summaryrefslogtreecommitdiffstats
path: root/src/video_core
diff options
context:
space:
mode:
Diffstat (limited to 'src/video_core')
-rw-r--r--src/video_core/renderer_opengl/gl_texture_cache.cpp31
-rw-r--r--src/video_core/renderer_opengl/gl_texture_cache.h4
-rw-r--r--src/video_core/renderer_opengl/util_shaders.cpp13
-rw-r--r--src/video_core/renderer_vulkan/fixed_pipeline_state.cpp83
-rw-r--r--src/video_core/renderer_vulkan/fixed_pipeline_state.h18
-rw-r--r--src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp3
-rw-r--r--src/video_core/renderer_vulkan/vk_rasterizer.cpp12
-rw-r--r--src/video_core/renderer_vulkan/vk_rasterizer.h3
-rw-r--r--src/video_core/renderer_vulkan/vk_resource_pool.cpp14
-rw-r--r--src/video_core/renderer_vulkan/vk_resource_pool.h2
-rw-r--r--src/video_core/renderer_vulkan/vk_state_tracker.cpp34
-rw-r--r--src/video_core/renderer_vulkan/vk_state_tracker.h5
-rw-r--r--src/video_core/shader/async_shaders.cpp1
13 files changed, 152 insertions, 71 deletions
diff --git a/src/video_core/renderer_opengl/gl_texture_cache.cpp b/src/video_core/renderer_opengl/gl_texture_cache.cpp
index 31eb54123..12434db67 100644
--- a/src/video_core/renderer_opengl/gl_texture_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_texture_cache.cpp
@@ -763,6 +763,37 @@ void Image::DownloadMemory(ImageBufferMap& map,
}
}
+GLuint Image::StorageHandle() noexcept {
+ switch (info.format) {
+ case PixelFormat::A8B8G8R8_SRGB:
+ case PixelFormat::B8G8R8A8_SRGB:
+ case PixelFormat::BC1_RGBA_SRGB:
+ case PixelFormat::BC2_SRGB:
+ case PixelFormat::BC3_SRGB:
+ case PixelFormat::BC7_SRGB:
+ case PixelFormat::ASTC_2D_4X4_SRGB:
+ case PixelFormat::ASTC_2D_8X8_SRGB:
+ case PixelFormat::ASTC_2D_8X5_SRGB:
+ case PixelFormat::ASTC_2D_5X4_SRGB:
+ case PixelFormat::ASTC_2D_5X5_SRGB:
+ case PixelFormat::ASTC_2D_10X8_SRGB:
+ case PixelFormat::ASTC_2D_6X6_SRGB:
+ case PixelFormat::ASTC_2D_10X10_SRGB:
+ case PixelFormat::ASTC_2D_12X12_SRGB:
+ case PixelFormat::ASTC_2D_8X6_SRGB:
+ case PixelFormat::ASTC_2D_6X5_SRGB:
+ if (store_view.handle != 0) {
+ return store_view.handle;
+ }
+ store_view.Create();
+ glTextureView(store_view.handle, ImageTarget(info), texture.handle, GL_RGBA8, 0,
+ info.resources.levels, 0, info.resources.layers);
+ return store_view.handle;
+ default:
+ return texture.handle;
+ }
+}
+
void Image::CopyBufferToImage(const VideoCommon::BufferImageCopy& copy, size_t buffer_offset) {
// Compressed formats don't have a pixel format or type
const bool is_compressed = gl_format == GL_NONE;
diff --git a/src/video_core/renderer_opengl/gl_texture_cache.h b/src/video_core/renderer_opengl/gl_texture_cache.h
index 874cf54f4..a6172f009 100644
--- a/src/video_core/renderer_opengl/gl_texture_cache.h
+++ b/src/video_core/renderer_opengl/gl_texture_cache.h
@@ -145,6 +145,8 @@ public:
void DownloadMemory(ImageBufferMap& map, std::span<const VideoCommon::BufferImageCopy> copies);
+ GLuint StorageHandle() noexcept;
+
GLuint Handle() const noexcept {
return texture.handle;
}
@@ -155,8 +157,8 @@ private:
void CopyImageToBuffer(const VideoCommon::BufferImageCopy& copy, size_t buffer_offset);
OGLTexture texture;
- OGLTextureView store_view;
OGLBuffer buffer;
+ OGLTextureView store_view;
GLenum gl_internal_format = GL_NONE;
GLenum gl_format = GL_NONE;
GLenum gl_type = GL_NONE;
diff --git a/src/video_core/renderer_opengl/util_shaders.cpp b/src/video_core/renderer_opengl/util_shaders.cpp
index 1b58e8617..31ec68505 100644
--- a/src/video_core/renderer_opengl/util_shaders.cpp
+++ b/src/video_core/renderer_opengl/util_shaders.cpp
@@ -93,7 +93,7 @@ void UtilShaders::BlockLinearUpload2D(Image& image, const ImageBufferMap& map,
glUniform1ui(7, params.block_height_mask);
glBindBufferRange(GL_SHADER_STORAGE_BUFFER, BINDING_INPUT_BUFFER, map.buffer, input_offset,
image.guest_size_bytes - swizzle.buffer_offset);
- glBindImageTexture(BINDING_OUTPUT_IMAGE, image.Handle(), swizzle.level, GL_TRUE, 0,
+ glBindImageTexture(BINDING_OUTPUT_IMAGE, image.StorageHandle(), swizzle.level, GL_TRUE, 0,
GL_WRITE_ONLY, store_format);
glDispatchCompute(num_dispatches_x, num_dispatches_y, image.info.resources.layers);
}
@@ -134,7 +134,7 @@ void UtilShaders::BlockLinearUpload3D(Image& image, const ImageBufferMap& map,
glUniform1ui(9, params.block_depth_mask);
glBindBufferRange(GL_SHADER_STORAGE_BUFFER, BINDING_INPUT_BUFFER, map.buffer, input_offset,
image.guest_size_bytes - swizzle.buffer_offset);
- glBindImageTexture(BINDING_OUTPUT_IMAGE, image.Handle(), swizzle.level, GL_TRUE, 0,
+ glBindImageTexture(BINDING_OUTPUT_IMAGE, image.StorageHandle(), swizzle.level, GL_TRUE, 0,
GL_WRITE_ONLY, store_format);
glDispatchCompute(num_dispatches_x, num_dispatches_y, num_dispatches_z);
}
@@ -164,7 +164,8 @@ void UtilShaders::PitchUpload(Image& image, const ImageBufferMap& map,
glUniform2i(LOC_DESTINATION, 0, 0);
glUniform1ui(LOC_BYTES_PER_BLOCK, bytes_per_block);
glUniform1ui(LOC_PITCH, pitch);
- glBindImageTexture(BINDING_OUTPUT_IMAGE, image.Handle(), 0, GL_FALSE, 0, GL_WRITE_ONLY, format);
+ glBindImageTexture(BINDING_OUTPUT_IMAGE, image.StorageHandle(), 0, GL_FALSE, 0, GL_WRITE_ONLY,
+ format);
for (const SwizzleParameters& swizzle : swizzles) {
const Extent3D num_tiles = swizzle.num_tiles;
const size_t input_offset = swizzle.buffer_offset + map.offset;
@@ -195,9 +196,9 @@ void UtilShaders::CopyBC4(Image& dst_image, Image& src_image, std::span<const Im
glUniform3ui(LOC_SRC_OFFSET, copy.src_offset.x, copy.src_offset.y, copy.src_offset.z);
glUniform3ui(LOC_DST_OFFSET, copy.dst_offset.x, copy.dst_offset.y, copy.dst_offset.z);
- glBindImageTexture(BINDING_INPUT_IMAGE, src_image.Handle(), copy.src_subresource.base_level,
- GL_FALSE, 0, GL_READ_ONLY, GL_RG32UI);
- glBindImageTexture(BINDING_OUTPUT_IMAGE, dst_image.Handle(),
+ glBindImageTexture(BINDING_INPUT_IMAGE, src_image.StorageHandle(),
+ copy.src_subresource.base_level, GL_FALSE, 0, GL_READ_ONLY, GL_RG32UI);
+ glBindImageTexture(BINDING_OUTPUT_IMAGE, dst_image.StorageHandle(),
copy.dst_subresource.base_level, GL_FALSE, 0, GL_WRITE_ONLY, GL_RGBA8UI);
glDispatchCompute(copy.extent.width, copy.extent.height, copy.extent.depth);
}
diff --git a/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp b/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp
index 5be6dabd9..362278f01 100644
--- a/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp
+++ b/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp
@@ -12,14 +12,15 @@
#include "common/cityhash.h"
#include "common/common_types.h"
#include "video_core/renderer_vulkan/fixed_pipeline_state.h"
+#include "video_core/renderer_vulkan/vk_state_tracker.h"
namespace Vulkan {
namespace {
-constexpr std::size_t POINT = 0;
-constexpr std::size_t LINE = 1;
-constexpr std::size_t POLYGON = 2;
+constexpr size_t POINT = 0;
+constexpr size_t LINE = 1;
+constexpr size_t POLYGON = 2;
constexpr std::array POLYGON_OFFSET_ENABLE_LUT = {
POINT, // Points
LINE, // Lines
@@ -40,10 +41,14 @@ constexpr std::array POLYGON_OFFSET_ENABLE_LUT = {
} // Anonymous namespace
-void FixedPipelineState::Fill(const Maxwell& regs, bool has_extended_dynamic_state) {
- const std::array enabled_lut = {regs.polygon_offset_point_enable,
- regs.polygon_offset_line_enable,
- regs.polygon_offset_fill_enable};
+void FixedPipelineState::Refresh(Tegra::Engines::Maxwell3D& maxwell3d,
+ bool has_extended_dynamic_state) {
+ const Maxwell& regs = maxwell3d.regs;
+ const std::array enabled_lut{
+ regs.polygon_offset_point_enable,
+ regs.polygon_offset_line_enable,
+ regs.polygon_offset_fill_enable,
+ };
const u32 topology_index = static_cast<u32>(regs.draw.topology.Value());
raw1 = 0;
@@ -64,45 +69,53 @@ void FixedPipelineState::Fill(const Maxwell& regs, bool has_extended_dynamic_sta
raw2 = 0;
const auto test_func =
- regs.alpha_test_enabled == 1 ? regs.alpha_test_func : Maxwell::ComparisonOp::Always;
+ regs.alpha_test_enabled != 0 ? regs.alpha_test_func : Maxwell::ComparisonOp::Always;
alpha_test_func.Assign(PackComparisonOp(test_func));
early_z.Assign(regs.force_early_fragment_tests != 0 ? 1 : 0);
alpha_test_ref = Common::BitCast<u32>(regs.alpha_test_ref);
point_size = Common::BitCast<u32>(regs.point_size);
- for (std::size_t index = 0; index < Maxwell::NumVertexArrays; ++index) {
- binding_divisors[index] =
- regs.instanced_arrays.IsInstancingEnabled(index) ? regs.vertex_array[index].divisor : 0;
+ if (maxwell3d.dirty.flags[Dirty::InstanceDivisors]) {
+ maxwell3d.dirty.flags[Dirty::InstanceDivisors] = false;
+ for (size_t index = 0; index < Maxwell::NumVertexArrays; ++index) {
+ const bool is_enabled = regs.instanced_arrays.IsInstancingEnabled(index);
+ binding_divisors[index] = is_enabled ? regs.vertex_array[index].divisor : 0;
+ }
}
-
- for (size_t index = 0; index < Maxwell::NumVertexAttributes; ++index) {
- const auto& input = regs.vertex_attrib_format[index];
- auto& attribute = attributes[index];
- attribute.raw = 0;
- attribute.enabled.Assign(input.IsConstant() ? 0 : 1);
- attribute.buffer.Assign(input.buffer);
- attribute.offset.Assign(input.offset);
- attribute.type.Assign(static_cast<u32>(input.type.Value()));
- attribute.size.Assign(static_cast<u32>(input.size.Value()));
- attribute.binding_index_enabled.Assign(regs.vertex_array[index].IsEnabled() ? 1 : 0);
+ if (maxwell3d.dirty.flags[Dirty::VertexAttributes]) {
+ maxwell3d.dirty.flags[Dirty::VertexAttributes] = false;
+ for (size_t index = 0; index < Maxwell::NumVertexAttributes; ++index) {
+ const auto& input = regs.vertex_attrib_format[index];
+ auto& attribute = attributes[index];
+ attribute.raw = 0;
+ attribute.enabled.Assign(input.IsConstant() ? 0 : 1);
+ attribute.buffer.Assign(input.buffer);
+ attribute.offset.Assign(input.offset);
+ attribute.type.Assign(static_cast<u32>(input.type.Value()));
+ attribute.size.Assign(static_cast<u32>(input.size.Value()));
+ }
}
-
- for (std::size_t index = 0; index < std::size(attachments); ++index) {
- attachments[index].Fill(regs, index);
+ if (maxwell3d.dirty.flags[Dirty::Blending]) {
+ maxwell3d.dirty.flags[Dirty::Blending] = false;
+ for (size_t index = 0; index < attachments.size(); ++index) {
+ attachments[index].Refresh(regs, index);
+ }
+ }
+ if (maxwell3d.dirty.flags[Dirty::ViewportSwizzles]) {
+ maxwell3d.dirty.flags[Dirty::ViewportSwizzles] = false;
+ const auto& transform = regs.viewport_transform;
+ std::ranges::transform(transform, viewport_swizzles.begin(), [](const auto& viewport) {
+ return static_cast<u16>(viewport.swizzle.raw);
+ });
}
-
- const auto& transform = regs.viewport_transform;
- std::transform(transform.begin(), transform.end(), viewport_swizzles.begin(),
- [](const auto& viewport) { return static_cast<u16>(viewport.swizzle.raw); });
-
if (!has_extended_dynamic_state) {
no_extended_dynamic_state.Assign(1);
- dynamic_state.Fill(regs);
+ dynamic_state.Refresh(regs);
}
}
-void FixedPipelineState::BlendingAttachment::Fill(const Maxwell& regs, std::size_t index) {
+void FixedPipelineState::BlendingAttachment::Refresh(const Maxwell& regs, size_t index) {
const auto& mask = regs.color_mask[regs.color_mask_common ? 0 : index];
raw = 0;
@@ -141,7 +154,7 @@ void FixedPipelineState::BlendingAttachment::Fill(const Maxwell& regs, std::size
enable.Assign(1);
}
-void FixedPipelineState::DynamicState::Fill(const Maxwell& regs) {
+void FixedPipelineState::DynamicState::Refresh(const Maxwell& regs) {
u32 packed_front_face = PackFrontFace(regs.front_face);
if (regs.screen_y_control.triangle_rast_flip != 0) {
// Flip front face
@@ -178,9 +191,9 @@ void FixedPipelineState::DynamicState::Fill(const Maxwell& regs) {
});
}
-std::size_t FixedPipelineState::Hash() const noexcept {
+size_t FixedPipelineState::Hash() const noexcept {
const u64 hash = Common::CityHash64(reinterpret_cast<const char*>(this), Size());
- return static_cast<std::size_t>(hash);
+ return static_cast<size_t>(hash);
}
bool FixedPipelineState::operator==(const FixedPipelineState& rhs) const noexcept {
diff --git a/src/video_core/renderer_vulkan/fixed_pipeline_state.h b/src/video_core/renderer_vulkan/fixed_pipeline_state.h
index 465a55fdb..a0eb83a68 100644
--- a/src/video_core/renderer_vulkan/fixed_pipeline_state.h
+++ b/src/video_core/renderer_vulkan/fixed_pipeline_state.h
@@ -58,7 +58,7 @@ struct FixedPipelineState {
BitField<30, 1, u32> enable;
};
- void Fill(const Maxwell& regs, std::size_t index);
+ void Refresh(const Maxwell& regs, size_t index);
constexpr std::array<bool, 4> Mask() const noexcept {
return {mask_r != 0, mask_g != 0, mask_b != 0, mask_a != 0};
@@ -96,8 +96,6 @@ struct FixedPipelineState {
BitField<6, 14, u32> offset;
BitField<20, 3, u32> type;
BitField<23, 6, u32> size;
- // Not really an element of a vertex attribute, but it can be packed here
- BitField<29, 1, u32> binding_index_enabled;
constexpr Maxwell::VertexAttribute::Type Type() const noexcept {
return static_cast<Maxwell::VertexAttribute::Type>(type.Value());
@@ -108,7 +106,7 @@ struct FixedPipelineState {
}
};
- template <std::size_t Position>
+ template <size_t Position>
union StencilFace {
BitField<Position + 0, 3, u32> action_stencil_fail;
BitField<Position + 3, 3, u32> action_depth_fail;
@@ -152,7 +150,7 @@ struct FixedPipelineState {
// Vertex stride is a 12 bits value, we have 4 bits to spare per element
std::array<u16, Maxwell::NumVertexArrays> vertex_strides;
- void Fill(const Maxwell& regs);
+ void Refresh(const Maxwell& regs);
Maxwell::ComparisonOp DepthTestFunc() const noexcept {
return UnpackComparisonOp(depth_test_func);
@@ -199,9 +197,9 @@ struct FixedPipelineState {
std::array<u16, Maxwell::NumViewports> viewport_swizzles;
DynamicState dynamic_state;
- void Fill(const Maxwell& regs, bool has_extended_dynamic_state);
+ void Refresh(Tegra::Engines::Maxwell3D& maxwell3d, bool has_extended_dynamic_state);
- std::size_t Hash() const noexcept;
+ size_t Hash() const noexcept;
bool operator==(const FixedPipelineState& rhs) const noexcept;
@@ -209,8 +207,8 @@ struct FixedPipelineState {
return !operator==(rhs);
}
- std::size_t Size() const noexcept {
- const std::size_t total_size = sizeof *this;
+ size_t Size() const noexcept {
+ const size_t total_size = sizeof *this;
return total_size - (no_extended_dynamic_state != 0 ? 0 : sizeof(DynamicState));
}
};
@@ -224,7 +222,7 @@ namespace std {
template <>
struct hash<Vulkan::FixedPipelineState> {
- std::size_t operator()(const Vulkan::FixedPipelineState& k) const noexcept {
+ size_t operator()(const Vulkan::FixedPipelineState& k) const noexcept {
return k.Hash();
}
};
diff --git a/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp b/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp
index d50dca604..fc6dd83eb 100644
--- a/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp
+++ b/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp
@@ -221,9 +221,6 @@ vk::Pipeline VKGraphicsPipeline::CreatePipeline(const SPIRVProgram& program,
std::vector<VkVertexInputBindingDescription> vertex_bindings;
std::vector<VkVertexInputBindingDivisorDescriptionEXT> vertex_binding_divisors;
for (std::size_t index = 0; index < Maxwell::NumVertexArrays; ++index) {
- if (state.attributes[index].binding_index_enabled == 0) {
- continue;
- }
const bool instanced = state.binding_divisors[index] != 0;
const auto rate = instanced ? VK_VERTEX_INPUT_RATE_INSTANCE : VK_VERTEX_INPUT_RATE_VERTEX;
vertex_bindings.push_back({
diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.cpp b/src/video_core/renderer_vulkan/vk_rasterizer.cpp
index 684d4e3a6..dfd38f575 100644
--- a/src/video_core/renderer_vulkan/vk_rasterizer.cpp
+++ b/src/video_core/renderer_vulkan/vk_rasterizer.cpp
@@ -267,8 +267,7 @@ void RasterizerVulkan::Draw(bool is_indexed, bool is_instanced) {
query_cache.UpdateCounters();
- GraphicsPipelineCacheKey key;
- key.fixed_state.Fill(maxwell3d.regs, device.IsExtExtendedDynamicStateSupported());
+ graphics_key.fixed_state.Refresh(maxwell3d, device.IsExtExtendedDynamicStateSupported());
std::scoped_lock lock{buffer_cache.mutex, texture_cache.mutex};
@@ -276,14 +275,15 @@ void RasterizerVulkan::Draw(bool is_indexed, bool is_instanced) {
texture_cache.UpdateRenderTargets(false);
const auto shaders = pipeline_cache.GetShaders();
- key.shaders = GetShaderAddresses(shaders);
+ graphics_key.shaders = GetShaderAddresses(shaders);
+
SetupShaderDescriptors(shaders, is_indexed);
const Framebuffer* const framebuffer = texture_cache.GetFramebuffer();
- key.renderpass = framebuffer->RenderPass();
+ graphics_key.renderpass = framebuffer->RenderPass();
- auto* const pipeline =
- pipeline_cache.GetGraphicsPipeline(key, framebuffer->NumColorBuffers(), async_shaders);
+ VKGraphicsPipeline* const pipeline = pipeline_cache.GetGraphicsPipeline(
+ graphics_key, framebuffer->NumColorBuffers(), async_shaders);
if (pipeline == nullptr || pipeline->GetHandle() == VK_NULL_HANDLE) {
// Async graphics pipeline was not ready.
return;
diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.h b/src/video_core/renderer_vulkan/vk_rasterizer.h
index 7fc6741da..acea1ba2d 100644
--- a/src/video_core/renderer_vulkan/vk_rasterizer.h
+++ b/src/video_core/renderer_vulkan/vk_rasterizer.h
@@ -20,6 +20,7 @@
#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"
@@ -173,6 +174,8 @@ private:
VKUpdateDescriptorQueue update_descriptor_queue;
BlitImageHelper blit_image;
+ 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 ee274ac59..a8bf7bda8 100644
--- a/src/video_core/renderer_vulkan/vk_resource_pool.cpp
+++ b/src/video_core/renderer_vulkan/vk_resource_pool.cpp
@@ -17,21 +17,21 @@ ResourcePool::~ResourcePool() = default;
size_t ResourcePool::CommitResource() {
// Refresh semaphore to query updated results
master_semaphore.Refresh();
-
- const auto search = [this](size_t begin, size_t end) -> std::optional<size_t> {
+ 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 (master_semaphore.IsFree(ticks[iterator])) {
+ if (gpu_tick >= ticks[iterator]) {
ticks[iterator] = master_semaphore.CurrentTick();
return iterator;
}
}
- return {};
+ return std::nullopt;
};
// Try to find a free resource from the hinted position to the end.
- auto found = search(free_iterator, ticks.size());
+ std::optional<size_t> found = search(hint_iterator, ticks.size());
if (!found) {
// Search from beginning to the hinted position.
- found = search(0, free_iterator);
+ found = search(0, hint_iterator);
if (!found) {
// Both searches failed, the pool is full; handle it.
const size_t free_resource = ManageOverflow();
@@ -41,7 +41,7 @@ size_t ResourcePool::CommitResource() {
}
}
// Free iterator is hinted to the resource after the one that's been commited.
- free_iterator = (*found + 1) % ticks.size();
+ hint_iterator = (*found + 1) % ticks.size();
return *found;
}
diff --git a/src/video_core/renderer_vulkan/vk_resource_pool.h b/src/video_core/renderer_vulkan/vk_resource_pool.h
index a018c7ec2..9d0bb3b4d 100644
--- a/src/video_core/renderer_vulkan/vk_resource_pool.h
+++ b/src/video_core/renderer_vulkan/vk_resource_pool.h
@@ -36,7 +36,7 @@ private:
MasterSemaphore& master_semaphore;
size_t grow_step = 0; ///< Number of new resources created after an overflow
- size_t free_iterator = 0; ///< Hint to where the next free resources is likely to be found
+ 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
};
diff --git a/src/video_core/renderer_vulkan/vk_state_tracker.cpp b/src/video_core/renderer_vulkan/vk_state_tracker.cpp
index e81fad007..956f86845 100644
--- a/src/video_core/renderer_vulkan/vk_state_tracker.cpp
+++ b/src/video_core/renderer_vulkan/vk_state_tracker.cpp
@@ -18,9 +18,7 @@
#define NUM(field_name) (sizeof(Maxwell3D::Regs::field_name) / (sizeof(u32)))
namespace Vulkan {
-
namespace {
-
using namespace Dirty;
using namespace VideoCommon::Dirty;
using Tegra::Engines::Maxwell3D;
@@ -128,6 +126,34 @@ void SetupDirtyStencilTestEnable(Tables& tables) {
tables[0][OFF(stencil_enable)] = StencilTestEnable;
}
+void SetupDirtyBlending(Tables& tables) {
+ tables[0][OFF(color_mask_common)] = Blending;
+ tables[0][OFF(independent_blend_enable)] = Blending;
+ FillBlock(tables[0], OFF(color_mask), NUM(color_mask), Blending);
+ FillBlock(tables[0], OFF(blend), NUM(blend), Blending);
+ FillBlock(tables[0], OFF(independent_blend), NUM(independent_blend), Blending);
+}
+
+void SetupDirtyInstanceDivisors(Tables& tables) {
+ static constexpr size_t divisor_offset = 3;
+ for (size_t index = 0; index < Regs::NumVertexArrays; ++index) {
+ tables[0][OFF(instanced_arrays) + index] = InstanceDivisors;
+ tables[0][OFF(vertex_array) + index * NUM(vertex_array[0]) + divisor_offset] =
+ InstanceDivisors;
+ }
+}
+
+void SetupDirtyVertexAttributes(Tables& tables) {
+ FillBlock(tables[0], OFF(vertex_attrib_format), NUM(vertex_attrib_format), VertexAttributes);
+}
+
+void SetupDirtyViewportSwizzles(Tables& tables) {
+ static constexpr size_t swizzle_offset = 6;
+ for (size_t index = 0; index < Regs::NumViewports; ++index) {
+ tables[0][OFF(viewport_transform) + index * NUM(viewport_transform[0]) + swizzle_offset] =
+ ViewportSwizzles;
+ }
+}
} // Anonymous namespace
StateTracker::StateTracker(Tegra::GPU& gpu)
@@ -148,6 +174,10 @@ StateTracker::StateTracker(Tegra::GPU& gpu)
SetupDirtyFrontFace(tables);
SetupDirtyStencilOp(tables);
SetupDirtyStencilTestEnable(tables);
+ SetupDirtyBlending(tables);
+ SetupDirtyInstanceDivisors(tables);
+ SetupDirtyVertexAttributes(tables);
+ SetupDirtyViewportSwizzles(tables);
}
} // namespace Vulkan
diff --git a/src/video_core/renderer_vulkan/vk_state_tracker.h b/src/video_core/renderer_vulkan/vk_state_tracker.h
index c335d2bdf..84e918a71 100644
--- a/src/video_core/renderer_vulkan/vk_state_tracker.h
+++ b/src/video_core/renderer_vulkan/vk_state_tracker.h
@@ -35,6 +35,11 @@ enum : u8 {
StencilOp,
StencilTestEnable,
+ Blending,
+ InstanceDivisors,
+ VertexAttributes,
+ ViewportSwizzles,
+
Last
};
static_assert(Last <= std::numeric_limits<u8>::max());
diff --git a/src/video_core/shader/async_shaders.cpp b/src/video_core/shader/async_shaders.cpp
index 3b40db9bc..02adcf9c7 100644
--- a/src/video_core/shader/async_shaders.cpp
+++ b/src/video_core/shader/async_shaders.cpp
@@ -64,6 +64,7 @@ void AsyncShaders::FreeWorkers() {
void AsyncShaders::KillWorkers() {
is_thread_exiting.store(true);
+ cv.notify_all();
for (auto& thread : worker_threads) {
thread.detach();
}
generated by cgit v1.2.3 (git 2.25.1) at 2024-07-08 05:43:36 +0200