 offset, u32 size, PixelFormat format) {
texture.Create(GL_TEXTURE_BUFFER);
const GLenum gl_format{MaxwellToGL::GetFormatTuple(format).internal_format};
const GLenum texture_format{GetTextureBufferFormat(gl_format)};
if (texture_format != gl_format) {
LOG_WARNING(Render_OpenGL, "Emulating SNORM texture buffer with UNORM.");
}
glTextureBufferRange(texture.handle, texture_format, buffer.handle, offset, size);
views.push_back({
.offset = offset,

View File

@@ -503,6 +503,17 @@ void GraphicsPipeline::ConfigureImpl(bool is_indexed) {
float_image_scaling_mask, down_factor, 0.0f);
}
}
if (info.uses_render_area) {
const auto render_area_width(static_cast<GLfloat>(regs.surface_clip.width));
const auto render_area_height(static_cast<GLfloat>(regs.surface_clip.height));
if (use_assembly) {
glProgramLocalParameter4fARB(AssemblyStage(stage), 1, render_area_width,
render_area_height, 0.0f, 0.0f);
} else {
glProgramUniform4f(source_programs[stage].handle, 1, render_area_width,
render_area_height, 0.0f, 0.0f);
}
}
}};
if constexpr (Spec::enabled_stages[0]) {
prepare_stage(0);

View File

@@ -618,6 +618,16 @@ void RasterizerOpenGL::SyncViewport() {
}
flags[Dirty::Viewport0 + index] = false;
if (!regs.viewport_scale_offset_enabled) {
const auto x = static_cast<GLfloat>(regs.surface_clip.x);
const auto y = static_cast<GLfloat>(regs.surface_clip.y);
const auto width = static_cast<GLfloat>(regs.surface_clip.width);
const auto height = static_cast<GLfloat>(regs.surface_clip.height);
glViewportIndexedf(static_cast<GLuint>(index), x, y, width != 0.0f ? width : 1.0f,
height != 0.0f ? height : 1.0f);
continue;
}
const auto& src = regs.viewport_transform[index];
GLfloat x = conv(src.translate_x - src.scale_x);
GLfloat y = conv(src.translate_y - src.scale_y);

View File

@@ -49,7 +49,7 @@ using VideoCommon::LoadPipelines;
using VideoCommon::SerializePipeline;
using Context = ShaderContext::Context;
constexpr u32 CACHE_VERSION = 6;
constexpr u32 CACHE_VERSION = 7;
template <typename Container>
auto MakeSpan(Container& container) {
@@ -76,7 +76,7 @@ Shader::RuntimeInfo MakeRuntimeInfo(const GraphicsPipelineKey& key,
}
break;
case Shader::Stage::TessellationEval:
// invert the face
// Flip the face, as opengl's drawing is also flipped
info.tess_clockwise = key.tessellation_clockwise == 0;
info.tess_primitive = [&key] {
switch (key.tessellation_primitive) {
@@ -219,6 +219,7 @@ ShaderCache::ShaderCache(RasterizerOpenGL& rasterizer_, Core::Frontend::EmuWindo
.support_float16 = false,
.support_int64 = device.HasShaderInt64(),
.needs_demote_reorder = device.IsAmd(),
.support_snorm_render_buffer = false,
} {
if (use_asynchronous_shaders) {
workers = CreateWorkers();

View File

@@ -70,8 +70,8 @@ void SetupDirtyViewports(Tables& tables) {
FillBlock(tables[1], OFF(viewport_transform), NUM(viewport_transform), Viewports);
FillBlock(tables[1], OFF(viewports), NUM(viewports), Viewports);
tables[0][OFF(viewport_scale_offset_enbled)] = ViewportTransform;
tables[1][OFF(viewport_scale_offset_enbled)] = Viewports;
tables[0][OFF(viewport_scale_offset_enabled)] = ViewportTransform;
tables[1][OFF(viewport_scale_offset_enabled)] = Viewports;
}
void SetupDirtyScissors(Tables& tables) {

View File

@@ -68,13 +68,15 @@ public:
}
vk::PipelineLayout CreatePipelineLayout(VkDescriptorSetLayout descriptor_set_layout) const {
using Shader::Backend::SPIRV::RenderAreaLayout;
using Shader::Backend::SPIRV::RescalingLayout;
const u32 size_offset = is_compute ? sizeof(RescalingLayout::down_factor) : 0u;
const VkPushConstantRange range{
.stageFlags = static_cast<VkShaderStageFlags>(
is_compute ? VK_SHADER_STAGE_COMPUTE_BIT : VK_SHADER_STAGE_ALL_GRAPHICS),
.offset = 0,
.size = static_cast<u32>(sizeof(RescalingLayout)) - size_offset,
.size = static_cast<u32>(sizeof(RescalingLayout)) - size_offset +
static_cast<u32>(sizeof(RenderAreaLayout)),
};
return device->GetLogical().CreatePipelineLayout({
.sType = VK_STRUCTURE_TYPE_PIPELINE_LAYOUT_CREATE_INFO,
@@ -167,6 +169,12 @@ private:
u32 image_bit{1u};
};
class RenderAreaPushConstant {
public:
bool uses_render_area{};
std::array<f32, 4> words{};
};
inline void PushImageDescriptors(TextureCache& texture_cache,
UpdateDescriptorQueue& update_descriptor_queue,
const Shader::Info& info, RescalingPushConstant& rescaling,

View File

@@ -31,6 +31,7 @@ namespace {
using boost::container::small_vector;
using boost::container::static_vector;
using Shader::ImageBufferDescriptor;
using Shader::Backend::SPIRV::RENDERAREA_LAYOUT_OFFSET;
using Shader::Backend::SPIRV::RESCALING_LAYOUT_DOWN_FACTOR_OFFSET;
using Shader::Backend::SPIRV::RESCALING_LAYOUT_WORDS_OFFSET;
using Tegra::Texture::TexturePair;
@@ -433,12 +434,19 @@ void GraphicsPipeline::ConfigureImpl(bool is_indexed) {
update_descriptor_queue.Acquire();
RescalingPushConstant rescaling;
RenderAreaPushConstant render_area;
const VkSampler* samplers_it{samplers.data()};
const VideoCommon::ImageViewInOut* views_it{views.data()};
const auto prepare_stage{[&](size_t stage) LAMBDA_FORCEINLINE {
buffer_cache.BindHostStageBuffers(stage);
PushImageDescriptors(texture_cache, update_descriptor_queue, stage_infos[stage], rescaling,
samplers_it, views_it);
const auto& info{stage_infos[0]};
if (info.uses_render_area) {
render_area.uses_render_area = true;
render_area.words = {static_cast<float>(regs.surface_clip.width),
static_cast<float>(regs.surface_clip.height)};
}
}};
if constexpr (Spec::enabled_stages[0]) {
prepare_stage(0);
@@ -455,10 +463,11 @@ void GraphicsPipeline::ConfigureImpl(bool is_indexed) {
if constexpr (Spec::enabled_stages[4]) {
prepare_stage(4);
}
ConfigureDraw(rescaling);
ConfigureDraw(rescaling, render_area);
}
void GraphicsPipeline::ConfigureDraw(const RescalingPushConstant& rescaling) {
void GraphicsPipeline::ConfigureDraw(const RescalingPushConstant& rescaling,
const RenderAreaPushConstant& render_area) {
texture_cache.UpdateRenderTargets(false);
scheduler.RequestRenderpass(texture_cache.GetFramebuffer());
@@ -474,7 +483,9 @@ void GraphicsPipeline::ConfigureDraw(const RescalingPushConstant& rescaling) {
const bool bind_pipeline{scheduler.UpdateGraphicsPipeline(this)};
const void* const descriptor_data{update_descriptor_queue.UpdateData()};
scheduler.Record([this, descriptor_data, bind_pipeline, rescaling_data = rescaling.Data(),
is_rescaling, update_rescaling](vk::CommandBuffer cmdbuf) {
is_rescaling, update_rescaling,
uses_render_area = render_area.uses_render_area,
render_area_data = render_area.words](vk::CommandBuffer cmdbuf) {
if (bind_pipeline) {
cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, *pipeline);
}
@@ -488,6 +499,11 @@ void GraphicsPipeline::ConfigureDraw(const RescalingPushConstant& rescaling) {
RESCALING_LAYOUT_DOWN_FACTOR_OFFSET, sizeof(scale_down_factor),
&scale_down_factor);
}
if (uses_render_area) {
cmdbuf.PushConstants(*pipeline_layout, VK_SHADER_STAGE_ALL_GRAPHICS,
RENDERAREA_LAYOUT_OFFSET, sizeof(render_area_data),
&render_area_data);
}
if (!descriptor_set_layout) {
return;
}

View File

@@ -62,6 +62,7 @@ class Device;
class PipelineStatistics;
class RenderPassCache;
class RescalingPushConstant;
class RenderAreaPushConstant;
class Scheduler;
class UpdateDescriptorQueue;
@@ -119,7 +120,8 @@ private:
template <typename Spec>
void ConfigureImpl(bool is_indexed);
void ConfigureDraw(const RescalingPushConstant& rescaling);
void ConfigureDraw(const RescalingPushConstant& rescaling,
const RenderAreaPushConstant& render_are);
void MakePipeline(VkRenderPass render_pass);

View File

@@ -53,7 +53,7 @@ using VideoCommon::FileEnvironment;
using VideoCommon::GenericEnvironment;
using VideoCommon::GraphicsEnvironment;
constexpr u32 CACHE_VERSION = 6;
constexpr u32 CACHE_VERSION = 7;
template <typename Container>
auto MakeSpan(Container& container) {
@@ -326,6 +326,7 @@ PipelineCache::PipelineCache(RasterizerVulkan& rasterizer_, const Device& device
.support_int64 = device.IsShaderInt64Supported(),
.needs_demote_reorder = driver_id == VK_DRIVER_ID_AMD_PROPRIETARY_KHR ||
driver_id == VK_DRIVER_ID_AMD_OPEN_SOURCE_KHR,
.support_snorm_render_buffer = true,
};
}

View File

@@ -683,6 +683,22 @@ void RasterizerVulkan::UpdateViewportsState(Tegra::Engines::Maxwell3D::Regs& reg
if (!state_tracker.TouchViewports()) {
return;
}
if (!regs.viewport_scale_offset_enabled) {
const auto x = static_cast<float>(regs.surface_clip.x);
const auto y = static_cast<float>(regs.surface_clip.y);
const auto width = static_cast<float>(regs.surface_clip.width);
const auto height = static_cast<float>(regs.surface_clip.height);
VkViewport viewport{
.x = x,
.y = y,
.width = width != 0.0f ? width : 1.0f,
.height = height != 0.0f ? height : 1.0f,
.minDepth = 0.0f,
.maxDepth = 1.0f,
};
scheduler.Record([viewport](vk::CommandBuffer cmdbuf) { cmdbuf.SetViewport(0, viewport); });
return;
}
const bool is_rescaling{texture_cache.IsRescaling()};
const float scale = is_rescaling ? Settings::values.resolution_info.up_factor : 1.0f;
const std::array viewports{

View File

@@ -51,7 +51,7 @@ Flags MakeInvalidationFlags() {
void SetupDirtyViewports(Tables& tables) {
FillBlock(tables[0], OFF(viewport_transform), NUM(viewport_transform), Viewports);
FillBlock(tables[0], OFF(viewports), NUM(viewports), Viewports);
tables[0][OFF(viewport_scale_offset_enbled)] = Viewports;
tables[0][OFF(viewport_scale_offset_enabled)] = Viewports;
tables[1][OFF(window_origin)] = Viewports;
}

View File

@@ -19,6 +19,7 @@
#include "video_core/engines/kepler_compute.h"
#include "video_core/memory_manager.h"
#include "video_core/shader_environment.h"
#include "video_core/texture_cache/format_lookup_table.h"
#include "video_core/textures/texture.h"
namespace VideoCommon {
@@ -33,7 +34,7 @@ static u64 MakeCbufKey(u32 index, u32 offset) {
return (static_cast<u64>(index) << 32) | offset;
}
static Shader::TextureType ConvertType(const Tegra::Texture::TICEntry& entry) {
static Shader::TextureType ConvertTextureType(const Tegra::Texture::TICEntry& entry) {
switch (entry.texture_type) {
case Tegra::Texture::TextureType::Texture1D:
return Shader::TextureType::Color1D;
@@ -59,6 +60,26 @@ static Shader::TextureType ConvertType(const Tegra::Texture::TICEntry& entry) {
}
}
static Shader::TexturePixelFormat ConvertTexturePixelFormat(const Tegra::Texture::TICEntry& entry) {
switch (PixelFormatFromTextureInfo(entry.format, entry.r_type, entry.g_type, entry.b_type,
entry.a_type, entry.srgb_conversion)) {
case VideoCore::Surface::PixelFormat::A8B8G8R8_SNORM:
return Shader::TexturePixelFormat::A8B8G8R8_SNORM;
case VideoCore::Surface::PixelFormat::R8_SNORM:
return Shader::TexturePixelFormat::R8_SNORM;
case VideoCore::Surface::PixelFormat::R8G8_SNORM:
return Shader::TexturePixelFormat::R8G8_SNORM;
case VideoCore::Surface::PixelFormat::R16G16B16A16_SNORM:
return Shader::TexturePixelFormat::R16G16B16A16_SNORM;
case VideoCore::Surface::PixelFormat::R16G16_SNORM:
return Shader::TexturePixelFormat::R16G16_SNORM;
case VideoCore::Surface::PixelFormat::R16_SNORM:
return Shader::TexturePixelFormat::R16_SNORM;
default:
return Shader::TexturePixelFormat::OTHER;
}
}
static std::string_view StageToPrefix(Shader::Stage stage) {
switch (stage) {
case Shader::Stage::VertexB:
@@ -178,22 +199,31 @@ void GenericEnvironment::Dump(u64 hash) {
void GenericEnvironment::Serialize(std::ofstream& file) const {
const u64 code_size{static_cast<u64>(CachedSize())};
const u64 num_texture_types{static_cast<u64>(texture_types.size())};
const u64 num_texture_pixel_formats{static_cast<u64>(texture_pixel_formats.size())};
const u64 num_cbuf_values{static_cast<u64>(cbuf_values.size())};
file.write(reinterpret_cast<const char*>(&code_size), sizeof(code_size))
.write(reinterpret_cast<const char*>(&num_texture_types), sizeof(num_texture_types))
.write(reinterpret_cast<const char*>(&num_texture_pixel_formats),
sizeof(num_texture_pixel_formats))
.write(reinterpret_cast<const char*>(&num_cbuf_values), sizeof(num_cbuf_values))
.write(reinterpret_cast<const char*>(&local_memory_size), sizeof(local_memory_size))
.write(reinterpret_cast<const char*>(&texture_bound), sizeof(texture_bound))
.write(reinterpret_cast<const char*>(&start_address), sizeof(start_address))
.write(reinterpret_cast<const char*>(&cached_lowest), sizeof(cached_lowest))
.write(reinterpret_cast<const char*>(&cached_highest), sizeof(cached_highest))
.write(reinterpret_cast<const char*>(&viewport_transform_state),
sizeof(viewport_transform_state))
.write(reinterpret_cast<const char*>(&stage), sizeof(stage))
.write(reinterpret_cast<const char*>(code.data()), code_size);
for (const auto& [key, type] : texture_types) {
file.write(reinterpret_cast<const char*>(&key), sizeof(key))
.write(reinterpret_cast<const char*>(&type), sizeof(type));
}
for (const auto& [key, format] : texture_pixel_formats) {
file.write(reinterpret_cast<const char*>(&key), sizeof(key))
.write(reinterpret_cast<const char*>(&format), sizeof(format));
}
for (const auto& [key, type] : cbuf_values) {
file.write(reinterpret_cast<const char*>(&key), sizeof(key))
.write(reinterpret_cast<const char*>(&type), sizeof(type));
@@ -237,15 +267,13 @@ std::optional<u64> GenericEnvironment::TryFindSize() {
return std::nullopt;
}
Shader::TextureType GenericEnvironment::ReadTextureTypeImpl(GPUVAddr tic_addr, u32 tic_limit,
bool via_header_index, u32 raw) {
Tegra::Texture::TICEntry GenericEnvironment::ReadTextureInfo(GPUVAddr tic_addr, u32 tic_limit,
bool via_header_index, u32 raw) {
const auto handle{Tegra::Texture::TexturePair(raw, via_header_index)};
const GPUVAddr descriptor_addr{tic_addr + handle.first * sizeof(Tegra::Texture::TICEntry)};
Tegra::Texture::TICEntry entry;
gpu_memory->ReadBlock(descriptor_addr, &entry, sizeof(entry));
const Shader::TextureType result{ConvertType(entry)};
texture_types.emplace(raw, result);
return result;
return entry;
}
GraphicsEnvironment::GraphicsEnvironment(Tegra::Engines::Maxwell3D& maxwell3d_,
@@ -305,8 +333,27 @@ u32 GraphicsEnvironment::ReadCbufValue(u32 cbuf_index, u32 cbuf_offset) {
Shader::TextureType GraphicsEnvironment::ReadTextureType(u32 handle) {
const auto& regs{maxwell3d->regs};
const bool via_header_index{regs.sampler_binding == Maxwell::SamplerBinding::ViaHeaderBinding};
return ReadTextureTypeImpl(regs.tex_header.Address(), regs.tex_header.limit, via_header_index,
handle);
auto entry =
ReadTextureInfo(regs.tex_header.Address(), regs.tex_header.limit, via_header_index, handle);
const Shader::TextureType result{ConvertTextureType(entry)};
texture_types.emplace(handle, result);
return result;
}
Shader::TexturePixelFormat GraphicsEnvironment::ReadTexturePixelFormat(u32 handle) {
const auto& regs{maxwell3d->regs};
const bool via_header_index{regs.sampler_binding == Maxwell::SamplerBinding::ViaHeaderBinding};
auto entry =
ReadTextureInfo(regs.tex_header.Address(), regs.tex_header.limit, via_header_index, handle);
const Shader::TexturePixelFormat result(ConvertTexturePixelFormat(entry));
texture_pixel_formats.emplace(handle, result);
return result;
}
u32 GraphicsEnvironment::ReadViewportTransformState() {
const auto& regs{maxwell3d->regs};
viewport_transform_state = regs.viewport_scale_offset_enabled;
return viewport_transform_state;
}
ComputeEnvironment::ComputeEnvironment(Tegra::Engines::KeplerCompute& kepler_compute_,
@@ -337,21 +384,41 @@ u32 ComputeEnvironment::ReadCbufValue(u32 cbuf_index, u32 cbuf_offset) {
Shader::TextureType ComputeEnvironment::ReadTextureType(u32 handle) {
const auto& regs{kepler_compute->regs};
const auto& qmd{kepler_compute->launch_description};
return ReadTextureTypeImpl(regs.tic.Address(), regs.tic.limit, qmd.linked_tsc != 0, handle);
auto entry = ReadTextureInfo(regs.tic.Address(), regs.tic.limit, qmd.linked_tsc != 0, handle);
const Shader::TextureType result{ConvertTextureType(entry)};
texture_types.emplace(handle, result);
return result;
}
Shader::TexturePixelFormat ComputeEnvironment::ReadTexturePixelFormat(u32 handle) {
const auto& regs{kepler_compute->regs};
const auto& qmd{kepler_compute->launch_description};
auto entry = ReadTextureInfo(regs.tic.Address(), regs.tic.limit, qmd.linked_tsc != 0, handle);
const Shader::TexturePixelFormat result(ConvertTexturePixelFormat(entry));
texture_pixel_formats.emplace(handle, result);
return result;
}
u32 ComputeEnvironment::ReadViewportTransformState() {
return viewport_transform_state;
}
void FileEnvironment::Deserialize(std::ifstream& file) {
u64 code_size{};
u64 num_texture_types{};
u64 num_texture_pixel_formats{};
u64 num_cbuf_values{};
file.read(reinterpret_cast<char*>(&code_size), sizeof(code_size))
.read(reinterpret_cast<char*>(&num_texture_types), sizeof(num_texture_types))
.read(reinterpret_cast<char*>(&num_texture_pixel_formats),
sizeof(num_texture_pixel_formats))
.read(reinterpret_cast<char*>(&num_cbuf_values), sizeof(num_cbuf_values))
.read(reinterpret_cast<char*>(&local_memory_size), sizeof(local_memory_size))
.read(reinterpret_cast<char*>(&texture_bound), sizeof(texture_bound))
.read(reinterpret_cast<char*>(&start_address), sizeof(start_address))
.read(reinterpret_cast<char*>(&read_lowest), sizeof(read_lowest))
.read(reinterpret_cast<char*>(&read_highest), sizeof(read_highest))
.read(reinterpret_cast<char*>(&viewport_transform_state), sizeof(viewport_transform_state))
.read(reinterpret_cast<char*>(&stage), sizeof(stage));
code = std::make_unique<u64[]>(Common::DivCeil(code_size, sizeof(u64)));
file.read(reinterpret_cast<char*>(code.get()), code_size);
@@ -362,6 +429,13 @@ void FileEnvironment::Deserialize(std::ifstream& file) {
.read(reinterpret_cast<char*>(&type), sizeof(type));
texture_types.emplace(key, type);
}
for (size_t i = 0; i < num_texture_pixel_formats; ++i) {
u32 key;
Shader::TexturePixelFormat format;
file.read(reinterpret_cast<char*>(&key), sizeof(key))
.read(reinterpret_cast<char*>(&format), sizeof(format));
texture_pixel_formats.emplace(key, format);
}
for (size_t i = 0; i < num_cbuf_values; ++i) {
u64 key;
u32 value;
@@ -409,6 +483,18 @@ Shader::TextureType FileEnvironment::ReadTextureType(u32 handle) {
return it->second;
}
Shader::TexturePixelFormat FileEnvironment::ReadTexturePixelFormat(u32 handle) {
const auto it{texture_pixel_formats.find(handle)};
if (it == texture_pixel_formats.end()) {
throw Shader::LogicError("Uncached read texture pixel format");
}
return it->second;
}
u32 FileEnvironment::ReadViewportTransformState() {
return viewport_transform_state;
}
u32 FileEnvironment::LocalMemorySize() const {
return local_memory_size;
}

View File

@@ -63,14 +63,15 @@ public:
protected:
std::optional<u64> TryFindSize();
Shader::TextureType ReadTextureTypeImpl(GPUVAddr tic_addr, u32 tic_limit, bool via_header_index,
u32 raw);
Tegra::Texture::TICEntry ReadTextureInfo(GPUVAddr tic_addr, u32 tic_limit,
bool via_header_index, u32 raw);
Tegra::MemoryManager* gpu_memory{};
GPUVAddr program_base{};
std::vector<u64> code;
std::unordered_map<u32, Shader::TextureType> texture_types;
std::unordered_map<u32, Shader::TexturePixelFormat> texture_pixel_formats;
std::unordered_map<u64, u32> cbuf_values;
u32 local_memory_size{};
@@ -85,6 +86,8 @@ protected:
u32 cached_highest = 0;
u32 initial_offset = 0;
u32 viewport_transform_state = 1;
bool has_unbound_instructions = false;
};
@@ -102,6 +105,10 @@ public:
Shader::TextureType ReadTextureType(u32 handle) override;
Shader::TexturePixelFormat ReadTexturePixelFormat(u32 handle) override;
u32 ReadViewportTransformState() override;
private:
Tegra::Engines::Maxwell3D* maxwell3d{};
size_t stage_index{};
@@ -120,6 +127,10 @@ public:
Shader::TextureType ReadTextureType(u32 handle) override;
Shader::TexturePixelFormat ReadTexturePixelFormat(u32 handle) override;
u32 ReadViewportTransformState() override;
private:
Tegra::Engines::KeplerCompute* kepler_compute{};
};
@@ -143,6 +154,10 @@ public:
[[nodiscard]] Shader::TextureType ReadTextureType(u32 handle) override;
[[nodiscard]] Shader::TexturePixelFormat ReadTexturePixelFormat(u32 handle) override;
[[nodiscard]] u32 ReadViewportTransformState() override;
[[nodiscard]] u32 LocalMemorySize() const override;
[[nodiscard]] u32 SharedMemorySize() const override;
@@ -156,6 +171,7 @@ public:
private:
std::unique_ptr<u64[]> code;
std::unordered_map<u32, Shader::TextureType> texture_types;
std::unordered_map<u32, Shader::TexturePixelFormat> texture_pixel_formats;
std::unordered_map<u64, u32> cbuf_values;
std::array<u32, 3> workgroup_size{};
u32 local_memory_size{};
@@ -164,6 +180,7 @@ private:
u32 read_lowest{};
u32 read_highest{};
u32 initial_offset{};
u32 viewport_transform_state = 1;
};
void SerializePipeline(std::span<const char> key, std::span<const GenericEnvironment* const> envs,

View File

@@ -516,7 +516,6 @@ void SwizzleBlockLinearImage(Tegra::MemoryManager& gpu_memory, GPUVAddr gpu_addr
const u32 num_blocks_per_layer = NumBlocks(level_size, tile_size);
const u32 host_bytes_per_layer = num_blocks_per_layer * bytes_per_block;
UNIMPLEMENTED_IF(info.tile_width_spacing > 0);
UNIMPLEMENTED_IF(copy.image_offset.x != 0);
UNIMPLEMENTED_IF(copy.image_offset.y != 0);
UNIMPLEMENTED_IF(copy.image_offset.z != 0);