Mirrors/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-12-27 06:39:25 +01:00
Code Issues Packages Projects Releases Wiki Activity

ggml : temporary disable llamafile sgemm until fixed

Browse Source 
ggml-ci
This commit is contained in:
Georgi Gerganov 2024-04-16 22:41:03 +03:00
parent 8cc91dc63c
commit f02ea667c1
No known key found for this signature in database
GPG Key ID: 449E073F9DC10735
3 changed files with 12 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
CMakeLists.txt
View File

@ -88,6 +88,7 @@ endif()
# 3rd party libs # 3rd party libs
option(LLAMA_ACCELERATE "llama: enable Accelerate framework" ON) option(LLAMA_ACCELERATE "llama: enable Accelerate framework" ON)
option(LLAMA_BLAS "llama: use BLAS" OFF) option(LLAMA_BLAS "llama: use BLAS" OFF)
option(LLAMA_LLAMAFILE "llama: use llamafile SGEMM" ON)
set(LLAMA_BLAS_VENDOR "Generic" CACHE STRING "llama: BLAS library vendor") set(LLAMA_BLAS_VENDOR "Generic" CACHE STRING "llama: BLAS library vendor")
option(LLAMA_CUDA "llama: use CUDA" OFF) option(LLAMA_CUDA "llama: use CUDA" OFF)
option(LLAMA_CUBLAS "llama: use CUDA (deprecated, use LLAMA_CUDA)" OFF) option(LLAMA_CUBLAS "llama: use CUDA (deprecated, use LLAMA_CUDA)" OFF)
@ -286,6 +287,7 @@ if (LLAMA_METAL)
${METALKIT_FRAMEWORK} ${METALKIT_FRAMEWORK}
) )
endif() endif()
if (LLAMA_BLAS) if (LLAMA_BLAS)
if (LLAMA_STATIC) if (LLAMA_STATIC)
set(BLA_STATIC ON) set(BLA_STATIC ON)
@ -368,6 +370,11 @@ if (LLAMA_BLAS)
endif() endif()
endif() endif()
if (LLAMA_LLAMAFILE)
# TODO: temporary disabled until test-backend-ops becomes green
#add_compile_definitions(GGML_USE_LLAMAFILE)
endif()
if (LLAMA_QKK_64) if (LLAMA_QKK_64)
add_compile_definitions(GGML_QKK_64) add_compile_definitions(GGML_QKK_64)
endif() endif()

4
Makefile
View File

@ -222,6 +222,10 @@ endif # LLAMA_DISABLE_LOGS
# disable ggml.c's use of sgemm.cpp # disable ggml.c's use of sgemm.cpp
ifdef LLAMA_NO_LLAMAFILE ifdef LLAMA_NO_LLAMAFILE
MK_CPPFLAGS += -DGGML_USE_LLAMAFILE=0 MK_CPPFLAGS += -DGGML_USE_LLAMAFILE=0
else
# TODO: temporary disabled until test-backend-ops becomes green
#MK_CPPFLAGS += -DGGML_USE_LLAMAFILE=1
MK_CPPFLAGS += -DGGML_USE_LLAMAFILE=0
endif endif
# warnings # warnings

6
ggml.c
View File

@ -33,12 +33,8 @@
#include <unistd.h> #include <unistd.h>
#endif #endif
#ifndef GGML_USE_LLAMAFILE
#ifdef __ARM_FEATURE_MATMUL_INT8 #ifdef __ARM_FEATURE_MATMUL_INT8
#define GGML_USE_LLAMAFILE 0 #undef GGML_USE_LLAMAFILE
#else
#define GGML_USE_LLAMAFILE 1
#endif
#endif #endif
#if defined(_MSC_VER) #if defined(_MSC_VER)