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

readme: cmake . -B build && cmake --build build

Browse source
This commit is contained in:
ochafik 2024-04-28 16:04:49 +01:00
parent 6e472f58e4
commit d39085440d
10 changed files with 68 additions and 103 deletions
Download patch file Download diff file Expand all files Collapse all files

8
.devops/main-intel.Dockerfile
View file

@ -10,14 +10,12 @@ WORKDIR /app
COPY . .
RUN mkdir build && \
cd build && \
if [ "${LLAMA_SYCL_F16}" = "ON" ]; then \
RUN if [ "${LLAMA_SYCL_F16}" = "ON" ]; then \
echo "LLAMA_SYCL_F16 is set" && \
export OPT_SYCL_F16="-DLLAMA_SYCL_F16=ON"; \
fi && \
cmake .. -DLLAMA_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx ${OPT_SYCL_F16} && \
cmake --build . --config Release --target main
cmake . -B build -DLLAMA_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx ${OPT_SYCL_F16} && \
cmake --build build --target main
FROM intel/oneapi-basekit:$ONEAPI_VERSION as runtime

6
.devops/main-vulkan.Dockerfile
View file

@ -14,10 +14,8 @@ RUN wget -qO - https://packages.lunarg.com/lunarg-signing-key-pub.asc | apt-key
# Build it
WORKDIR /app
COPY . .
RUN mkdir build && \
cd build && \
cmake .. -DLLAMA_VULKAN=1 && \
cmake --build . --config Release --target main
RUN cmake . -B build -DLLAMA_VULKAN=1 && \
cmake --build build --target main
# Clean up
WORKDIR /

8
.devops/server-intel.Dockerfile
View file

@ -10,14 +10,12 @@ WORKDIR /app
COPY . .
RUN mkdir build && \
cd build && \
if [ "${LLAMA_SYCL_F16}" = "ON" ]; then \
RUN if [ "${LLAMA_SYCL_F16}" = "ON" ]; then \
echo "LLAMA_SYCL_F16 is set" && \
export OPT_SYCL_F16="-DLLAMA_SYCL_F16=ON"; \
fi && \
cmake .. -DLLAMA_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DLLAMA_CURL=ON ${OPT_SYCL_F16} && \
cmake --build . --config Release --target server
cmake . -B build -DLLAMA_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DLLAMA_CURL=ON ${OPT_SYCL_F16} && \
cmake --build build --target server
FROM intel/oneapi-basekit:$ONEAPI_VERSION as runtime

6
.devops/server-vulkan.Dockerfile
View file

@ -18,10 +18,8 @@ RUN apt-get update && \
# Build it
WORKDIR /app
COPY . .
RUN mkdir build && \
cd build && \
cmake .. -DLLAMA_VULKAN=1 -DLLAMA_CURL=1 && \
cmake --build . --config Release --target server
RUN cmake . -B build -DLLAMA_VULKAN=1 -DLLAMA_CURL=1 && \
cmake --build build --target server
# Clean up
WORKDIR /

6
.github/workflows/bench.yml vendored
View file

@ -96,9 +96,7 @@ jobs:
id: cmake_build
run: |
set -eux
mkdir build
cd build
cmake .. \
cmake . -B build \
-DLLAMA_NATIVE=OFF \
-DLLAMA_BUILD_SERVER=ON \
-DLLAMA_CURL=ON \
@ -109,7 +107,7 @@ jobs:
-DLLAMA_FATAL_WARNINGS=OFF \
-DLLAMA_ALL_WARNINGS=OFF \
-DCMAKE_BUILD_TYPE=Release;
cmake --build . --config Release -j $(nproc) --target server
cmake --build build -j $(nproc) --target server
- name: Download the dataset
id: download_dataset

12
.github/workflows/server.yml vendored
View file

@ -94,15 +94,13 @@ jobs:
- name: Build
id: cmake_build
run: |
mkdir build
cd build
cmake .. \
cmake . -B build \
-DLLAMA_NATIVE=OFF \
-DLLAMA_BUILD_SERVER=ON \
-DLLAMA_CURL=ON \
-DCMAKE_BUILD_TYPE=${{ matrix.build_type }} \
-DLLAMA_SANITIZE_${{ matrix.sanitizer }}=ON ;
cmake --build . --config ${{ matrix.build_type }} -j $(nproc) --target server
cmake --build build -j $(nproc) --target server
- name: Tests
@ -143,10 +141,8 @@ jobs:
- name: Build
id: cmake_build
run: |
mkdir build
cd build
cmake .. -DLLAMA_CURL=ON -DCURL_LIBRARY="$env:RUNNER_TEMP/libcurl/lib/libcurl.dll.a" -DCURL_INCLUDE_DIR="$env:RUNNER_TEMP/libcurl/include"
cmake --build . --config Release -j ${env:NUMBER_OF_PROCESSORS} --target server
cmake . -B build -DLLAMA_CURL=ON -DCURL_LIBRARY="$env:RUNNER_TEMP/libcurl/lib/libcurl.dll.a" -DCURL_INCLUDE_DIR="$env:RUNNER_TEMP/libcurl/include"
cmake --build build -j ${env:NUMBER_OF_PROCESSORS} --target server
- name: Python setup
id: setup_python

23
README-sycl.md
View file

@ -185,9 +185,8 @@ Upon a successful installation, SYCL is enabled for the available intel devices,
```sh
git clone https://github.com/oneapi-src/oneMKL
cd oneMKL
mkdir -p buildWithCublas && cd buildWithCublas
cmake ../ -DCMAKE_CXX_COMPILER=icpx -DCMAKE_C_COMPILER=icx -DENABLE_MKLGPU_BACKEND=OFF -DENABLE_MKLCPU_BACKEND=OFF -DENABLE_CUBLAS_BACKEND=ON -DTARGET_DOMAINS=blas
make
cmake -B buildWithCublas . -DCMAKE_CXX_COMPILER=icpx -DCMAKE_C_COMPILER=icx -DENABLE_MKLGPU_BACKEND=OFF -DENABLE_MKLCPU_BACKEND=OFF -DENABLE_CUBLAS_BACKEND=ON -DTARGET_DOMAINS=blas
cmake --build buildWithCublas
```
@ -227,16 +226,15 @@ Similarly, user targeting Nvidia GPUs should expect at least one SYCL-CUDA devic
source /opt/intel/oneapi/setvars.sh
# Build LLAMA with MKL BLAS acceleration for intel GPU
mkdir -p build && cd build
# Option 1: Use FP32 (recommended for better performance in most cases)
cmake .. -DLLAMA_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx
cmake . -B build -DLLAMA_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx
# Option 2: Use FP16
cmake .. -DLLAMA_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DLLAMA_SYCL_F16=ON
cmake . -B build -DLLAMA_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DLLAMA_SYCL_F16=ON
#build all binary
cmake --build . --config Release -j -v
# build all binary
cmake --build build -j -v
```
#### Nvidia GPU
@ -248,16 +246,15 @@ export CPLUS_INCLUDE_DIR=/path/to/oneMKL/buildWithCublas/include:$CPLUS_INCLUDE_
export CPLUS_INCLUDE_DIR=/path/to/oneMKL/include:$CPLUS_INCLUDE_DIR
# Build LLAMA with Nvidia BLAS acceleration through SYCL
mkdir -p build && cd build
# Option 1: Use FP32 (recommended for better performance in most cases)
cmake .. -DLLAMA_SYCL=ON -DLLAMA_SYCL_TARGET=NVIDIA -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx
cmake . -B build -DLLAMA_SYCL=ON -DLLAMA_SYCL_TARGET=NVIDIA -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx
# Option 2: Use FP16
cmake .. -DLLAMA_SYCL=ON -DLLAMA_SYCL_TARGET=NVIDIA -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DLLAMA_SYCL_F16=ON
cmake . -B build -DLLAMA_SYCL=ON -DLLAMA_SYCL_TARGET=NVIDIA -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DLLAMA_SYCL_F16=ON
#build all binary
cmake --build . --config Release -j -v
# build all binary
cmake --build build -j -v
```

73
README.md
View file

@ -321,10 +321,8 @@ In order to build llama.cpp you have three different options.
- Using `CMake`:
```bash
mkdir build
cd build
cmake ..
cmake --build . --config Release
cmake . -B build # Note: add -DCMAKE_BUILD_TYPE=Debug here for debug builds
cmake --build build
```
- Using `Zig` (version 0.11 or later):
@ -438,10 +436,8 @@ Building the program with BLAS support may lead to some performance improvements
- Using `CMake` on Linux:
```bash
mkdir build
cd build
cmake .. -DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS
cmake --build . --config Release
cmake . -B build -DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS
cmake --build build
```
- #### BLIS
@ -461,11 +457,9 @@ Building the program with BLAS support may lead to some performance improvements
- Using manual oneAPI installation:
By default, `LLAMA_BLAS_VENDOR` is set to `Generic`, so if you already sourced intel environment script and assign `-DLLAMA_BLAS=ON` in cmake, the mkl version of Blas will automatically been selected. Otherwise please install oneAPI and follow the below steps:
```bash
mkdir build
cd build
source /opt/intel/oneapi/setvars.sh # You can skip this step if in oneapi-basekit docker image, only required for manual installation
cmake .. -DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=Intel10_64lp -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DLLAMA_NATIVE=ON
cmake --build . --config Release
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
```
- Using oneAPI docker image:
@ -486,10 +480,8 @@ Building the program with BLAS support may lead to some performance improvements
- Using `CMake`:
```bash
mkdir build
cd build
cmake .. -DLLAMA_CUDA=ON
cmake --build . --config Release
cmake . -B build -DLLAMA_CUDA=ON
cmake --build build .
```
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:
@ -563,15 +555,14 @@ Building the program with BLAS support may lead to some performance improvements
```sh
git clone --recurse-submodules https://github.com/KhronosGroup/OpenCL-SDK.git
mkdir OpenCL-SDK/build
cd OpenCL-SDK/build
cmake .. -DBUILD_DOCS=OFF \
cd OpenCL-SDK
cmake . -B build -DBUILD_DOCS=OFF \
-DBUILD_EXAMPLES=OFF \
-DBUILD_TESTING=OFF \
-DOPENCL_SDK_BUILD_SAMPLES=OFF \
-DOPENCL_SDK_TEST_SAMPLES=OFF
cmake --build . --config Release
cmake --install . --prefix /some/path
cmake --build build
cmake --install build --prefix /some/path
```
</details>
@ -593,23 +584,23 @@ Building the program with BLAS support may lead to some performance improvements
```cmd
set OPENCL_SDK_ROOT="C:/OpenCL-SDK-v2023.04.17-Win-x64"
git clone https://github.com/CNugteren/CLBlast.git
mkdir CLBlast\build
cd CLBlast\build
cmake .. -DBUILD_SHARED_LIBS=OFF -DOVERRIDE_MSVC_FLAGS_TO_MT=OFF -DTUNERS=OFF -DOPENCL_ROOT=%OPENCL_SDK_ROOT% -G "Visual Studio 17 2022" -A x64
cmake --build . --config Release
cmake --install . --prefix C:/CLBlast
cd CLBlast
cmake . -B build -DBUILD_SHARED_LIBS=OFF -DOVERRIDE_MSVC_FLAGS_TO_MT=OFF -DTUNERS=OFF -DOPENCL_ROOT=%OPENCL_SDK_ROOT% -G "Visual Studio 17 2022" -A x64
cmake --build build --config Release
cmake --install build --prefix C:/CLBlast
```
(note: `--config Release` at build time is the default and only relevant for Visual Studio builds - or multi-config Ninja builds)
- <details>
<summary>Unix:</summary>
```sh
git clone https://github.com/CNugteren/CLBlast.git
mkdir CLBlast/build
cd CLBlast/build
cmake .. -DBUILD_SHARED_LIBS=OFF -DTUNERS=OFF
cmake --build . --config Release
cmake --install . --prefix /some/path
cd CLBlast
cmake . -B build -DBUILD_SHARED_LIBS=OFF -DTUNERS=OFF
cmake --build build
cmake --install build --prefix /some/path
```
Where `/some/path` is where the built library will be installed (default is `/usr/local`).
@ -623,21 +614,17 @@ Building the program with BLAS support may lead to some performance improvements
```
- CMake (Unix):
```sh
mkdir build
cd build
cmake .. -DLLAMA_CLBLAST=ON -DCLBlast_DIR=/some/path
cmake --build . --config Release
cmake . -B build -DLLAMA_CLBLAST=ON -DCLBlast_DIR=/some/path
cmake --build build
```
- CMake (Windows):
```cmd
set CL_BLAST_CMAKE_PKG="C:/CLBlast/lib/cmake/CLBlast"
git clone https://github.com/ggerganov/llama.cpp
cd llama.cpp
mkdir build
cd build
cmake .. -DBUILD_SHARED_LIBS=OFF -DLLAMA_CLBLAST=ON -DCMAKE_PREFIX_PATH=%CL_BLAST_CMAKE_PKG% -G "Visual Studio 17 2022" -A x64
cmake --build . --config Release
cmake --install . --prefix C:/LlamaCPP
cmake . -B build -DBUILD_SHARED_LIBS=OFF -DLLAMA_CLBLAST=ON -DCMAKE_PREFIX_PATH=%CL_BLAST_CMAKE_PKG% -G "Visual Studio 17 2022" -A x64
cmake --build build --config Release
cmake --install build --prefix C:/LlamaCPP
```
##### Running Llama with CLBlast
@ -693,10 +680,8 @@ Building the program with BLAS support may lead to some performance improvements
Then, build llama.cpp using the cmake command below:
```bash
mkdir -p build
cd build
cmake .. -DLLAMA_VULKAN=1
cmake --build . --config Release
cmake . -B build -DLLAMA_VULKAN=1
cmake --build build
# 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

16
examples/main-cmake-pkg/README.md
View file

@ -17,11 +17,9 @@ In this case, CLBlast was already installed so the CMake package is referenced i
```cmd
git clone https://github.com/ggerganov/llama.cpp
cd llama.cpp
mkdir build
cd build
cmake .. -DBUILD_SHARED_LIBS=OFF -DLLAMA_CLBLAST=ON -DCMAKE_PREFIX_PATH=C:/CLBlast/lib/cmake/CLBlast -G "Visual Studio 17 2022" -A x64
cmake --build . --config Release
cmake --install . --prefix C:/LlamaCPP
cmake . -B build -DBUILD_SHARED_LIBS=OFF -DLLAMA_CLBLAST=ON -DCMAKE_PREFIX_PATH=C:/CLBlast/lib/cmake/CLBlast -G "Visual Studio 17 2022" -A x64
cmake --build build --config Release
cmake --install build --prefix C:/LlamaCPP
```
### Build main-cmake-pkg
@ -29,9 +27,7 @@ cmake --install . --prefix C:/LlamaCPP
```cmd
cd ..\examples\main-cmake-pkg
mkdir build
cd build
cmake .. -DBUILD_SHARED_LIBS=OFF -DCMAKE_PREFIX_PATH="C:/CLBlast/lib/cmake/CLBlast;C:/LlamaCPP/lib/cmake/Llama" -G "Visual Studio 17 2022" -A x64
cmake --build . --config Release
cmake --install . --prefix C:/MyLlamaApp
cmake . -B build -DBUILD_SHARED_LIBS=OFF -DCMAKE_PREFIX_PATH="C:/CLBlast/lib/cmake/CLBlast;C:/LlamaCPP/lib/cmake/Llama" -G "Visual Studio 17 2022" -A x64
cmake --build build --config Release
cmake --install build --prefix C:/MyLlamaApp
```

13
examples/server/README.md
View file

@ -74,15 +74,18 @@ page cache before using this. See https://github.com/ggerganov/llama.cpp/issues/
- Using `make`:
```bash
make
make server
```
- Using `CMake`:
```bash
cmake --build . --config Release
cmake . -B build
cmake --build build -t server
```
Binary is at `./build/bin/server`
## Build with SSL
`server` can also be built with SSL support using OpenSSL 3
@ -99,10 +102,8 @@ page cache before using this. See https://github.com/ggerganov/llama.cpp/issues/
- Using `CMake`:
```bash
mkdir build
cd build
cmake .. -DLLAMA_SERVER_SSL=ON
make server
cmake . -B build -DLLAMA_SERVER_SSL=ON
cmake --build build -t server
```
## Quick Start