JDLOBrew/eden
16
0
Fork 0
Code Pull Requests Actions Packages Releases 4 Activity

[vk, pipeline_cache, qcom] Resolving pipeline usage for QCOM

Browse Source
This commit is contained in:
PavelBARABANOV
2025-11-18 22:30:43 +03:00
parent 0d3cef65ad
commit ea5e15cde5
2 changed files with 28 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

23
src/video_core/renderer_vulkan/vk_pipeline_cache.cpp
View File

@@ -677,7 +677,17 @@ std::unique_ptr<GraphicsPipeline> PipelineCache::CreateGraphicsPipeline(
const auto runtime_info{MakeRuntimeInfo(programs, key, program, previous_stage)};
ConvertLegacyToGeneric(program, runtime_info);
const std::vector<u32> code{EmitSPIRV(profile, runtime_info, program, binding, this->optimize_spirv_output)};
// Adreno don't support subgroup operations in vertex stages
// Disable subgroup features for vertex shaders if not supported by the device
Shader::Profile stage_profile = profile;
if (program.stage == Shader::Stage::VertexA || program.stage == Shader::Stage::VertexB) {
if (!device.IsSubgroupSupportedForStage(VK_SHADER_STAGE_VERTEX_BIT)) {
stage_profile.support_vote = false;
}
}
const std::vector<u32> code{EmitSPIRV(stage_profile, runtime_info, program, binding, this->optimize_spirv_output)};
device.SaveShader(code);
modules[stage_index] = BuildShader(device, code);
if (device.HasDebuggingToolAttached()) {
@@ -771,6 +781,17 @@ std::unique_ptr<ComputePipeline> PipelineCache::CreateComputePipeline(
}
auto program{TranslateProgram(pools.inst, pools.block, env, cfg, host_info)};
// Adreno have lower shared memory limits (32KB)
// Clamp shared memory usage to device maximum to avoid validation errors
const u32 max_shared_memory = device.GetMaxComputeSharedMemorySize();
if (program.shared_memory_size > max_shared_memory) {
LOG_WARNING(Render_Vulkan,
"Compute shader 0x{:016x} requests {}KB shared memory but device max is {}KB - clamping",
key.unique_hash, program.shared_memory_size / 1024, max_shared_memory / 1024);
program.shared_memory_size = max_shared_memory;
}
const std::vector<u32> code{EmitSPIRV(profile, program, this->optimize_spirv_output)};
device.SaveShader(code);
vk::ShaderModule spv_module{BuildShader(device, code)};

6
src/video_core/vulkan_common/vulkan_device.h
View File

@@ -365,6 +365,12 @@ public:
return properties.subgroup_properties.supportedOperations & feature;
}
/// Returns true if subgroup operations are supported in the specified shader stage.
/// Mobile GPUs (Qualcomm Adreno) often only support subgroups in fragment/compute stages.
bool IsSubgroupSupportedForStage(VkShaderStageFlagBits stage) const {
return properties.subgroup_properties.supportedStages & stage;
}
/// Returns the maximum number of push descriptors.
u32 MaxPushDescriptors() const {
return properties.push_descriptor.maxPushDescriptors;