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

rocblas init stuff

Browse source
This commit is contained in:
Henri Vasserman 2023-07-29 14:16:46 +03:00
parent d2ade639f4
commit f8e3fc6c74
No known key found for this signature in database
GPG key ID: 2995FC0F58B1A986
1 changed files with 10 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

13
ggml-cuda.cu
View file

@ -10,7 +10,10 @@
#include <hip/hip_runtime.h> #include <hip/hip_runtime.h>
#include <hipblas/hipblas.h> #include <hipblas/hipblas.h>
#include <hip/hip_fp16.h> #include <hip/hip_fp16.h>
#ifdef __HIP_PLATFORM_AMD__
// for rocblas_initialize()
#include "rocblas/rocblas.h" #include "rocblas/rocblas.h"
#endif
#define CUBLAS_COMPUTE_32F HIPBLAS_R_32F #define CUBLAS_COMPUTE_32F HIPBLAS_R_32F
#define CUBLAS_COMPUTE_32F_FAST_16F HIPBLAS_R_32F #define CUBLAS_COMPUTE_32F_FAST_16F HIPBLAS_R_32F
#define CUBLAS_GEMM_DEFAULT HIPBLAS_GEMM_DEFAULT #define CUBLAS_GEMM_DEFAULT HIPBLAS_GEMM_DEFAULT
@ -2746,10 +2749,14 @@ void ggml_init_cublas() {
static bool initialized = false; static bool initialized = false;
if (!initialized) { if (!initialized) {
#ifdef GGML_USE_HIPBLAS
rocblas_initialize(); #ifdef __HIP_PLATFORM_AMD__
hipDeviceSynchronize(); // Workaround for a rocBLAS bug when using multiple graphics cards:
// https://github.com/ROCmSoftwarePlatform/rocBLAS/issues/1346
rocblas_initialize();
CUDA_CHECK(cudaDeviceSynchronize());
#endif #endif
CUDA_CHECK(cudaGetDeviceCount(&g_device_count)); CUDA_CHECK(cudaGetDeviceCount(&g_device_count));
GGML_ASSERT(g_device_count <= GGML_CUDA_MAX_DEVICES); GGML_ASSERT(g_device_count <= GGML_CUDA_MAX_DEVICES);
int64_t total_vram = 0; int64_t total_vram = 0;