ver4a/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

backend cpu: add online flow for aarch64 Q4_0 GEMV/GEMM kernels (#9921)

Browse source 
* backend-cpu: add online flow for aarch64 Q4_0 GEMV/GEMM kernels

---------

Co-authored-by: Diego Devesa <slarengh@gmail.com>
This commit is contained in:
Charles Xu 2024-11-15 01:28:50 +01:00 • committed by GitHub
parent ae8de6d50a
commit 1607a5e5b0
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
9 changed files with 273 additions and 22 deletions
Download patch file Download diff file Expand all files Collapse all files

5
ggml/src/ggml-cpu/CMakeLists.txt
View file

@ -236,6 +236,11 @@ else()
message(STATUS "Unknown architecture")
endif()
if (GGML_CPU_AARCH64)
message(STATUS "Using runtime weight conversion of Q4_0 to Q4_0_x_x to enable optimized GEMM/GEMV kernels")
add_compile_definitions(GGML_USE_CPU_AARCH64)
endif()
target_compile_options(ggml-cpu PRIVATE "$<$<COMPILE_LANGUAGE:CXX>:${ARCH_FLAGS}>")
target_compile_options(ggml-cpu PRIVATE "$<$<COMPILE_LANGUAGE:C>:${ARCH_FLAGS}>")