mirror of
https://github.com/ggerganov/llama.cpp.git
synced 2024-10-31 23:28:51 +01:00
de17e3f745
* disable mmap to fix memcpy crash, add missed cmd in guide, fix softmax * refactor to disable mmap for SYCL backend * fix compile error in other os * refactor the solution, use host buf to fix it, instead of disable mmap * keep to support mmap() * use host buff to reduce malloc times * revert to malloc/free solution, for threaad safe
38 lines
1.2 KiB
Bash
Executable File
38 lines
1.2 KiB
Bash
Executable File
#!/bin/bash
|
|
|
|
# MIT license
|
|
# Copyright (C) 2024 Intel Corporation
|
|
# SPDX-License-Identifier: MIT
|
|
|
|
INPUT2="Building a website can be done in 10 simple steps:\nStep 1:"
|
|
source /opt/intel/oneapi/setvars.sh
|
|
|
|
if [ $# -gt 0 ]; then
|
|
GGML_SYCL_DEVICE=$1
|
|
GGML_SYCL_SINGLE_GPU=1
|
|
else
|
|
GGML_SYCL_DEVICE=0
|
|
GGML_SYCL_SINGLE_GPU=0
|
|
fi
|
|
|
|
#export GGML_SYCL_DEBUG=1
|
|
|
|
|
|
#ZES_ENABLE_SYSMAN=1, Support to get free memory of GPU by sycl::aspect::ext_intel_free_memory. Recommended to use when --split-mode = layer.
|
|
|
|
if [ $GGML_SYCL_SINGLE_GPU -eq 1 ]; then
|
|
echo "use $GGML_SYCL_DEVICE as main GPU"
|
|
#use signle GPU only
|
|
ZES_ENABLE_SYSMAN=1 ./build/bin/main -m models/llama-2-7b.Q4_0.gguf -p "${INPUT2}" -n 400 -e -ngl 33 -s 0 -mg $GGML_SYCL_DEVICE -sm none
|
|
else
|
|
#use multiple GPUs with same max compute units
|
|
ZES_ENABLE_SYSMAN=1 ./build/bin/main -m models/llama-2-7b.Q4_0.gguf -p "${INPUT2}" -n 400 -e -ngl 33 -s 0
|
|
fi
|
|
|
|
#use main GPU only
|
|
#ZES_ENABLE_SYSMAN=1 ./build/bin/main -m models/llama-2-7b.Q4_0.gguf -p "${INPUT2}" -n 400 -e -ngl 33 -s 0 -mg $GGML_SYCL_DEVICE -sm none
|
|
|
|
#use multiple GPUs with same max compute units
|
|
#ZES_ENABLE_SYSMAN=1 ./build/bin/main -m models/llama-2-7b.Q4_0.gguf -p "${INPUT2}" -n 400 -e -ngl 33 -s 0
|
|
|