[ROCm][CI] Remove duplicate MI355 DPX jobs (#59256)
Signed-off-by: aarushjain29 <aarushi.jain2@amd.com> Co-authored-by: Andreas Karatzas <akaratza@amd.com>
Aarushi Jain committed
Sep 30, 2026 at 12:51 UTC
ed3f6d1a56272a966bd9e514cb262c0ce3ee6822
1 file changed
+3
-381
.buildkite/test-amd.yaml
+3
-381
@@ -3177,6 +3177,7 @@ steps:
3177
dind: false
3178
agent_pool: mi355_2
3179
num_gpus: 2
3180
+ optional: true
3181
working_dir: "/vllm-workspace/"
3182
source_file_dependencies:
3183
- benchmarks/attention_benchmarks/
@@ -3921,78 +3922,6 @@ steps:
3922
- bash ../tools/install_torchcodec_rocm.sh || exit 1
3923
- pytest -s entrypoints/openai/correctness/
3924
3924
-- label: ":amd: (MI355 DPX) Entrypoints Integration (API Server) %N (MI355 suite)"
3925
- timeout_in_minutes: 65
3926
- parallelism: 4
3927
- mirror_hardwares: [amdexperimental, amdproduction, amdgfx950nightly, amdmi355]
3928
- dind: false
3929
- agent_pool: mi355_dpx
3930
- optional: true
3931
- fast_check: true
3932
- working_dir: "/vllm-workspace/tests"
3933
- source_file_dependencies:
3934
- - vllm/
3935
- - tests/entrypoints/serve
3936
- - tests/entrypoints/scale_out
3937
- commands:
3938
- - export VLLM_WORKER_MULTIPROC_METHOD=spawn
3939
- - pytest -v -s entrypoints/serve --ignore=entrypoints/serve/dev/rpc --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --shard-id=$$BUILDKITE_PARALLEL_JOB
3940
- - if [ "$$BUILDKITE_PARALLEL_JOB" = "1" ]; then PYTHONPATH=/vllm-workspace pytest -v -s entrypoints/serve/dev/rpc; fi
3941
- - pytest -v -s entrypoints/scale_out --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --shard-id=$$BUILDKITE_PARALLEL_JOB
3942
-
3943
-- label: ":amd: (MI355 DPX) Entrypoints Integration (OpenAI API completion) (MI355 suite)" # TBD
3944
- timeout_in_minutes: 65
3945
- mirror_hardwares: [amdexperimental, amdproduction, amdgfx950nightly, amdmi355]
3946
- dind: false
3947
- agent_pool: mi355_dpx
3948
- fast_check: true
3949
- optional: true
3950
- working_dir: "/vllm-workspace/tests"
3951
- source_file_dependencies:
3952
- - vllm/
3953
- - tests/entrypoints/openai
3954
- commands:
3955
- - export VLLM_WORKER_MULTIPROC_METHOD=spawn
3956
- - pytest -v -s entrypoints/openai/completion --ignore=entrypoints/openai/completion/test_tensorizer_entrypoint.py
3957
- - pytest -v -s entrypoints/openai --ignore=entrypoints/openai/completion --ignore=entrypoints/openai/chat_completion --ignore=entrypoints/openai/responses --ignore=entrypoints/openai/correctness
3958
-
3959
-- label: ":amd: (MI355 DPX) Entrypoints Integration (OpenAI API chat_completion) (MI355 suite)" # TBD
3960
- timeout_in_minutes: 70
3961
- mirror_hardwares: [amdexperimental, amdproduction, amdgfx950nightly, amdmi355]
3962
- dind: false
3963
- agent_pool: mi355_dpx
3964
- fast_check: true
3965
- optional: true
3966
- working_dir: "/vllm-workspace/tests"
3967
- source_file_dependencies:
3968
- - vllm/
3969
- - tests/entrypoints/openai
3970
- commands:
3971
- - export VLLM_WORKER_MULTIPROC_METHOD=spawn
3972
- - pytest -v -s entrypoints/openai/chat_completion
3973
-
3974
-- label: ":amd: (MI355 DPX) Entrypoints Integration (API Server Generate) (MI355 suite)" # TBD
3975
- timeout_in_minutes: 65
3976
- mirror_hardwares: [amdexperimental, amdproduction, amdgfx950nightly, amdmi355]
3977
- dind: false
3978
- agent_pool: mi355_dpx
3979
- fast_check: true
3980
- optional: true
3981
- working_dir: "/vllm-workspace/tests"
3982
- source_file_dependencies:
3983
- - vllm/
3984
- - tests/tool_use
3985
- - tests/entrypoints/tool_parsers
3986
- - tests/entrypoints/anthropic
3987
- - tests/entrypoints/generate
3988
- - tests/entrypoints/cohere
3989
- commands:
3990
- - pytest -v -s tool_use
3991
- - pytest -v -s entrypoints/tool_parsers
3992
- - pytest -v -s entrypoints/generate
3993
- - pytest -v -s entrypoints/anthropic
3994
- - pytest -v -s entrypoints/cohere
3995
-
3925
- label: ":amd: (MI355 DPX) Entrypoints Integration (Speech to Text)" # TBD
3926
timeout_in_minutes: 60
3927
mirror_hardwares: [amdexperimental, amdproduction, amdgfx950nightly, amdmi355]
@@ -4435,6 +4364,7 @@ steps:
4364
dind: false
4365
agent_pool: mi355_2
4366
num_gpus: 2
4367
+ optional: true
4368
working_dir: "/vllm-workspace"
4369
source_file_dependencies:
4370
- vllm/model_executor/models/
@@ -4898,7 +4828,7 @@ steps:
4828
commands:
4829
- pytest -v -s kernels/attention --ignore=kernels/attention/test_triton_unified_attention_diffkv.py --ignore=kernels/attention/test_rocm_aiter_mla_decode.py --ignore=kernels/attention/test_rocm_aiter_mla_decode_metadata.py --ignore=kernels/attention/test_rocm_aiter_mla_fp8_support.py --ignore=kernels/attention/test_rocm_aiter_mla_op_registration.py --shard-id=$$BUILDKITE_PARALLEL_JOB --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT
4830
4901
-- label: ":amd: (MI355 DPX) MoE Kernels Shard %N (MI355 suite)" # TBD
4831
+- label: ":amd: (MI355 DPX) MoE Kernels Shard %N" # TBD
4832
timeout_in_minutes: 55
4833
mirror_hardwares: [amdexperimental, amdproduction, amdgfx950nightly, amdmi355]
4834
dind: false
@@ -5313,20 +5243,6 @@ steps:
5243
commands:
5244
- pytest -v -s models/language/generation_ppl_test
5245
5316
-- label: ":amd: (MI355 DPX) Language Models (Standard) (MI355 suite)" # TBD
5317
- timeout_in_minutes: 60
5318
- mirror_hardwares: [amdexperimental, amdproduction, amdgfx950nightly, amdmi355]
5319
- dind: false
5320
- agent_pool: mi355_dpx
5321
- working_dir: "/vllm-workspace/tests"
5322
- source_file_dependencies:
5323
- - vllm/
5324
- - tests/models/language
5325
- optional: true
5326
- commands:
5327
- - pip freeze | grep -E 'torch'
5328
- - pytest -v -s models/language -m 'core_model and (not slow_test)'
5329
-
5246
#---------------------------------------------------- mi355 · models / multimodal ----------------------------------------------------#
5247
5248
- label: ":amd: (MI355 DPX) Multimodal Models (Standard) 1: qwen2" # TBD
@@ -5399,20 +5315,6 @@ steps:
5315
commands:
5316
- pytest -v -s models/multimodal/pooling -m 'not core_model'
5317
5402
-- label: ":amd: (MI355 DPX) Multimodal Models (Standard) 1: qwen2 (MI355 suite)" # TBD
5403
- timeout_in_minutes: 65
5404
- mirror_hardwares: [amdexperimental, amdproduction, amdgfx950nightly, amdmi355]
5405
- dind: false
5406
- agent_pool: mi355_dpx
5407
- optional: true
5408
- working_dir: "/vllm-workspace/tests"
5409
- source_file_dependencies:
5410
- - vllm/
5411
- - tests/models/multimodal
5412
- commands:
5413
- - pytest -v -s models/multimodal/generation/test_common.py -m core_model -k "qwen2"
5414
- - pytest -v -s models/multimodal/generation/test_ultravox.py -m core_model
5415
-
5318
- label: ":amd: (MI355 DPX) Multimodal Models (Standard) 4: other + whisper" # TBD
5319
timeout_in_minutes: 70
5320
mirror_hardwares: [amdexperimental, amdproduction, amdgfx950nightly, amdmi355]
@@ -5933,224 +5835,6 @@ steps:
5835
- pytest -v -s v1/attention --ignore=v1/attention/test_mla_backends.py --shard-id=$$BUILDKITE_PARALLEL_JOB --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT
5836
- VLLM_ROCM_USE_AITER=1 pytest -v -s v1/attention/test_mla_backends.py --shard-id=$$BUILDKITE_PARALLEL_JOB --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT
5837
5936
-- label: ":amd: (MI355 DPX) V1 Core (MI355 suite)"
5937
- timeout_in_minutes: 60
5938
- mirror_hardwares: [amdexperimental, amdproduction, amdgfx950nightly, amdmi355]
5939
- dind: false
5940
- agent_pool: mi355_dpx
5941
- optional: true
5942
- working_dir: "/vllm-workspace/tests"
5943
- source_file_dependencies:
5944
- - vllm/
5945
- - tests/v1/core
5946
- - tests/v1/executor
5947
- - tests/v1/kv_offload
5948
- - tests/v1/worker
5949
- - tests/v1/cudagraph
5950
- - tests/v1/kv_connector/unit
5951
- - tests/v1/metrics
5952
- - tests/entrypoints/openai/correctness/test_lmeval.py
5953
- - vllm/config/
5954
- - vllm/distributed/
5955
- - vllm/engine/
5956
- - vllm/entrypoints/pooling/
5957
- - vllm/inputs/
5958
- - vllm/lora/
5959
- - vllm/model_executor/
5960
- - vllm/multimodal/
5961
- - vllm/outputs.py
5962
- - vllm/platforms/
5963
- - vllm/pooling_params.py
5964
- - vllm/profiler/
5965
- - vllm/sampling_params.py
5966
- - vllm/tokenizers/
5967
- - vllm/transformers_utils/
5968
- - vllm/utils/
5969
- - vllm/v1/
5970
- - tests/v1/simple_kv_offload
5971
- - tests/v1/streaming_input
5972
- - tests/v1/ec_connector/unit
5973
- commands:
5974
- - export VLLM_WORKER_MULTIPROC_METHOD=spawn
5975
- - pytest -v -s -m 'not cpu_test' v1/core
5976
-
5977
-- label: ":amd: (MI355 DPX) V1 Executor + Worker (MI355 suite)"
5978
- timeout_in_minutes: 60
5979
- mirror_hardwares: [amdexperimental, amdproduction, amdgfx950nightly, amdmi355]
5980
- dind: false
5981
- agent_pool: mi355_dpx
5982
- optional: true
5983
- working_dir: /vllm-workspace/tests
5984
- source_file_dependencies:
5985
- - vllm/
5986
- - tests/v1/core
5987
- - tests/v1/executor
5988
- - tests/v1/kv_offload
5989
- - tests/v1/worker
5990
- - tests/v1/cudagraph
5991
- - tests/v1/kv_connector/unit
5992
- - tests/v1/metrics
5993
- - tests/entrypoints/openai/correctness/test_lmeval.py
5994
- - vllm/config/
5995
- - vllm/distributed/
5996
- - vllm/engine/
5997
- - vllm/entrypoints/pooling/
5998
- - vllm/inputs/
5999
- - vllm/lora/
6000
- - vllm/model_executor/
6001
- - vllm/multimodal/
6002
- - vllm/outputs.py
6003
- - vllm/platforms/
6004
- - vllm/pooling_params.py
6005
- - vllm/profiler/
6006
- - vllm/sampling_params.py
6007
- - vllm/tokenizers/
6008
- - vllm/transformers_utils/
6009
- - vllm/utils/
6010
- - vllm/v1/
6011
- - tests/v1/simple_kv_offload
6012
- - tests/v1/streaming_input
6013
- - tests/v1/ec_connector/unit
6014
- commands:
6015
- - export VLLM_WORKER_MULTIPROC_METHOD=spawn
6016
- - pytest -v -s v1/executor
6017
- - pytest -v -s v1/worker
6018
- - pytest -v -s v1/streaming_input
6019
-
6020
-- label: ":amd: (MI355 DPX) V1 KV Offload (MI355 suite)"
6021
- timeout_in_minutes: 60
6022
- mirror_hardwares: [amdexperimental, amdproduction, amdgfx950nightly, amdmi355]
6023
- dind: false
6024
- agent_pool: mi355_dpx
6025
- optional: true
6026
- working_dir: /vllm-workspace/tests
6027
- source_file_dependencies:
6028
- - vllm/
6029
- - tests/v1/core
6030
- - tests/v1/executor
6031
- - tests/v1/kv_offload
6032
- - tests/v1/worker
6033
- - tests/v1/cudagraph
6034
- - tests/v1/kv_connector/unit
6035
- - tests/v1/metrics
6036
- - tests/entrypoints/openai/correctness/test_lmeval.py
6037
- - vllm/config/
6038
- - vllm/distributed/
6039
- - vllm/engine/
6040
- - vllm/entrypoints/pooling/
6041
- - vllm/inputs/
6042
- - vllm/lora/
6043
- - vllm/model_executor/
6044
- - vllm/multimodal/
6045
- - vllm/outputs.py
6046
- - vllm/platforms/
6047
- - vllm/pooling_params.py
6048
- - vllm/profiler/
6049
- - vllm/sampling_params.py
6050
- - vllm/tokenizers/
6051
- - vllm/transformers_utils/
6052
- - vllm/utils/
6053
- - vllm/v1/
6054
- - tests/v1/simple_kv_offload
6055
- - tests/v1/streaming_input
6056
- - tests/v1/ec_connector/unit
6057
- commands:
6058
- - export VLLM_WORKER_MULTIPROC_METHOD=spawn
6059
- - bash /vllm-workspace/.buildkite/scripts/install-kv-offload.sh
6060
- - pytest -v -s v1/kv_offload
6061
- - pytest -v -s v1/simple_kv_offload
6062
-
6063
-- label: ":amd: (MI355 DPX) V1 KV Connectors Shard %N (MI355 suite)"
6064
- timeout_in_minutes: 60
6065
- parallelism: 4
6066
- mirror_hardwares: [amdexperimental, amdproduction, amdgfx950nightly, amdmi355]
6067
- dind: false
6068
- agent_pool: mi355_dpx
6069
- optional: true
6070
- working_dir: /vllm-workspace/tests
6071
- source_file_dependencies:
6072
- - vllm/
6073
- - tests/v1/core
6074
- - tests/v1/executor
6075
- - tests/v1/kv_offload
6076
- - tests/v1/worker
6077
- - tests/v1/cudagraph
6078
- - tests/v1/kv_connector/unit
6079
- - tests/v1/metrics
6080
- - tests/entrypoints/openai/correctness/test_lmeval.py
6081
- - vllm/config/
6082
- - vllm/distributed/
6083
- - vllm/engine/
6084
- - vllm/entrypoints/pooling/
6085
- - vllm/inputs/
6086
- - vllm/lora/
6087
- - vllm/model_executor/
6088
- - vllm/multimodal/
6089
- - vllm/outputs.py
6090
- - vllm/platforms/
6091
- - vllm/pooling_params.py
6092
- - vllm/profiler/
6093
- - vllm/sampling_params.py
6094
- - vllm/tokenizers/
6095
- - vllm/transformers_utils/
6096
- - vllm/utils/
6097
- - vllm/v1/
6098
- - tests/v1/simple_kv_offload
6099
- - tests/v1/streaming_input
6100
- - tests/v1/ec_connector/unit
6101
- commands:
6102
- - bash /vllm-workspace/.buildkite/scripts/install-kv-connectors.sh
6103
- - export VLLM_WORKER_MULTIPROC_METHOD=spawn
6104
- - pytest -v -s -m 'not cpu_test' v1/kv_connector/unit --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --shard-id=$$BUILDKITE_PARALLEL_JOB
6105
- - pytest -v -s -m 'not cpu_test' v1/ec_connector/unit --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --shard-id=$$BUILDKITE_PARALLEL_JOB
6106
-
6107
-- label: ":amd: (MI355 DPX) V1 Metrics + LM Eval (MI355 suite)"
6108
- timeout_in_minutes: 65
6109
- mirror_hardwares: [amdexperimental, amdproduction, amdgfx950nightly, amdmi355]
6110
- dind: false
6111
- agent_pool: mi355_dpx
6112
- optional: true
6113
- working_dir: /vllm-workspace/tests
6114
- source_file_dependencies:
6115
- - vllm/
6116
- - tests/v1/core
6117
- - tests/v1/executor
6118
- - tests/v1/kv_offload
6119
- - tests/v1/worker
6120
- - tests/v1/cudagraph
6121
- - tests/v1/kv_connector/unit
6122
- - tests/v1/metrics
6123
- - tests/entrypoints/openai/correctness/test_lmeval.py
6124
- - vllm/config/
6125
- - vllm/distributed/
6126
- - vllm/engine/
6127
- - vllm/entrypoints/pooling/
6128
- - vllm/inputs/
6129
- - vllm/lora/
6130
- - vllm/model_executor/
6131
- - vllm/multimodal/
6132
- - vllm/outputs.py
6133
- - vllm/platforms/
6134
- - vllm/pooling_params.py
6135
- - vllm/profiler/
6136
- - vllm/sampling_params.py
6137
- - vllm/tokenizers/
6138
- - vllm/transformers_utils/
6139
- - vllm/utils/
6140
- - vllm/v1/
6141
- - tests/v1/simple_kv_offload
6142
- - tests/v1/streaming_input
6143
- - tests/v1/ec_connector/unit
6144
- commands:
6145
- - export GIT_CONFIG_COUNT=1
6146
- - export GIT_CONFIG_KEY_0=http.version
6147
- - export GIT_CONFIG_VALUE_0=HTTP/1.1
6148
- - export GIT_TERMINAL_PROMPT=0
6149
- - export VLLM_WORKER_MULTIPROC_METHOD=spawn
6150
- - pytest -v -s -m 'not cpu_test' v1/metrics
6151
- - pip install -U git+https://github.com/vllm-project/lm-evaluation-harness.git@streaming-api
6152
- - pytest -v -s entrypoints/openai/correctness/test_lmeval.py::test_lm_eval_accuracy_v1_engine
6153
-
5838
- label: ":amd: (MI355 DPX) CUDAGraph"
5839
timeout_in_minutes: 55
5840
mirror_hardwares: [amdexperimental, amdproduction, amdgfx950nightly, amdmi355]
@@ -6181,68 +5865,6 @@ steps:
5865
- pytest -v -s v1/cudagraph/test_encoder_cudagraph.py
5866
5867
6184
-- label: ":amd: (MI355 DPX) V1 Sample (MI355 suite)"
6185
- timeout_in_minutes: 70
6186
- mirror_hardwares: [amdexperimental, amdproduction, amdgfx950nightly, amdmi355]
6187
- dind: false
6188
- agent_pool: mi355_dpx
6189
- optional: true
6190
- working_dir: "/vllm-workspace/tests"
6191
- source_file_dependencies:
6192
- - vllm/
6193
- - tests/v1/sample
6194
- - tests/v1/logits_processors
6195
- - tests/v1/test_oracle.py
6196
- - tests/v1/test_request.py
6197
- - tests/v1/test_outputs.py
6198
- - vllm/config/
6199
- - vllm/distributed/
6200
- - vllm/engine/
6201
- - vllm/inputs/
6202
- - vllm/logger.py
6203
- - vllm/model_executor/
6204
- - vllm/platforms/
6205
- - vllm/sampling_params.py
6206
- - vllm/transformers_utils/
6207
- - vllm/utils/
6208
- - vllm/v1/
6209
- commands:
6210
- - export VLLM_WORKER_MULTIPROC_METHOD=spawn
6211
- - pytest -v -s v1/sample
6212
-
6213
-- label: ":amd: (MI355 DPX) V1 Logits + Oracle (MI355 suite)"
6214
- timeout_in_minutes: 70
6215
- mirror_hardwares: [amdexperimental, amdproduction, amdgfx950nightly, amdmi355]
6216
- dind: false
6217
- agent_pool: mi355_dpx
6218
- optional: true
6219
- working_dir: /vllm-workspace/tests
6220
- source_file_dependencies:
6221
- - vllm/
6222
- - tests/v1/sample
6223
- - tests/v1/logits_processors
6224
- - tests/v1/test_oracle.py
6225
- - tests/v1/test_request.py
6226
- - tests/v1/test_outputs.py
6227
- - vllm/config/
6228
- - vllm/distributed/
6229
- - vllm/engine/
6230
- - vllm/inputs/
6231
- - vllm/logger.py
6232
- - vllm/model_executor/
6233
- - vllm/platforms/
6234
- - vllm/sampling_params.py
6235
- - vllm/transformers_utils/
6236
- - vllm/utils/
6237
- - vllm/v1/
6238
- commands:
6239
- - export VLLM_WORKER_MULTIPROC_METHOD=spawn
6240
- - pytest -v -s v1/logits_processors
6241
- - pytest -v -s v1/test_oracle.py
6242
- - pytest -v -s v1/test_request.py
6243
- - pytest -v -s v1/test_outputs.py
6244
-
6245
-
5868
- label: ":amd: (MI355 DPX) V1 Spec Decode" # TBD
5869
timeout_in_minutes: 50
5870
mirror_hardwares: [amdexperimental, amdproduction, amdgfx950nightly, amdmi355]