llama.cpp/ggml-cuda.h
slaren 58b367c2d7
cuBLAS: refactor and optimize f16 mat mul performance (#1259)
* cuBLAS: refactor, convert fp16 to fp32 on device

* cuBLAS: use multiple streams, choose smartly between mul_mat_q and mul_mat_f16

* fix build

* cuBLAS: update block_q5_1
2023-05-01 18:11:07 +02:00

19 lines
638 B
C

#include "ggml.h"
#ifdef __cplusplus
extern "C" {
#endif
void ggml_init_cublas(void);
bool ggml_cuda_can_mul_mat(const struct ggml_tensor * src0, const struct ggml_tensor * src1, struct ggml_tensor * dst);
size_t ggml_cuda_mul_mat_get_wsize(const struct ggml_tensor * src0, const struct ggml_tensor * src1, struct ggml_tensor * dst);
void ggml_cuda_mul_mat(const struct ggml_tensor * src0, const struct ggml_tensor * src1, struct ggml_tensor * dst, void * wdata, size_t wsize);
// TODO: export these with GGML_API
void * ggml_cuda_host_malloc(size_t size);
void ggml_cuda_host_free(void * ptr);
#ifdef __cplusplus
}
#endif