vulkan: opt mul_mat_vecq for mi50 (#22933)
This commit is contained in:
@@ -4719,7 +4719,7 @@ static void ggml_vk_load_shaders(vk_device& device, vk_pipeline requested) {
|
||||
}
|
||||
uint32_t rm_iq = 2 * rm_kq;
|
||||
|
||||
const bool use_subgroups = device->subgroup_arithmetic && device->architecture != vk_device_architecture::AMD_GCN;
|
||||
const bool use_subgroups = device->subgroup_arithmetic;
|
||||
// Ensure a subgroup size >= 16 is available
|
||||
const bool use_subgroups16 = use_subgroups && subgroup_min_size_16;
|
||||
|
||||
|
||||
Reference in New Issue
Block a user