iHore/vLLM
8
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

[ROCm][CI] Purging away redundant test group definitions (#46418)

Browse Source 
Signed-off-by: Andreas Karatzas <akaratza@amd.com>
This commit is contained in:
Andreas Karatzas
2026-06-22 15:42:47 -07:00
committed by GitHub
parent e2fe837572
commit 70ef4d3009
1 changed files with 131 additions and 283 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.buildkite/test-amd.yaml
+131 -283
View File
@@ -161,24 +161,6 @@ steps:
commands:
- "find compile/ -maxdepth 1 -name 'test_*.py' -print0 | xargs -0 -n1 -I{} pytest -s -v '{}'"
- label: PyTorch Fullgraph # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx90anightly, amdmi250]
agent_pool: mi250_1
optional: true
torch_nightly: true
working_dir: "/vllm-workspace/tests"
source_file_dependencies:
- vllm/compilation/
- vllm/model_executor/
- vllm/v1/attention/
- vllm/config/compilation.py
- csrc/
- tests/compile
- vllm/platforms/rocm.py
commands:
- pytest -v -s compile/fullgraph/test_full_graph.py -k 'not test_fp8_kv_scale_compile'
- label: PyTorch Fullgraph Smoke Test # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx90anightly, amdmi250]
@@ -263,37 +245,6 @@ steps:
- CUDA_VISIBLE_DEVICES=0,1 pytest -v -s v1/shutdown
- pytest -v -s v1/worker/test_worker_memory_snapshot.py
- label: Elastic EP Scaling Test # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx90anightly, amdmi250]
agent_pool: mi250_4
num_gpus: 4
working_dir: "/vllm-workspace/tests"
source_file_dependencies:
- vllm/distributed/
- vllm/engine/
- vllm/executor/
- vllm/compilation/
- tests/distributed/
- vllm/platforms/rocm.py
commands:
- pytest -v -s distributed/test_elastic_ep.py
- label: EPLB Execution # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx90anightly, amdmi250]
agent_pool: mi250_4
num_gpus: 4
working_dir: "/vllm-workspace/tests"
source_file_dependencies:
- vllm/distributed/eplb
- tests/distributed/test_eplb_execute.py
- tests/distributed/test_eplb_spec_decode.py
- vllm/platforms/rocm.py
commands:
- pytest -v -s distributed/test_eplb_execute.py
- pytest -v -s distributed/test_eplb_spec_decode.py
- label: Pipeline + Context Parallelism (4 GPUs) # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx90anightly, amdmi250]
@@ -315,78 +266,8 @@ steps:
- pytest -v -s distributed/test_pp_cudagraph.py
- pytest -v -s distributed/test_pipeline_parallel.py
#----------------------------------------------------------- mi250 · evals -----------------------------------------------------------#
- label: Multi-Modal Accuracy Eval (Small Models) # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx90anightly, amdmi250]
agent_pool: mi250_1
optional: true
working_dir: "/vllm-workspace/.buildkite/lm-eval-harness"
source_file_dependencies:
- vllm/multimodal/
- vllm/inputs/
- vllm/v1/core/
- vllm/platforms/rocm.py
- vllm/model_executor/model_loader/
commands:
- pytest -s -v test_lm_eval_correctness.py --config-list-file=configs/models-mm-small.txt --tp-size=1
#--------------------------------------------------------- mi250 · examples ----------------------------------------------------------#
- label: Examples # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx90anightly, amdmi250]
agent_pool: mi250_1
optional: true
working_dir: "/vllm-workspace/examples"
source_file_dependencies:
- vllm/entrypoints
- vllm/multimodal
- examples/
- vllm/platforms/rocm.py
commands:
- pip install tensorizer
# Basic
- python3 basic/offline_inference/chat.py --attention-backend TRITON_ATTN
- python3 basic/offline_inference/generate.py --model facebook/opt-125m
- python3 basic/offline_inference/generate.py --model meta-llama/Llama-2-13b-chat-hf --cpu-offload-gb 10
- python3 basic/offline_inference/classify.py
- python3 basic/offline_inference/embed.py
- python3 basic/offline_inference/score.py
# Multi-modal models
- python3 generate/multimodal/audio_language_offline.py --seed 0
- python3 generate/multimodal/vision_language_offline.py --seed 0
- python3 generate/multimodal/vision_language_multi_image_offline.py --seed 0
- python3 generate/multimodal/encoder_decoder_multimodal_offline.py --model-type whisper --seed 0
# Pooling models
- python3 pooling/embed/vision_embedding_offline.py --seed 0
# Features demo
- python3 features/automatic_prefix_caching/prefix_caching_offline.py
- python3 deployment/llm_engine_example.py
- python3 features/tensorize_vllm_model.py --model facebook/opt-125m serialize --serialized-directory /tmp/ --suffix v1 && python3 features/tensorize_vllm_model.py --model facebook/opt-125m deserialize --path-to-tensors /tmp/vllm/facebook/opt-125m/v1/model.tensors
- python3 features/speculative_decoding/spec_decode_offline.py --test --method eagle --num_spec_tokens 3 --dataset-name hf --dataset-path philschmid/mt-bench --num-prompts 80 --temp 0 --top-p 1.0 --top-k -1 --tp 1 --enable-chunked-prefill --max-model-len 2048
- python3 features/speculative_decoding/spec_decode_offline.py --test --method eagle3 --num_spec_tokens 3 --dataset-name hf --dataset-path philschmid/mt-bench --num-prompts 80 --temp 0 --top-p 1.0 --top-k -1 --tp 1 --enable-chunked-prefill --max-model-len 1536
#---------------------------------------------------------- mi250 · kernels ----------------------------------------------------------#
- label: Kernels Core Operation Test # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx90anightly, amdmi250]
agent_pool: mi250_1
optional: true
working_dir: "/vllm-workspace/tests"
source_file_dependencies:
- csrc/
- tests/kernels/core
- tests/kernels/test_top_k_per_row.py
- tests/kernels/test_concat_mla_q.py
- vllm/model_executor/layers/rotary_embedding/
- vllm/_aiter_ops.py
- vllm/platforms/rocm.py
commands:
- pytest -v -s kernels/core --ignore=kernels/core/test_minimax_reduce_rms.py kernels/test_concat_mla_q.py kernels/test_top_k_per_row.py
- label: Kernels Helion Test # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx90anightly, amdmi250]
@@ -432,50 +313,6 @@ steps:
commands:
- pytest -v -s models/test_utils.py models/test_vision.py
- label: Basic Models Tests (Extra Initialization) %N # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx90anightly, amdmi250]
agent_pool: mi250_1
torch_nightly: true
parallelism: 2
working_dir: "/vllm-workspace/tests"
source_file_dependencies:
- vllm/model_executor/models/
- vllm/model_executor/layers/
- tests/models/test_initialization.py
- tests/models/registry.py
- vllm/_aiter_ops.py
- vllm/platforms/rocm.py
commands:
- pytest -v -s models/test_initialization.py -k 'not test_can_initialize_small_subset' --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --shard-id=$$BUILDKITE_PARALLEL_JOB
- label: Basic Models Tests (Initialization) # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx90anightly, amdmi250]
agent_pool: mi250_1
torch_nightly: true
working_dir: "/vllm-workspace/tests"
source_file_dependencies:
- vllm/
- tests/models/test_initialization.py
- tests/models/registry.py
commands:
- pytest -v -s models/test_initialization.py::test_can_initialize_small_subset
- label: Basic Models Tests (Other) # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx90anightly, amdmi250]
agent_pool: mi250_1
torch_nightly: true
working_dir: "/vllm-workspace/tests"
source_file_dependencies:
- vllm/
- tests/models/test_terratorch.py
- tests/models/test_transformers.py
- tests/models/test_registry.py
commands:
- pytest -v -s models/test_terratorch.py models/test_transformers.py models/test_registry.py
#----------------------------------------------------- mi250 · models / language -----------------------------------------------------#
- label: Language Models Test (MTEB) # TBD
@@ -500,53 +337,8 @@ steps:
commands:
- pytest -v -s models/language/generation_ppl_test
- label: Language Models Tests (Extra Standard) %N # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx90anightly, amdmi250]
agent_pool: mi250_1
torch_nightly: true
parallelism: 2
working_dir: "/vllm-workspace/tests"
source_file_dependencies:
- vllm/model_executor/models/
- vllm/model_executor/model_loader/
- vllm/model_executor/layers/
- vllm/v1/attention/backends/
- vllm/v1/attention/selector.py
- tests/models/language/pooling/test_embedding.py
- tests/models/language/generation/test_common.py
- tests/models/language/pooling/test_classification.py
- vllm/_aiter_ops.py
- vllm/platforms/rocm.py
commands:
- pip freeze | grep -E 'torch'
- pytest -v -s models/language -m 'core_model and slow_test' --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --shard-id=$$BUILDKITE_PARALLEL_JOB
#---------------------------------------------------- mi250 · models / multimodal ----------------------------------------------------#
- label: Multi-Modal Models (Extended Generation 2) # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx90anightly, amdmi250]
agent_pool: mi250_1
working_dir: "/vllm-workspace/tests"
source_file_dependencies:
- vllm/
- tests/models/multimodal/generation
commands:
- pip install git+https://github.com/TIGER-AI-Lab/Mantis.git
- pytest -v -s models/multimodal/generation/test_common.py -m 'split(group=0) and not core_model'
- label: Multi-Modal Models (Extended Pooling) # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx90anightly, amdmi250]
agent_pool: mi250_1
working_dir: "/vllm-workspace/tests"
source_file_dependencies:
- vllm/
- tests/models/multimodal/pooling
commands:
- pytest -v -s models/multimodal/pooling -m 'not core_model'
- label: "Multi-Modal Models (Standard) 3: llava + qwen2_vl" # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx90anightly, amdmi250]
@@ -780,21 +572,6 @@ steps:
- TP_SIZE=1 DP_SIZE=2 pytest -v -s v1/distributed/test_external_lb_dp.py
- DP_SIZE=2 pytest -v -s entrypoints/openai/test_multi_api_servers.py
- label: NixlConnector PD + Spec Decode acceptance (2 GPUs) # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx90anightly, amdmi250]
agent_pool: mi250_2
num_gpus: 2
working_dir: "/vllm-workspace/tests"
source_file_dependencies:
- vllm/distributed/kv_transfer/kv_connector/v1/nixl_connector.py
- vllm/v1/worker/kv_connector_model_runner_mixin.py
- tests/v1/kv_connector/nixl_integration/
- vllm/platforms/rocm.py
commands:
- uv pip install --system -r /vllm-workspace/requirements/kv_connectors_rocm.txt
- ATTENTION_BACKEND=TRITON_ATTN bash v1/kv_connector/nixl_integration/spec_decode_acceptance_test.sh
- label: V1 e2e (2 GPUs) # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx90anightly, amdmi250]
@@ -807,20 +584,6 @@ steps:
commands:
- pytest -v -s v1/e2e/spec_decode/test_spec_decode.py -k "tensor_parallelism"
- label: Distributed NixlConnector PD accuracy (4 GPUs) # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx90anightly, amdmi250]
agent_pool: mi250_4
num_gpus: 4
working_dir: "/vllm-workspace/tests"
source_file_dependencies:
- vllm/distributed/kv_transfer/kv_connector/v1/nixl_connector.py
- tests/v1/kv_connector/nixl_integration/
- vllm/platforms/rocm.py
commands:
- uv pip install --system -r /vllm-workspace/requirements/kv_connectors_rocm.txt
- ATTENTION_BACKEND=TRITON_ATTN bash v1/kv_connector/nixl_integration/config_sweep_accuracy_test.sh
#------------------------------------------------------------- mi250 · misc ------------------------------------------------------------#
- label: Async Engine, Inputs, Utils, Worker, Config (CPU) # TBD
@@ -980,6 +743,24 @@ steps:
commands:
- pytest -s -v compile/passes --ignore compile/passes/distributed
- label: PyTorch Fullgraph # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx942nightly, amdmi300]
agent_pool: mi300_1
optional: true
torch_nightly: true
working_dir: "/vllm-workspace/tests"
source_file_dependencies:
- vllm/compilation/
- vllm/model_executor/
- vllm/v1/attention/
- vllm/config/compilation.py
- csrc/
- tests/compile
- vllm/platforms/rocm.py
commands:
- pytest -v -s compile/fullgraph/test_full_graph.py -k 'not test_fp8_kv_scale_compile'
- label: Pytorch Nightly Dependency Override Check # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx942nightly, amdmi300]
@@ -1057,6 +838,21 @@ steps:
- pytest -v -s distributed/test_eplb_algo.py
- pytest -v -s distributed/test_eplb_utils.py
- label: EPLB Execution # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx942nightly, amdmi300]
agent_pool: mi300_4
num_gpus: 4
working_dir: "/vllm-workspace/tests"
source_file_dependencies:
- vllm/distributed/eplb
- tests/distributed/test_eplb_execute.py
- tests/distributed/test_eplb_spec_decode.py
- vllm/platforms/rocm.py
commands:
- pytest -v -s distributed/test_eplb_execute.py
- pytest -v -s distributed/test_eplb_spec_decode.py
- label: Distributed Tests (2xH100-2xMI250) # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx942nightly, amdmi300]
@@ -1410,6 +1206,21 @@ steps:
commands:
- pytest -s -v test_lm_eval_correctness.py --config-list-file=configs/models-small-rocm.txt
- label: Multi-Modal Accuracy Eval (Small Models) # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx942nightly, amdmi300]
agent_pool: mi300_1
optional: true
working_dir: "/vllm-workspace/.buildkite/lm-eval-harness"
source_file_dependencies:
- vllm/multimodal/
- vllm/inputs/
- vllm/v1/core/
- vllm/platforms/rocm.py
- vllm/model_executor/model_loader/
commands:
- pytest -s -v test_lm_eval_correctness.py --config-list-file=configs/models-mm-small.txt --tp-size=1
- label: GPQA Eval (GPT-OSS) (2xH100-2xMI300) # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx942nightly, amdmi300]
@@ -1742,6 +1553,52 @@ steps:
- pytest -v -s model_executor -m '(not slow_test)'
- pytest -v -s entrypoints/openai/completion/test_tensorizer_entrypoint.py
#------------------------------------------------------ mi300 · models / basic -------------------------------------------------------#
- label: Basic Models Tests (Extra Initialization) %N # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx942nightly, amdmi300]
agent_pool: mi300_1
torch_nightly: true
parallelism: 2
working_dir: "/vllm-workspace/tests"
source_file_dependencies:
- vllm/model_executor/models/
- vllm/model_executor/layers/
- tests/models/test_initialization.py
- tests/models/registry.py
- vllm/_aiter_ops.py
- vllm/platforms/rocm.py
commands:
- pytest -v -s models/test_initialization.py -k 'not test_can_initialize_small_subset' --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --shard-id=$$BUILDKITE_PARALLEL_JOB
- label: Basic Models Tests (Initialization) # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx942nightly, amdmi300]
agent_pool: mi300_1
torch_nightly: true
working_dir: "/vllm-workspace/tests"
source_file_dependencies:
- vllm/
- tests/models/test_initialization.py
- tests/models/registry.py
commands:
- pytest -v -s models/test_initialization.py::test_can_initialize_small_subset
- label: Basic Models Tests (Other) # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx942nightly, amdmi300]
agent_pool: mi300_1
torch_nightly: true
working_dir: "/vllm-workspace/tests"
source_file_dependencies:
- vllm/
- tests/models/test_terratorch.py
- tests/models/test_transformers.py
- tests/models/test_registry.py
commands:
- pytest -v -s models/test_terratorch.py models/test_transformers.py models/test_registry.py
#----------------------------------------------------- mi300 · models / language -----------------------------------------------------#
- label: Language Models Test (Extended Pooling) # TBD
@@ -1770,6 +1627,28 @@ steps:
- pip freeze | grep -E 'torch'
- pytest -v -s models/language -m 'core_model and (not slow_test)'
- label: Language Models Tests (Extra Standard) %N # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx942nightly, amdmi300]
agent_pool: mi300_1
torch_nightly: true
parallelism: 2
working_dir: "/vllm-workspace/tests"
source_file_dependencies:
- vllm/model_executor/models/
- vllm/model_executor/model_loader/
- vllm/model_executor/layers/
- vllm/v1/attention/backends/
- vllm/v1/attention/selector.py
- tests/models/language/pooling/test_embedding.py
- tests/models/language/generation/test_common.py
- tests/models/language/pooling/test_classification.py
- vllm/_aiter_ops.py
- vllm/platforms/rocm.py
commands:
- pip freeze | grep -E 'torch'
- pytest -v -s models/language -m 'core_model and slow_test' --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --shard-id=$$BUILDKITE_PARALLEL_JOB
#---------------------------------------------------- mi300 · models / multimodal ----------------------------------------------------#
- label: Multi-Modal Models (Extended Generation 1) # TBD
@@ -2336,6 +2215,21 @@ steps:
- TP_SIZE=1 DP_SIZE=2 pytest -v -s v1/distributed/test_external_lb_dp.py
- DP_SIZE=2 pytest -v -s entrypoints/openai/test_multi_api_servers.py
- label: NixlConnector PD + Spec Decode acceptance (2 GPUs) # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx942nightly, amdmi300]
agent_pool: mi300_2
num_gpus: 2
working_dir: "/vllm-workspace/tests"
source_file_dependencies:
- vllm/distributed/kv_transfer/kv_connector/v1/nixl_connector.py
- vllm/v1/worker/kv_connector_model_runner_mixin.py
- tests/v1/kv_connector/nixl_integration/
- vllm/platforms/rocm.py
commands:
- uv pip install --system -r /vllm-workspace/requirements/kv_connectors_rocm.txt
- ATTENTION_BACKEND=TRITON_ATTN bash v1/kv_connector/nixl_integration/spec_decode_acceptance_test.sh
- label: Distributed Tests (2xH100-2xMI300) # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx942nightly, amdmi300]
@@ -3411,52 +3305,6 @@ steps:
commands:
- pytest -v -s -m 'not slow_test' v1/spec_decode
- label: NixlConnector PD + Spec Decode acceptance (2 GPUs) # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx950nightly, amdmi355]
agent_pool: mi355_2
num_gpus: 2
optional: true
working_dir: "/vllm-workspace/tests"
source_file_dependencies:
- vllm/distributed/kv_transfer/kv_connector/v1/nixl_connector.py
- vllm/v1/worker/kv_connector_model_runner_mixin.py
- tests/v1/kv_connector/nixl_integration/
- vllm/platforms/rocm.py
commands:
- uv pip install --system -r /vllm-workspace/requirements/kv_connectors_rocm.txt
- ATTENTION_BACKEND=TRITON_ATTN bash v1/kv_connector/nixl_integration/spec_decode_acceptance_test.sh
- label: Distributed NixlConnector PD accuracy (4 GPUs) # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx950nightly, amdmi355]
agent_pool: mi355_4
num_gpus: 4
optional: true
working_dir: "/vllm-workspace/tests"
source_file_dependencies:
- vllm/distributed/kv_transfer/kv_connector/v1/nixl_connector.py
- tests/v1/kv_connector/nixl_integration/
- vllm/platforms/rocm.py
commands:
- uv pip install --system -r /vllm-workspace/requirements/kv_connectors_rocm.txt
- ATTENTION_BACKEND=TRITON_ATTN bash v1/kv_connector/nixl_integration/config_sweep_accuracy_test.sh
- label: DP EP Distributed NixlConnector PD accuracy tests (4 GPUs) # TBD
timeout_in_minutes: 180
mirror_hardwares: [amdexperimental, amdproduction, amdgfx950nightly, amdmi355]
agent_pool: mi355_4
num_gpus: 4
optional: true
working_dir: "/vllm-workspace/tests"
source_file_dependencies:
- vllm/distributed/kv_transfer/kv_connector/v1/nixl_connector.py
- tests/v1/kv_connector/nixl_integration/
- vllm/platforms/rocm.py
commands:
- uv pip install --system -r /vllm-workspace/requirements/kv_connectors_rocm.txt
- DP_EP=1 ATTENTION_BACKEND=TRITON_ATTN bash v1/kv_connector/nixl_integration/config_sweep_accuracy_test.sh
#------------------------------------------------------ mi355 · weight_loading -------------------------------------------------------#
- label: Weight Loading Multiple GPU # TBD