diff options
| author | Kawrakow <48489457+ikawrakow@users.noreply.github.com> | 2024-07-27 07:55:01 +0200 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-07-27 07:55:01 +0200 |
| commit | 154e0d75fccf1784fe9ff6fd76a630b66563da3d (patch) | |
| tree | 81ce6dbb5b1900c1aa78a879f0593c694cab9d27 /ggml-cuda.h | |
| parent | 0684c3e9c70d49323b4fc517128cbe222cab7f96 (diff) | |
Merge mainline llama.cpp (#3)
* Merging mainline - WIP
* Merging mainline - WIP
AVX2 and CUDA appear to work.
CUDA performance seems slightly (~1-2%) lower as it is so often
the case with llama.cpp/ggml after some "improvements" have been made.
* Merging mainline - fix Metal
* Remove check
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
Diffstat (limited to 'ggml-cuda.h')
| -rw-r--r-- | ggml-cuda.h | 44 |
1 files changed, 0 insertions, 44 deletions
diff --git a/ggml-cuda.h b/ggml-cuda.h deleted file mode 100644 index d7903c66..00000000 --- a/ggml-cuda.h +++ /dev/null @@ -1,44 +0,0 @@ -#pragma once - -#include "ggml.h" -#include "ggml-backend.h" - -#ifdef GGML_USE_HIPBLAS -#define GGML_CUDA_NAME "ROCm" -#define GGML_CUBLAS_NAME "hipBLAS" -#else -#define GGML_CUDA_NAME "CUDA" -#define GGML_CUBLAS_NAME "cuBLAS" -#endif - -#ifdef __cplusplus -extern "C" { -#endif - -#define GGML_CUDA_MAX_DEVICES 16 - -// backend API -GGML_API GGML_CALL ggml_backend_t ggml_backend_cuda_init(int device); - -GGML_API GGML_CALL bool ggml_backend_is_cuda(ggml_backend_t backend); - -// device buffer -GGML_API GGML_CALL ggml_backend_buffer_type_t ggml_backend_cuda_buffer_type(int device); - -// split tensor buffer that splits matrices by rows across multiple devices -GGML_API GGML_CALL ggml_backend_buffer_type_t ggml_backend_cuda_split_buffer_type(const float * tensor_split); - -// pinned host buffer for use with the CPU backend for faster copies between CPU and GPU -GGML_API GGML_CALL ggml_backend_buffer_type_t ggml_backend_cuda_host_buffer_type(void); - -GGML_API GGML_CALL int ggml_backend_cuda_get_device_count(void); -GGML_API GGML_CALL void ggml_backend_cuda_get_device_description(int device, char * description, size_t description_size); -GGML_API GGML_CALL void ggml_backend_cuda_get_device_memory(int device, size_t * free, size_t * total); - -GGML_API GGML_CALL bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size); -GGML_API GGML_CALL void ggml_backend_cuda_unregister_host_buffer(void * buffer); - -GGML_API void ggml_backend_cuda_log_set_callback(ggml_log_callback log_callback, void * user_data); -#ifdef __cplusplus -} -#endif |