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

llamafile : improve sgemm.cpp (#6796)

Browse Source 
* llamafile : improve sgemm.cpp

- Re-enable by default
- Fix issue described in #6716
- Make code more abstract, elegant, and maintainable
- Faster handling of weirdly shaped `m` an `n` edge cases

* Address review comments

* Help clang produce fma instructions

* Address review comments
This commit is contained in:
Justine Tunney
2024-04-22 15:00:36 -04:00
committed by GitHub
parent e931888d50
commit 192090bae4
4 changed files with 406 additions and 567 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
CMakeLists.txt
View File

@ -43,17 +43,11 @@ else()
set(LLAMA_METAL_DEFAULT OFF)
endif()
# TODO: fix this for Android CI
# https://github.com/ggerganov/llama.cpp/pull/6716#issuecomment-2061509191
#if (CMAKE_SYSTEM_NAME MATCHES "ANDROID")
# set(LLAMA_LLAMAFILE_DEFAULT OFF)
#else()
# set(LLAMA_LLAMAFILE_DEFAULT ON)
#endif()
# TODO: temporary disable until MoE is fixed
# https://github.com/ggerganov/llama.cpp/pull/6716
set(LLAMA_LLAMAFILE_DEFAULT OFF)
if (CMAKE_SYSTEM_NAME MATCHES "ANDROID")
set(LLAMA_LLAMAFILE_DEFAULT OFF)
else()
set(LLAMA_LLAMAFILE_DEFAULT ON)
endif()
# general
option(BUILD_SHARED_LIBS "build shared libraries" OFF)