mrq/bitsandbytes-rocm
1
0
Fork 1
Code Issues 1 Pull Requests Packages Projects Releases Wiki Activity

drop support for gfx903 because depending on hipblaslt gums up too many things

Browse Source
This commit is contained in:
mrq 2023-10-12 19:16:14 -05:00
parent e38b9e91b7
commit c88f97a9c8
2 changed files with 6 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
csrc/ops.cuh
View File

@ -21,8 +21,8 @@
#include <hip/hip_runtime_api.h>
#include <hip/hip_fp16.h>
#include <hipblas/hipblas.h>
#include <hipblaslt/hipblaslt.h> //only supports gfx903
#include <hipsparse/hipsparse.h>
#include <hip/hip_bfloat16.h>
#define cudaPeekAtLastError hipPeekAtLastError
#define cudaMemset hipMemset
@ -36,7 +36,7 @@
#define cublasStatus_t hipblasStatus_t
#define cublasGemmStridedBatchedEx hipblasGemmStridedBatchedEx
#define cublasOperation_t hipblasOperation_t
#define cublasLtMatrixLayoutCreate hipblasLtMatrixLayoutCreate
#define cublasLtMatrixLayoutCreate hipblasMatrixLayoutCreate
#define cudaError_t hipError_t
#define cudaGetErrorString hipGetErrorString
#define cudaSuccess hipSuccess
@ -49,8 +49,8 @@
#define cusparseHandle_t hipsparseHandle_t
#define cusparseCreate hipsparseCreate
#define __nv_bfloat16 hip_bfloat16
#define cublasLtHandle_t hipblasLtHandle_t
#define cublasLtCreate hipblasLtCreate
#define cublasLtHandle_t hipblasHandle_t
#define cublasLtCreate hipblasCreate
#define CUBLAS_GEMM_DEFAULT HIPBLAS_GEMM_DEFAULT
#define CUBLAS_GEMM_DEFAULT_TENSOR_OP HIPBLAS_GEMM_DEFAULT //TODO: HIP didn't have the right one, might cause issues

4
csrc/pythonInterface.c
View File

@ -7,8 +7,8 @@
#include <ops.cuh>
#ifdef BITS_AND_BYTES_USE_ROCM
#include <hipblaslt/hipblaslt.h>
#define cublasLtHandle_t hipblasLtHandle_t
#include <hipblas/hipblas.h>
#define cublasLtHandle_t hipblasHandle_t
#define cudaMemAttachHost hipMemAttachHost
#define cudaMallocManaged hipMallocManaged
#define cudaDevAttrConcurrentManagedAccess hipDeviceAttributeConcurrentManagedAccess