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 2, 01:00 UTC
[ROCm][CI] Move Model Executor test step from MI250 to MI300 (gfx942) (#44370)
Signed-off-by: JartX <sagformas@epdcenter.es>
JartX committed
Jun 3, 2026 at 19:23 UTC
5b2a2beade03a029a9cae2dd11abe24bb41f39f7
1 file changed
+23
-23
.buildkite/test-amd.yaml
+23
-23
@@ -449,29 +449,6 @@ steps:
449
commands:
450
- pytest -v -s lora --shard-id=$$BUILDKITE_PARALLEL_JOB --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --ignore=lora/test_chatglm3_tp.py --ignore=lora/test_llama_tp.py --ignore=lora/test_qwen3_with_multi_loras.py --ignore=lora/test_olmoe_tp.py --ignore=lora/test_deepseekv2_tp.py --ignore=lora/test_gptoss_tp.py --ignore=lora/test_qwen3moe_tp.py --ignore=lora/test_qwen35_densemodel_lora.py
451
452
-#------------------------------------------------------ mi250 · model_executor -------------------------------------------------------#
453
-
454
-- label: Model Executor # TBD
455
- timeout_in_minutes: 180
456
- mirror_hardwares: [amdexperimental, amdproduction, amdgfx90anightly, amdmi250]
457
- agent_pool: mi250_1
458
- optional: true
459
- torch_nightly: true
460
- working_dir: "/vllm-workspace/tests"
461
- source_file_dependencies:
462
- - vllm/engine/arg_utils.py
463
- - vllm/config/model.py
464
- - vllm/model_executor
465
- - tests/model_executor
466
- - tests/entrypoints/openai/completion/test_tensorizer_entrypoint.py
467
- - vllm/_aiter_ops.py
468
- - vllm/platforms/rocm.py
469
- commands:
470
- - apt-get update && apt-get install -y curl libsodium23
471
- - export VLLM_WORKER_MULTIPROC_METHOD=spawn
472
- - pytest -v -s model_executor -m '(not slow_test)'
473
- - pytest -v -s entrypoints/openai/completion/test_tensorizer_entrypoint.py
474
-
452
#------------------------------------------------------ mi250 · models / basic -------------------------------------------------------#
453
454
- label: Basic Models Test (Other CPU) # TBD
@@ -1739,6 +1716,29 @@ steps:
1716
- pytest -v -s -x lora/test_gptoss_tp.py
1717
- pytest -v -s -x lora/test_qwen35_densemodel_lora.py
1718
1719
+#------------------------------------------------------ mi300 · model_executor -------------------------------------------------------#
1720
+
1721
+- label: Model Executor # TBD
1722
+ timeout_in_minutes: 180
1723
+ mirror_hardwares: [amdexperimental, amdproduction, amdgfx942nightly, amdmi300]
1724
+ agent_pool: mi300_1
1725
+ optional: true
1726
+ torch_nightly: true
1727
+ working_dir: "/vllm-workspace/tests"
1728
+ source_file_dependencies:
1729
+ - vllm/engine/arg_utils.py
1730
+ - vllm/config/model.py
1731
+ - vllm/model_executor
1732
+ - tests/model_executor
1733
+ - tests/entrypoints/openai/completion/test_tensorizer_entrypoint.py
1734
+ - vllm/_aiter_ops.py
1735
+ - vllm/platforms/rocm.py
1736
+ commands:
1737
+ - apt-get update && apt-get install -y curl libsodium23
1738
+ - export VLLM_WORKER_MULTIPROC_METHOD=spawn
1739
+ - pytest -v -s model_executor -m '(not slow_test)'
1740
+ - pytest -v -s entrypoints/openai/completion/test_tensorizer_entrypoint.py
1741
+
1742
#----------------------------------------------------- mi300 · models / language -----------------------------------------------------#
1743
1744
- label: Language Models Test (Extended Pooling) # TBD