A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 17, 05:30 UTC
CI: make sure benchmark script exit on error (#2449)
Simon Mo committed
Jan 16, 2024 at 09:50 UTC
947f0b23ccea1377a5f08900fe71d7de118e5042
1 file changed
+11
.buildkite/run-benchmarks.sh
+11
@@ -7,8 +7,10 @@ cd "$(dirname "${BASH_SOURCE[0]}")/.."
7
8
# run benchmarks and upload the result to buildkite
9
python3 benchmarks/benchmark_latency.py 2>&1 | tee benchmark_latency.txt
10
+bench_latency_exit_code=$?
11
12
python3 benchmarks/benchmark_throughput.py --input-len 256 --output-len 256 2>&1 | tee benchmark_throughput.txt
13
+bench_throughput_exit_code=$?
14
15
# write the results into a markdown file
16
echo "### Latency Benchmarks" >> benchmark_results.md
@@ -22,3 +24,12 @@ sed -n '$p' benchmark_throughput.txt >> benchmark_results.md
24
25
# upload the results to buildkite
26
/workspace/buildkite-agent annotate --style "info" --context "benchmark-results" < benchmark_results.md
27
+
28
+# exit with the exit code of the benchmarks
29
+if [ $bench_latency_exit_code -ne 0 ]; then
30
+ exit $bench_latency_exit_code
31
+fi
32
+
33
+if [ $bench_throughput_exit_code -ne 0 ]; then
34
+ exit $bench_throughput_exit_code
35
+fi