matt/ollama37
1
0
Fork 0
mirror of https://github.com/dogkeeper886/ollama37.git synced 2025-12-11 00:07:07 +00:00
Code Issues Packages Projects Releases Wiki Activity

Support multiple variants for a given llm lib type

Browse Source 
In some cases we may want multiple variants for a given GPU type or CPU.
This adds logic to have an optional Variant which we can use to select
an optimal library, but also allows us to try multiple variants in case
some fail to load.

This can be useful for scenarios such as ROCm v5 vs v6 incompatibility
or potentially CPU features.
This commit is contained in:
Daniel Hiltgen
2024-01-05 12:13:08 -08:00
parent b24e8d17b2
commit 8da7bef05f
16 changed files with 428 additions and 212 deletions
Download Patch File Download Diff File Expand all files Collapse all files

33
llm/generate/gen_linux.sh
View File

@@ -48,23 +48,31 @@ init_vars
git_module_setup
apply_patches
#
# CPU first for the default library
#
CMAKE_DEFS="${COMMON_CMAKE_DEFS} ${CMAKE_DEFS}"
BUILD_DIR="${LLAMACPP_DIR}/build/linux/cpu"
if [ -z "${OLLAMA_SKIP_CPU_GENERATE}" ]; then
#
# CPU first for the default library
#
CMAKE_DEFS="${COMMON_CMAKE_DEFS} ${CMAKE_DEFS}"
BUILD_DIR="${LLAMACPP_DIR}/build/linux/cpu"
build
install
build
install
# Placeholder to keep go embed happy until we start building dynamic CPU lib variants
touch ${BUILD_DIR}/lib/dummy.so
# Placeholder to keep go embed happy until we start building dynamic CPU lib variants
touch ${BUILD_DIR}/lib/dummy.so
else
echo "Skipping CPU generation step as requested"
fi
if [ -d /usr/local/cuda/lib64/ ]; then
echo "CUDA libraries detected - building dynamic CUDA library"
init_vars
CUDA_MAJOR=$(ls /usr/local/cuda/lib64/libcudart.so.* | head -1 | cut -f3 -d. || true)
if [ -n "${CUDA_MAJOR}" ]; then
CUDA_VARIANT=_v${CUDA_MAJOR}
fi
CMAKE_DEFS="-DLLAMA_CUBLAS=on ${COMMON_CMAKE_DEFS} ${CMAKE_DEFS}"
BUILD_DIR="${LLAMACPP_DIR}/build/linux/cuda"
BUILD_DIR="${LLAMACPP_DIR}/build/linux/cuda${CUDA_VARIANT}"
CUDA_LIB_DIR=/usr/local/cuda/lib64
build
install
@@ -96,9 +104,12 @@ fi
if [ -d "${ROCM_PATH}" ]; then
echo "ROCm libraries detected - building dynamic ROCm library"
if [ -f ${ROCM_PATH}/lib/librocm_smi64.so.? ]; then
ROCM_VARIANT=_v$(ls ${ROCM_PATH}/lib/librocm_smi64.so.? | cut -f3 -d. || true)
fi
init_vars
CMAKE_DEFS="${COMMON_CMAKE_DEFS} ${CMAKE_DEFS} -DLLAMA_HIPBLAS=on -DCMAKE_C_COMPILER=$ROCM_PATH/llvm/bin/clang -DCMAKE_CXX_COMPILER=$ROCM_PATH/llvm/bin/clang++ -DAMDGPU_TARGETS=$(amdGPUs) -DGPU_TARGETS=$(amdGPUs)"
BUILD_DIR="${LLAMACPP_DIR}/build/linux/rocm"
BUILD_DIR="${LLAMACPP_DIR}/build/linux/rocm${ROCM_VARIANT}"
build
install
gcc -fPIC -g -shared -o ${BUILD_DIR}/lib/libext_server.so \