raphiaura/KytyPS5
1
0
Fork 0
mirror of https://github.com/KytyPS5/KytyPS5.git synced 2026-10-20 10:56:58 +02:00
Code Issues Packages Projects Releases Wiki Activity

refactor(graphics)

Browse Source
This commit is contained in:
nmzik
2026-09-12 13:38:47 +04:00
parent f2455c875e
commit 25a44371e3
53 changed files with 708 additions and 1315 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/graphics/host_gpu/renderer/renderCompute.cpp
+12 -18
View File
@@ -36,18 +36,12 @@
#include <vector>
namespace Libs::Graphics {
static uint64_t BufferDescriptorSize(const ShaderBufferResource& descriptor) {
const uint64_t records = descriptor.NumRecords();
const uint64_t stride = descriptor.Stride();
return stride == 0 ? records : records * stride;
}
static bool FillSourcesDisjoint(std::span<const ShaderRecompiler::IR::DescriptorValue> sources,
GuestRange destination, uint32_t output_buffer = UINT32_MAX) {
for (uint32_t i = 0; i < sources.size(); ++i) {
if (i == output_buffer) continue;
const auto source = DecodeNativeDescriptor<ShaderBufferResource>(sources[i]);
const auto bytes = BufferDescriptorSize(source);
const auto bytes = source.GetSize();
if (source.Base48() < destination.End() && destination.address < source.Base48() + bytes)
return false;
}
@@ -67,7 +61,7 @@ bool RenderExecutor::TryConsumeComputeMetaClear(const ShaderComputeInputInfo& in
const auto descriptor = DecodeNativeDescriptor<ShaderBufferResource>(resources.buffers[i]);
// A metadata resource that is also read is not proven to be a full overwrite. Execute it
// conservatively instead of replacing the dispatch with a coarse full-surface clear.
if (cache.IsMeta(descriptor.Base48()) && (!resource.written || resource.read)) {
if ((!resource.written || resource.read) && cache.IsMeta(descriptor.Base48())) {
return false;
}
}
@@ -115,7 +109,7 @@ bool ResolveComputeBufferFill(const ShaderComputeInputInfo& input, uint32_t grou
const uint64_t invocations = input.dispatch_thread_dimensions
? group_x
: static_cast<uint64_t>(group_x) * input.threads_num[0];
const auto size = BufferDescriptorSize(descriptor);
const auto size = descriptor.GetSize();
if (invocations != descriptor.NumRecords() || size == 0 || size > UINT32_MAX ||
(input.dispatch_thread_dimensions &&
(group_x % input.threads_num[0] != 0 || input.dispatch_threads_num[0] != group_x ||
@@ -264,9 +258,6 @@ void RenderExecutor::DispatchDirect(uint64_t submit_id, CommandBuffer& buffer,
input_info.dispatch_threads_num[2] = thread_group_z;
}
const uint32_t frame_num = static_cast<uint32_t>(m_context.GetGpu().GetFrameNum());
const bool large_workgroup =
(input_info.threads_num[0] * input_info.threads_num[1] * input_info.threads_num[2] >= 512);
const auto& program = *input_info.stage.program;
const auto& resources = input_info.stage.resources;
if (TryConsumeComputeMetaClear(input_info, buffer)) {
@@ -278,14 +269,17 @@ void RenderExecutor::DispatchDirect(uint64_t submit_id, CommandBuffer& buffer,
ResetBindings();
return;
}
const auto sampled_images = std::count_if(
program.info.images.begin(), program.info.images.end(), [](const auto& image) {
return image.resource_class == ShaderRecompiler::IR::ImageResourceClass::Sampled;
});
const bool large_workgroup =
(input_info.threads_num[0] * input_info.threads_num[1] * input_info.threads_num[2] >= 512);
const bool has_sampler = !program.info.samplers.empty();
static std::atomic<uint32_t> dispatch_log_count {0};
if ((large_workgroup || has_sampler) &&
dispatch_log_count.fetch_add(1, std::memory_order_relaxed) < 512) {
const auto sampled_images = std::count_if(
program.info.images.begin(), program.info.images.end(), [](const auto& image) {
return image.resource_class == ShaderRecompiler::IR::ImageResourceClass::Sampled;
});
const uint32_t frame_num = static_cast<uint32_t>(m_context.GetGpu().GetFrameNum());
LOGF("GraphicsRenderDispatchDirect: frame=%u shader=0x%016" PRIx64
" groups=%ux%ux%u mode=0x%08" PRIx32 " local=%ux%ux%u "
"buffers=%zu textures=%zu sampled=%zu storage=%zu samplers=%zu push=%u\n",
@@ -374,11 +368,11 @@ void RenderExecutor::DispatchDirect(uint64_t submit_id, CommandBuffer& buffer,
buffer.EndRendering();
auto& pipeline =
m_context.GetPipelineCache().CreateComputePipeline(input_info, compute_program);
m_context.GetPipelineCache().GetComputePipeline(input_info, compute_program);
auto bindings = PrepareBindings(input_info.stage);
FindBuffers(bindings);
if (program.info.uses_dma) {
m_context.GetGpuResources().PrepareBda();
m_context.PrepareBda();
}
RebindBuffers(bindings);
RebindImages(bindings);