mirror of
https://github.com/ikawrakow/ik_llama.cpp.git
synced 2026-01-26 17:20:01 +00:00
0ceeb11721
* Merging mainline - WIP * Merging mainline - WIP AVX2 and CUDA appear to work. CUDA performance seems slightly (~1-2%) lower as it is so often the case with llama.cpp/ggml after some "improvements" have been made. * Merging mainline - fix Metal * Remove check --------- Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
24 lines
582 B
Bash
Executable File
24 lines
582 B
Bash
Executable File
|
|
# MIT license
|
|
# Copyright (C) 2024 Intel Corporation
|
|
# SPDX-License-Identifier: MIT
|
|
|
|
mkdir -p build
|
|
cd build
|
|
source /opt/intel/oneapi/setvars.sh
|
|
|
|
#for FP16
|
|
#cmake .. -DGGML_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DGGML_SYCL_F16=ON # faster for long-prompt inference
|
|
|
|
#for FP32
|
|
cmake .. -DGGML_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx
|
|
|
|
#build example/main
|
|
#cmake --build . --config Release --target main
|
|
|
|
#build example/llama-bench
|
|
#cmake --build . --config Release --target llama-bench
|
|
|
|
#build all binary
|
|
cmake --build . --config Release -j -v
|