25b6402315
* Add Windows GPU UT * temporarily remove ut on arc * retry * retry * retry * fix * retry * retry * fix * retry * retry * retry * retry * retry * retry * retry * retry * retry * retry * retry * retry * retry * fix * retry * retry * retry * retry * retry * retry * merge into single workflow * retry inference test * retry * retrigger * try to fix inference test * retry * retry * retry * retry * retry * retry * retry * retry * retry * retry * retry * check lower_bound * retry * retry * try example test * try fix example test * retry * fix * seperate function into shell script * remove cygpath * try remove all cygpath * retry * retry * Revert "try remove all cygpath" This reverts commit 7ceeff3e48f08429062ecef548c1a3ad3488756f. * Revert "retry" This reverts commit 40ea2457843bff6991b8db24316cde5de1d35418. * Revert "retry" This reverts commit 817d0db3e5aec3bd449d3deaf4fb01d3ecfdc8a3. * enable ut * fix * retrigger * retrigger * update download url * fix * fix * retry * add comment * fix
56 lines
1.8 KiB
Bash
56 lines
1.8 KiB
Bash
#!/bin/bash
|
|
|
|
export ANALYTICS_ZOO_ROOT=${ANALYTICS_ZOO_ROOT}
|
|
export LLM_HOME=${ANALYTICS_ZOO_ROOT}/python/llm/src
|
|
export LLM_INFERENCE_TEST_DIR=${ANALYTICS_ZOO_ROOT}/python/llm/test/inference_gpu
|
|
|
|
if [[ $RUNNER_OS == "Linux" ]]; then
|
|
export USE_XETLA=OFF
|
|
export SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1
|
|
elif [[ $RUNNER_OS == "Windows" ]]; then
|
|
export ANALYTICS_ZOO_ROOT=$(cygpath -m ${ANALYTICS_ZOO_ROOT})
|
|
export LLM_INFERENCE_TEST_DIR=${ANALYTICS_ZOO_ROOT}/python/llm/test/inference_gpu
|
|
export SYCL_CACHE_PERSISTENT=1
|
|
fi
|
|
|
|
export DEVICE='xpu'
|
|
|
|
set -e
|
|
|
|
echo "# Start testing inference"
|
|
start=$(date "+%s")
|
|
|
|
# if [ -z "$THREAD_NUM" ]; then
|
|
# THREAD_NUM=2
|
|
# fi
|
|
# export OMP_NUM_THREADS=$THREAD_NUM
|
|
|
|
# import pytest_check_error function
|
|
source ${ANALYTICS_ZOO_ROOT}/python/llm/test/run-llm-check-function.sh
|
|
|
|
pytest_check_error pytest ${LLM_INFERENCE_TEST_DIR}/test_transformers_api.py -v -s
|
|
pytest_check_error pytest ${LLM_INFERENCE_TEST_DIR}/test_transformers_api_layernorm.py -v -s
|
|
|
|
export BIGDL_LLM_XMX_DISABLED=1
|
|
pytest_check_error pytest ${LLM_INFERENCE_TEST_DIR}/test_transformers_api_final_logits.py -v -s
|
|
pytest_check_error pytest ${LLM_INFERENCE_TEST_DIR}/test_transformers_api_attention.py -v -s
|
|
pytest_check_error pytest ${LLM_INFERENCE_TEST_DIR}/test_transformers_api_mlp.py -v -s
|
|
pytest_check_error pytest ${LLM_INFERENCE_TEST_DIR}/test_transformers_api_RMSNorm.py -v -s
|
|
unset BIGDL_LLM_XMX_DISABLED
|
|
|
|
now=$(date "+%s")
|
|
time=$((now-start))
|
|
|
|
echo "Bigdl-llm gpu inference tests finished"
|
|
echo "Time used:$time seconds"
|
|
|
|
echo "# Start testing layers.fast_rope_embedding"
|
|
start=$(date "+%s")
|
|
|
|
pytest_check_error pytest ${LLM_INFERENCE_TEST_DIR}/test_layer_fast_rope.py -v -s
|
|
|
|
now=$(date "+%s")
|
|
time=$((now-start))
|
|
|
|
echo "Bigdl-llm gpu layers.fast_rope_embedding tests finished"
|
|
echo "Time used:$time seconds"
|