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

build: revert more --config Release

Browse source
This commit is contained in:
Olivier Chafik 2024-04-29 16:22:36 +01:00
parent cbf99fcf3b
commit 4b39329431
2 changed files with 8 additions and 8 deletions
Download patch file Download diff file Expand all files Collapse all files

2
.github/workflows/server.yml vendored
View file

@ -100,7 +100,7 @@ jobs:
-DLLAMA_CURL=ON \ -DLLAMA_CURL=ON \
-DCMAKE_BUILD_TYPE=${{ matrix.build_type }} \ -DCMAKE_BUILD_TYPE=${{ matrix.build_type }} \
-DLLAMA_SANITIZE_${{ matrix.sanitizer }}=ON ; -DLLAMA_SANITIZE_${{ matrix.sanitizer }}=ON ;
cmake --build build --config Release -j $(nproc) --target server cmake --build build --config ${{ matrix.build_type }} -j $(nproc) --target server
- name: Tests - name: Tests

14
README.md
View file

@ -455,7 +455,7 @@ Building the program with BLAS support may lead to some performance improvements
```bash ```bash
cmake -B build -DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS cmake -B build -DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS
cmake --build build cmake --build build --config Release
``` ```
- #### BLIS - #### BLIS
@ -477,7 +477,7 @@ Building the program with BLAS support may lead to some performance improvements
```bash ```bash
source /opt/intel/oneapi/setvars.sh # You can skip this step if in oneapi-basekit docker image, only required for manual installation source /opt/intel/oneapi/setvars.sh # You can skip this step if in oneapi-basekit docker image, only required for manual installation
cmake -B build -DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=Intel10_64lp -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DLLAMA_NATIVE=ON cmake -B build -DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=Intel10_64lp -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DLLAMA_NATIVE=ON
cmake --build build cmake --build build --config Release
``` ```
- Using oneAPI docker image: - Using oneAPI docker image:
@ -499,7 +499,7 @@ Building the program with BLAS support may lead to some performance improvements
```bash ```bash
cmake -B build -DLLAMA_CUDA=ON cmake -B build -DLLAMA_CUDA=ON
cmake --build build cmake --build build --config Release
``` ```
The environment variable [`CUDA_VISIBLE_DEVICES`](https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#env-vars) can be used to specify which GPU(s) will be used. The following compilation options are also available to tweak performance: The environment variable [`CUDA_VISIBLE_DEVICES`](https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#env-vars) can be used to specify which GPU(s) will be used. The following compilation options are also available to tweak performance:
@ -527,7 +527,7 @@ Building the program with BLAS support may lead to some performance improvements
```bash ```bash
CC=/opt/rocm/llvm/bin/clang CXX=/opt/rocm/llvm/bin/clang++ \ CC=/opt/rocm/llvm/bin/clang CXX=/opt/rocm/llvm/bin/clang++ \
cmake -H. -Bbuild -DLLAMA_HIPBLAS=ON -DAMDGPU_TARGETS=gfx1030 -DCMAKE_BUILD_TYPE=Release \ cmake -H. -Bbuild -DLLAMA_HIPBLAS=ON -DAMDGPU_TARGETS=gfx1030 -DCMAKE_BUILD_TYPE=Release \
&& cmake --build build --config Release -j 16 && cmake --build build --config Release -- -j 16
``` ```
On Linux it is also possible to use unified memory architecture (UMA) to share main memory between the CPU and integrated GPU by setting `-DLLAMA_HIP_UMA=ON"`. On Linux it is also possible to use unified memory architecture (UMA) to share main memory between the CPU and integrated GPU by setting `-DLLAMA_HIP_UMA=ON"`.
However, this hurts performance for non-integrated GPUs (but enables working with integrated GPUs). However, this hurts performance for non-integrated GPUs (but enables working with integrated GPUs).
@ -617,7 +617,7 @@ Building the program with BLAS support may lead to some performance improvements
git clone https://github.com/CNugteren/CLBlast.git git clone https://github.com/CNugteren/CLBlast.git
cd CLBlast cd CLBlast
cmake -B build -DBUILD_SHARED_LIBS=OFF -DTUNERS=OFF cmake -B build -DBUILD_SHARED_LIBS=OFF -DTUNERS=OFF
cmake --build build cmake --build build --config Release
cmake --install build --prefix /some/path cmake --install build --prefix /some/path
``` ```
@ -633,7 +633,7 @@ Building the program with BLAS support may lead to some performance improvements
- CMake (Unix): - CMake (Unix):
```sh ```sh
cmake -B build -DLLAMA_CLBLAST=ON -DCLBlast_DIR=/some/path cmake -B build -DLLAMA_CLBLAST=ON -DCLBlast_DIR=/some/path
cmake --build build cmake --build build --config Release
``` ```
- CMake (Windows): - CMake (Windows):
```cmd ```cmd
@ -699,7 +699,7 @@ Building the program with BLAS support may lead to some performance improvements
```bash ```bash
cmake -B build -DLLAMA_VULKAN=1 cmake -B build -DLLAMA_VULKAN=1
cmake --build build cmake --build build --config Release
# Test the output binary (with "-ngl 33" to offload all layers to GPU) # Test the output binary (with "-ngl 33" to offload all layers to GPU)
./bin/main -m "PATH_TO_MODEL" -p "Hi you how are you" -n 50 -e -ngl 33 -t 4 ./bin/main -m "PATH_TO_MODEL" -p "Hi you how are you" -n 50 -e -ngl 33 -t 4