happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Workaround for subgroup arithmetic failing on MoltenVK with AMD GPUs (issue 15846) (#15886)

This commit is contained in:
lksj92hs 2025-09-09 15:01:15 +03:00 committed by GitHub
parent a972faebed
commit ed54e32558
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 6 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
ggml/src/ggml-vulkan/ggml-vulkan.cpp
View File

@ -3736,6 +3736,12 @@ static vk_device ggml_vk_get_device(size_t idx) {
device->subgroup_arithmetic = (vk11_props.subgroupSupportedStages & vk::ShaderStageFlagBits::eCompute) &&
(vk11_props.subgroupSupportedOperations & vk::SubgroupFeatureFlagBits::eArithmetic);
#ifdef __APPLE__
// Workaround for subgroup arithmetic failing on MoltenVK with AMD GPUs (issue 15846)
if (device->vendor_id == VK_VENDOR_ID_AMD) {
device->subgroup_arithmetic = false;
}
#endif
device->subgroup_shuffle = (vk11_props.subgroupSupportedStages & vk::ShaderStageFlagBits::eCompute) &&
(vk11_props.subgroupSupportedOperations & vk::SubgroupFeatureFlagBits::eShuffle);
device->subgroup_clustered = (vk11_props.subgroupSupportedStages & vk::ShaderStageFlagBits::eCompute) &&