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 16, 23:15 UTC
[ci] Mount buildkite agent on Docker container to upload benchmark results (#5330)
Signed-off-by: kevin <kevin@anyscale.com>
Kevin H. Luu committed
Jun 10, 2024 at 09:22 UTC
c5602f0baa8fa42df11853a0d422bc140cf04c9a
2 files changed
+6
-3
.buildkite/run-benchmarks.sh
+3
-3
@@ -54,12 +54,12 @@ tail -n 24 benchmark_serving.txt >> benchmark_results.md # last 24 lines
54
echo '```' >> benchmark_results.md
55
56
# if the agent binary is not found, skip uploading the results, exit 0
57
-if [ ! -f /workspace/buildkite-agent ]; then
57
+if [ ! -f buildkite-agent ]; then
58
exit 0
59
fi
60
61
# upload the results to buildkite
62
-/workspace/buildkite-agent annotate --style "info" --context "benchmark-results" < benchmark_results.md
62
+buildkite-agent annotate --style "info" --context "benchmark-results" < benchmark_results.md
63
64
# exit with the exit code of the benchmarks
65
if [ $bench_latency_exit_code -ne 0 ]; then
@@ -75,4 +75,4 @@ if [ $bench_serving_exit_code -ne 0 ]; then
75
fi
76
77
rm ShareGPT_V3_unfiltered_cleaned_split.json
78
-/workspace/buildkite-agent artifact upload "*.json"
78
+buildkite-agent artifact upload "*.json"
.buildkite/test-template-aws.j2
+3
@@ -49,6 +49,9 @@ steps:
49
{% if not step.no_gpu %}
50
gpus: all
51
{% endif %}
52
+ {% if step.label == "Benchmarks" %}
53
+ mount-buildkite-agent: true
54
+ {% endif %}
55
command: ["bash", "-c", "cd {{ (step.working_dir or default_working_dir) | safe }} && {{ step.command or (step.commands | join(' && ')) | safe }}"]
56
environment:
57
- VLLM_USAGE_SOURCE=ci-test