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 10, 01:30 UTC
[CI] Fix OOM in Hopper Fusion E2E Tests (H100) (#32489)
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
Lucas Wilkinson committed
Jan 16, 2026 at 14:27 UTC
ca21288080882714d7d10ee8588641225b08756c
1 file changed
+2
-1
.buildkite/test-pipeline.yaml
+2
-1
@@ -1064,7 +1064,8 @@ steps:
1064
- tests/compile/test_fusion_attn.py
1065
commands:
1066
- export VLLM_TEST_CLEAN_GPU_MEMORY=1
1067
- - pytest -v -s tests/compile/test_fusion_attn.py
1067
+ # skip Llama-4 since it does not fit on this device
1068
+ - pytest -v -s tests/compile/test_fusion_attn.py -k 'not Llama-4'
1069
1070
- label: Hopper Fusion Distributed E2E Tests (2xH100) # 70min
1071
timeout_in_minutes: 70