Mirrors/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-12-24 13:28:50 +01:00
Code Issues Packages Projects Releases Wiki Activity

make : add LLAMA_HIP_UMA option (#4587)

Browse Source 
NB: LLAMA_HIP_UMA=1 (or any value) adds MK_CPPFLAG -DGGML_HIP_UMA
This commit is contained in:
Michael Kesper 2023-12-22 09:03:25 +01:00 committed by GitHub
parent f31b984898
commit 28cb35a0ec
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 10 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
Makefile
View File

@ -452,6 +452,9 @@ ifdef LLAMA_HIPBLAS
LLAMA_CUDA_MMV_Y ?= 1
LLAMA_CUDA_KQUANTS_ITER ?= 2
MK_CPPFLAGS += -DGGML_USE_HIPBLAS -DGGML_USE_CUBLAS
ifdef LLAMA_HIP_UMA
MK_CPPFLAGS += -DGGML_HIP_UMA
endif # LLAMA_HIP_UMA
MK_LDFLAGS += -L$(ROCM_PATH)/lib -Wl,-rpath=$(ROCM_PATH)/lib
MK_LDFLAGS += -lhipblas -lamdhip64 -lrocblas
HIPFLAGS += $(addprefix --offload-arch=,$(GPU_TARGETS))

8
README.md
View File

@ -440,7 +440,13 @@ Building the program with BLAS support may lead to some performance improvements
&& cmake --build build -- -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"`.
However, this hurts performance for non-integrated GPUs.
However, this hurts performance for non-integrated GPUs (but enables working with integrated GPUs).
- Using `make` (example for target gfx1030, build with 16 CPU threads):
```bash
make -j16 LLAMA_HIPBLAS=1 LLAMA_HIP_UMA=1 AMDGPU_TARGETS=gxf1030
```
- Using `CMake` for Windows (using x64 Native Tools Command Prompt for VS, and assuming a gfx1100-compatible AMD GPU):
```bash
set PATH=%HIP_PATH%\bin;%PATH%