tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-06-26 19:55:04 +00:00
Code Issues Packages Projects Releases Wiki Activity

CUDA: enable peer access between devices (#2470)

Browse Source
This commit is contained in:
Johannes Gäßler
2023-09-17 16:37:53 +02:00
committed by GitHub
parent 8b428c9bc8
commit 111163e246
4 changed files with 63 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
CMakeLists.txt
View File

@ -80,6 +80,8 @@ set(LLAMA_CUDA_DMMV_X "32" CACHE STRING "llama: x stride for dmmv CUDA kern
set(LLAMA_CUDA_MMV_Y "1" CACHE STRING "llama: y block size for mmv CUDA kernels")
option(LLAMA_CUDA_F16 "llama: use 16 bit floats for some calculations" OFF)
set(LLAMA_CUDA_KQUANTS_ITER "2" CACHE STRING "llama: iters./thread per block for Q2_K/Q6_K")
set(LLAMA_CUDA_PEER_MAX_BATCH_SIZE "128" CACHE STRING
"llama: max. batch size for using peer access")
option(LLAMA_HIPBLAS "llama: use hipBLAS" OFF)
option(LLAMA_CLBLAST "llama: use CLBlast" OFF)
option(LLAMA_METAL "llama: use Metal" ${LLAMA_METAL_DEFAULT})
@ -304,6 +306,7 @@ if (LLAMA_CUBLAS)
add_compile_definitions(GGML_CUDA_F16)
endif()
add_compile_definitions(K_QUANTS_PER_ITERATION=${LLAMA_CUDA_KQUANTS_ITER})
add_compile_definitions(GGML_CUDA_PEER_MAX_BATCH_SIZE=${LLAMA_CUDA_PEER_MAX_BATCH_SIZE})
if (LLAMA_STATIC)
set(LLAMA_EXTRA_LIBS ${LLAMA_EXTRA_LIBS} CUDA::cudart_static CUDA::cublas_static CUDA::cublasLt_static)