[ROCm][CI] Split MI300 Distributed Compile by graph partition mode (#53602)
Signed-off-by: Aarushi Jain <142941703+aarushjain29@users.noreply.github.com> Signed-off-by: Aarushi Jain <aarushi.jain2@amd.com>
Aarushi Jain committed
Sep 9, 2026 at 17:07 UTC
6983a0883dc831d33ffa2713b955159894a350d7
1 file changed
+22
-3
.buildkite/test-amd.yaml
+22
-3
@@ -659,8 +659,27 @@ steps:
659
commands:
660
- bash standalone_tests/pytorch_nightly_dependency.sh
661
662
-- label: ":amd: (MI300) Distributed Compile" # TBD
663
- timeout_in_minutes: 95
662
+- label: ":amd: (MI300) Distributed Compile (Inductor Partition)" # TBD
663
+ timeout_in_minutes: 65
664
+ mirror_hardwares: [amdexperimental, amdproduction, amdgfx942nightly, amdmi300]
665
+ dind: false
666
+ agent_pool: mi300_2
667
+ num_gpus: 2
668
+ working_dir: "/vllm-workspace/"
669
+ source_file_dependencies:
670
+ - vllm/compilation/
671
+ - vllm/model_executor/layers
672
+ - tests/compile/passes/distributed/
673
+ - tests/compile/fusions_e2e/
674
+ - vllm/_aiter_ops.py
675
+ - vllm/platforms/rocm.py
676
+ commands:
677
+ - export VLLM_TEST_CLEAN_GPU_MEMORY=1
678
+ - VLLM_TEST_CLEAN_GPU_MEMORY=1 pytest -v -s tests/compile/passes/distributed/test_async_tp.py
679
+ - pytest -v -s tests/compile/fusions_e2e/test_tp2_ar_rms.py::test_tp2_ar_rms_fusions -k "inductor_partition"
680
+
681
+- label: ":amd: (MI300) Distributed Compile (Dynamo Partition)" # TBD
682
+ timeout_in_minutes: 65
683
mirror_hardwares: [amdexperimental, amdproduction, amdgfx942nightly, amdmi300]
684
dind: false
685
agent_pool: mi300_2
@@ -676,7 +695,7 @@ steps:
695
commands:
696
- export VLLM_TEST_CLEAN_GPU_MEMORY=1
697
- VLLM_TEST_CLEAN_GPU_MEMORY=1 pytest -v -s tests/compile/passes/distributed/test_async_tp.py
679
- - pytest -v -s tests/compile/fusions_e2e/test_tp2_ar_rms.py::test_tp2_ar_rms_fusions
698
+ - pytest -v -s tests/compile/fusions_e2e/test_tp2_ar_rms.py::test_tp2_ar_rms_fusions -k "dynamo_partition"
699
700
- label: ":amd: (MI300) Distributed Compile + RPC" # TBD
701
timeout_in_minutes: 45