LM/llama.cpp for TurboQuant

Download source

sudo apt install git cmake
sudo apt install libssl-dev openssl
mkdir -p ~/llama.cpp/src

git clone https://github.com/TheTom/llama-cpp-turboquant ~/llama.cpp/src-tbq

List available options

grep -rn "option(GGML_SYCL" CMakeLists.txt ggml/CMakeLists.txt
grep -rn "SYCL" CMakeLists.txt ggml/CMakeLists.txt

Build

sudo apt install libvulkan-dev glslc spirv-headers

# set options
cmake -B build/vulkan -G Ninja \
  -DCMAKE_BUILD_TYPE=Release \
  -DGGML_VULKAN=ON \
  -DGGML_NATIVE=ON \
  -DGGML_VULKAN_CHECK_RESULTS=OFF \
  -DCMAKE_BUILD_TYPE=Release \
  -DCMAKE_INSTALL_RPATH="\$ORIGIN/../lib"

# build
# !!! don't use option --parallel
cmake --build build/vulkan --parallel

# install 
cmake --install build/vulkan --prefix ~/llama.cpp/vulkan

# clean
rm -rf build

# test GPU
cd ~/llama.cpp/vulkan/bin
./llama-cli --list-devices

Got It!

Available devices:
  SYCL0: Intel(R) Arc(TM) Pro B70 Graphics (32656 MiB, 32128 MiB free)