File tree Expand file tree Collapse file tree 1 file changed +1
-8
lines changed
Filter options
Expand file tree Collapse file tree 1 file changed +1
-8
lines changed
Original file line number Diff line number Diff line change @@ -9,23 +9,16 @@ RUN python3 -m pip install --upgrade pip cmake scikit-build-core[pyproject] setu
9
9
10
10
COPY . .
11
11
12
- # Install to /usr so that easily findable by cmake
13
- RUN mv /OpenBLAS /opt/OpenBLAS && cd /opt/OpenBLAS && make install PREFIX=/usr/ && cd /
14
-
15
- # Have to disable GGML_LLAMAFILE for Q4_0_4_4 quantization
16
12
ENV march=armv8.2-a+crypto+fp16+rcpc+dotprod
17
13
ENV mcpu=cortex-a78c+crypto+noprofile+nossbs+noflagm+nopauth
18
14
ENV mtune=cortex-a78c
19
15
20
16
ENV compiler_flags="-march=${march} -mcpu=${mcpu} -mtune=${mtune}"
21
17
22
- # This is the build that uses OpenBLAS + LLAMAFILE
23
- # RUN PKG_CONFIG_PATH="/opt/OpenBLAS/install/lib/pkgconfig" CMAKE_ARGS="-DGGML_BLAS=ON -DGGML_BLAS_VENDOR=OpenBLAS -DGGML_LLAMAFILE=ON -DCMAKE_C_FLAGS='${compiler_flags}' -DCMAKE_CXX_FLAGS='${compiler_flags}'" pip install -v -e .[server] 2>&1 | tee buildlog.txt
24
- # This is a release build that works
18
+ # This is a release build that works (have to disable GGML_LLAMAFILE for Q4_0_4_4 quantization)
25
19
RUN CMAKE_ARGS="-DGGML_LLAMAFILE=OFF -DCMAKE_C_FLAGS='${compiler_flags}' -DCMAKE_CXX_FLAGS='${compiler_flags}' -DCMAKE_BUILD_TYPE=Release" pip install -v -e .[server] 2>&1 | tee buildlog.txt
26
20
27
21
RUN cd /root && pyinstaller -DF /llama_cpp/server/__main__.py \
28
- --add-data /usr/lib/libopenblas.so:. \
29
22
--add-data /llama_cpp/lib/libllama.so:llama_cpp/lib \
30
23
--add-data /llama_cpp/lib/libggml.so:llama_cpp/lib \
31
24
-n llama-cpp-py-server
You can’t perform that action at this time.
0 commit comments