mirror of
https://github.com/ggerganov/whisper.cpp.git
synced 2024-12-19 04:37:51 +00:00
Allow all RDNA2 archs to use sdot4 intrinsic (llama/8629)
The check gating the use of `__builtin_amdgc_sdot4` specifically checks for gfx1030. This causes a severe perf regression for anything gfx103? that's not gfx1030 and not using `HSA_OVERRIDE_GFX_VERSION` (if you've built ROCm to support it). We already have a generic RDNA2 define, let's use it.
This commit is contained in:
parent
11182fae34
commit
86506b0c5c
@ -459,7 +459,7 @@ static __device__ __forceinline__ uint32_t __hgt2_mask(const half2 a, const half
|
||||
|
||||
static __device__ __forceinline__ int ggml_cuda_dp4a(const int a, const int b, int c) {
|
||||
#if defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__)
|
||||
#if defined(__gfx906__) || defined(__gfx908__) || defined(__gfx90a__) || defined(__gfx1030__)
|
||||
#if defined(__gfx906__) || defined(__gfx908__) || defined(__gfx90a__) || defined(RDNA2)
|
||||
c = __builtin_amdgcn_sdot4(a, b, c, false);
|
||||
#elif defined(RDNA3)
|
||||
c = __builtin_amdgcn_sudot4( true, a, true, b, c, false);
|
||||
|
Loading…
Reference in New Issue
Block a user