forked from GATEOverflow/cm4mlops
-
Notifications
You must be signed in to change notification settings - Fork 0
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Merge branch 'GATEOverflow:mlperf-inference' into mlperf-inference
- Loading branch information
Showing
17 changed files
with
471 additions
and
48 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -10,6 +10,7 @@ on: | |
| branches: | ||
| - main | ||
| - docs | ||
| - mlperf-inference | ||
|
|
||
| jobs: | ||
|
|
||
|
|
||
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
75 changes: 75 additions & 0 deletions
75
script/app-mlperf-inference-intel/run_gptj_harness_v4_0.sh
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,75 @@ | ||
| #!/bin/bash | ||
| export PATH=${CM_CONDA_BIN_PATH}:$PATH | ||
|
|
||
| export KMP_BLOCKTIME=1 | ||
| export KMP_AFFINITY=granularity=fine,compact,1,0 | ||
| export LD_PRELOAD=${LD_PRELOAD}:${CONDA_PREFIX}/lib/libiomp5.so | ||
| # export LD_PRELOAD=${LD_PRELOAD}:${CONDA_PREFIX}/lib/libtcmalloc.so | ||
| # | ||
|
|
||
| BATCH_SIZE=${CM_MLPERF_LOADGEN_BATCH_SIZE} | ||
|
|
||
| DIR_SCRIPT=$(dirname "${BASH_SOURCE[0]}") | ||
| [ -z $DIR_NS ] && DIR_NS="$DIR_SCRIPT/gpt-j-env/neural-speed" | ||
| [ -z $VALIDATION_DATA_JSON ] && VALIDATION_DATA_JSON="$DIR_SCRIPT/gpt-j-env/cnn_dailymail_validation.json" | ||
| [ -z $CHECKPOINT_DIR ] && CHECKPOINT_DIR="$DIR_SCRIPT/gpt-j-env/finetuned_gptj" | ||
|
|
||
| # num_numa=$(numactl --hardware|grep available|awk -F' ' '{ print $2 }') | ||
|
|
||
| export num_physical_cores=$(lscpu -b -p=Core,Socket | grep -v '^#' | sort -u | wc -l) | ||
| IFS=', ' read -r -a available_nodes_list <<<"$(numactl -s | grep nodebind | sed -E 's/^nodebind: (.+)$/\1/')" | ||
| declare -p available_nodes_list | ||
| num_numa="${#available_nodes_list[@]}" | ||
| declare -p num_numa | ||
|
|
||
| find "$DIR_NS" -name CMakeCache.txt -exec rm {} \; | ||
| CMAKE_ARGS="-DNS_PROFILING=ON" pip install -e "$DIR_NS" | ||
|
|
||
| [ -z $NUM_PROC ] && NUM_PROC=$num_numa | ||
| CPUS_PER_PROC=$((num_physical_cores / num_numa)) | ||
| [ -z $WORKERS_PER_PROC ] && WORKERS_PER_PROC=1 | ||
| [ -z $CPUS_PER_WORKER ] && CPUS_PER_WORKER= # e.g. 8:8:8:8:8:8:8 | ||
| [ -z $BATCH_PROC_ALLOC ] && BATCH_PROC_ALLOC= # e.g. 12:12:12:12:12:12:12 | ||
| [ -z $LOGICAL_CORES_START ] && LOGICAL_CORES_START=-1 # set to -1 to disable / or use $num_physical_cores | ||
| [ -z $CORES_OFFSET ] && CORES_OFFSET=0 | ||
|
|
||
| [ -z $BATCH_SIZE ] && BATCH_SIZE=12 | ||
| [ -z $BEAM_SIZE ] && BEAM_SIZE=4 | ||
|
|
||
| OUTPUT_DIR="${CM_MLPERF_OUTPUT_DIR}" | ||
| MODEL_PATH="${CM_ML_MODEL_FILE_WITH_PATH}" | ||
| cd ${CM_HARNESS_CODE_ROOT} | ||
| export WORKLOAD_DATA=${CM_HARNESS_CODE_ROOT}/data | ||
| export VALIDATION_DATA_JSON=${WORKLOAD_DATA}/validation-data/cnn_dailymail_validation.json | ||
|
|
||
|
|
||
| for i in $(seq 0 $(($NUM_PROC - 1))); do | ||
| [[ ! -e "${MODEL_PATH}${i}" ]] && ln -fs "$(basename $MODEL_PATH)" "${MODEL_PATH}${i}" | ||
| done | ||
|
|
||
| echo "Start time: $(date)" | ||
| cmd="python runner.py --workload-name gptj \ | ||
| --scenario ${CM_MLPERF_LOADGEN_SCENARIO} \ | ||
| --mode ${LOADGEN_MODE} \ | ||
| --num-proc ${NUM_PROC} \ | ||
| --cpus-per-proc ${CPUS_PER_PROC} \ | ||
| --dataset-path ${VALIDATION_DATA_JSON} \ | ||
| --model-path ${MODEL_PATH} \ | ||
| --model-checkpoint ${CHECKPOINT_DIR} \ | ||
| --batch-size ${BATCH_SIZE} \ | ||
| --beam-size ${BEAM_SIZE} \ | ||
| --mlperf-conf ${CM_MLPERF_CONF} \ | ||
| --user-conf ${CM_MLPERF_USER_CONF} \ | ||
| --workers-per-proc ${WORKERS_PER_PROC} \ | ||
| --total-sample-count ${TOTAL_SAMPLE_COUNT} \ | ||
| --output-dir ${OUTPUT_DIR} \ | ||
| --cores-offset ${CORES_OFFSET} \ | ||
| --logical-cores-start \"${LOGICAL_CORES_START}\" \ | ||
| --cpus-per-worker \"${CPUS_PER_WORKER}\" \ | ||
| --batch-proc-alloc \"${BATCH_PROC_ALLOC}\" \ | ||
| 2>&1 | tee ${OUTPUT_DIR}.log" | ||
| echo "$cmd" | ||
| eval "$cmd" | ||
| test $? -eq 0 || exit $? | ||
| echo "End time: $(date)" | ||
|
|
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Oops, something went wrong.