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, 03:15 UTC
[perf-benchmark] Fix dependency for steps in benchmark pipeline (#11710)
Kevin H. Luu committed
Jan 3, 2025 at 13:38 UTC
fd3a62a122fcbc9331d000b325e72687629ef1bd
1 file changed
+4
-3
.buildkite/nightly-benchmarks/benchmark-pipeline.yaml
+4
-3
@@ -1,5 +1,6 @@
1
steps:
2
- label: "Wait for container to be ready"
3
+ key: wait-for-container-image
4
agents:
5
queue: A100
6
plugins:
@@ -10,12 +11,11 @@ steps:
11
command:
12
- sh .buildkite/nightly-benchmarks/scripts/wait-for-image.sh
13
13
- - wait
14
-
14
- label: "A100"
15
# skip: "use this flag to conditionally skip the benchmark step, useful for PR testing"
16
agents:
17
queue: A100
18
+ depends_on: wait-for-container-image
19
plugins:
20
- kubernetes:
21
podSpec:
@@ -49,6 +49,7 @@ steps:
49
# skip: "use this flag to conditionally skip the benchmark step, useful for PR testing"
50
agents:
51
queue: H200
52
+ depends_on: wait-for-container-image
53
plugins:
54
- docker#v5.12.0:
55
image: public.ecr.aws/q9t5s3a7/vllm-ci-postmerge-repo:$BUILDKITE_COMMIT
@@ -73,7 +74,7 @@ steps:
74
# skip: "use this flag to conditionally skip the benchmark step, useful for PR testing"
75
agents:
76
queue: H100
76
- depends_on: ~
77
+ depends_on: wait-for-container-image
78
plugins:
79
- docker#v5.12.0:
80
image: public.ecr.aws/q9t5s3a7/vllm-ci-postmerge-repo:$BUILDKITE_COMMIT