jeans/ollama
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Adapted rocm support to cgo based llama.cpp

Browse Source
This commit is contained in:
Daniel Hiltgen
2023-11-29 11:00:37 -08:00
parent f8ef4439e9
commit 35934b2e05
37 changed files with 1688 additions and 658 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
llm/llama.cpp/gen_common.sh
View File

@@ -1,10 +1,11 @@
# common logic accross linux and darwin
init_vars() {
LLAMACPP_DIR=gguf
PATCHES="0001-Expose-callable-API-for-server.patch"
CMAKE_DEFS="-DLLAMA_ACCELERATE=on"
# TODO - LLAMA_K_QUANTS is stale and needs to be mapped to newer cmake settings
CMAKE_TARGETS="--target ggml --target ggml_static --target llama --target build_info --target common --target ext_server"
CMAKE_TARGETS="--target ggml --target ggml_static --target llama --target build_info --target common --target ext_server --target llava_static"
if echo "${CGO_CFLAGS}" | grep -- '-g' > /dev/null ; then
CMAKE_DEFS="-DCMAKE_BUILD_TYPE=RelWithDebInfo -DCMAKE_VERBOSE_MAKEFILE=on -DLLAMA_GPROF=on ${CMAKE_DEFS}"
else
@@ -29,6 +30,6 @@ apply_patches() {
}
build() {
cmake -S gguf -B ${BUILD_DIR} ${CMAKE_DEFS}
cmake --build ${BUILD_DIR} ${CMAKE_TARGETS} -j8
cmake -S ${LLAMACPP_DIR} -B ${BUILD_DIR} ${CMAKE_DEFS}
cmake --build ${BUILD_DIR} ${CMAKE_TARGETS} -j8
}