vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

fix A800 not supper 'movmatrix' module

Browse source
This commit is contained in:
glide-the 2025-02-10 18:08:28 +08:00
parent d7b31a9d84
commit 28dd5c505d
2 changed files with 2 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

1
ggml/src/ggml-cuda/common.cuh
View file

@ -47,6 +47,7 @@
#define GGML_CUDA_CC_TURING 750
#define GGML_CUDA_CC_AMPERE 800
#define GGML_CUDA_CC_OFFSET_AMD 0x1000000
#define GGML_CUDA_CC_HOPPER 900
// GCN/CNDA, wave size is 64
#define GGML_CUDA_CC_GCN4 (GGML_CUDA_CC_OFFSET_AMD + 0x803) // Tonga, Fiji, Polaris, minimum for fast fp16

2
ggml/src/ggml-cuda/mma.cuh
View file

@ -16,7 +16,7 @@
#include "common.cuh"
#if CUDART_VERSION >= 11080
#if (CUDART_VERSION >= 11080) && (__CUDA_ARCH__ >= GGML_CUDA_CC_HOPPER)
static __device__ __forceinline__ int ggml_cuda_movmatrix(const int x) {
int ret = 0;