[CI] Standardize test job labels by device (#52659)

Signed-off-by: Kevin Luu <51931015+khluu@users.noreply.github.com> Co-authored-by: OpenAI Codex <codex@openai.com>
2026-08-18 12:50:32 -07:00
parent 9842d70145
commit 8d6b18329a
35 changed files with 268 additions and 210 deletions
@@ -2,7 +2,7 @@ group: Attention
 depends_on: 
  - image-build
 steps:
- label: V1 attention (H100-MI300)
+- label: ":nvidia: (H100) V1 Attention Shard %N"
  key: v1-attention-h100-mi300
  timeout_in_minutes: 85
  device: h100
@@ -16,6 +16,7 @@ steps:
  parallelism: 2
  mirror:
    amd:
+      label: ":amd: (MI300) V1 Attention Shard %N"
      dind: false
      device: mi300_1
      timeout_in_minutes: 125
@@ -30,7 +31,7 @@ steps:
      - vllm/envs.py
      - vllm/platforms/rocm.py

- label: V1 attention (B200)
+- label: ":nvidia: (B200) V1 Attention Shard %N"
  key: v1-attention-b200
  timeout_in_minutes: 80
  device: b200-k8s
@@ -2,7 +2,7 @@ group: Basic Correctness
 depends_on: 
  - image-build
 steps:
- label: Basic Correctness
+- label: ":nvidia: (H200) Basic Correctness"
  key: basic-correctness
  timeout_in_minutes: 68
  device: h200_18gb
@@ -19,6 +19,7 @@ steps:
  - pytest -v -s basic_correctness/test_cpu_offload.py
  mirror:
    amd:
+      label: ":amd: (MI300) Basic Correctness"
      dind: false
      device: mi300_1
      timeout_in_minutes: 60
@@ -2,7 +2,7 @@ group: Benchmarks
 depends_on: 
  - image-build
 steps:
- label: Benchmarks CLI Test
+- label: ":nvidia: (H200) Benchmarks CLI"
  key: benchmarks-cli-test
  timeout_in_minutes: 45
  device: h200_18gb
@@ -14,13 +14,14 @@ steps:
  - pytest -v -s benchmarks/
  mirror:
    amd:
+      label: ":amd: (MI300) Benchmarks CLI"
      dind: false
      device: mi300_1
      timeout_in_minutes: 40
      depends_on:
      - image-build-amd

- label: Attention Benchmarks Smoke Test (B200)
+- label: ":nvidia: (B200) Attention Benchmark Smoke"
  key: attention-benchmarks-smoke-test-b200
  device: b200-k8s
  num_gpus: 2
@@ -2,7 +2,7 @@ group: Compile
 depends_on: 
  - image-build
 steps:
- label: Sequence Parallel Correctness Tests (2xB200)
+- label: ":nvidia: (B200) Sequence Parallel Correctness"
  key: sequence-parallel-correctness-tests-2xb200
  timeout_in_minutes: 45
  working_dir: "/vllm-workspace/"
@@ -13,7 +13,7 @@ steps:
  - export VLLM_TEST_CLEAN_GPU_MEMORY=1
  - pytest -v -s tests/compile/correctness_e2e/test_sequence_parallel.py

- label: AsyncTP Correctness Tests (2xB200)
+- label: ":nvidia: (B200) AsyncTP Correctness"
  key: asynctp-correctness-tests-b200
  timeout_in_minutes: 30
  working_dir: "/vllm-workspace/"
@@ -24,7 +24,7 @@ steps:
  - export VLLM_TEST_CLEAN_GPU_MEMORY=1
  - pytest -v -s tests/compile/correctness_e2e/test_async_tp.py

- label: Distributed Compile Unit Tests (2xH100)
+- label: ":nvidia: (H100) Distributed Compile"
  key: distributed-compile-unit-tests-2xh100
  timeout_in_minutes: 45
  working_dir: "/vllm-workspace/"
@@ -38,7 +38,7 @@ steps:
  - export VLLM_TEST_CLEAN_GPU_MEMORY=1
  - pytest -s -v tests/compile/passes/distributed

- label: Fusion and Compile Unit Tests (2xB200)
+- label: ":nvidia: (B200) Fusion and Compile"
  key: fusion-and-compile-unit-tests-2xb200
  timeout_in_minutes: 30
  working_dir: "/vllm-workspace/"
@@ -65,7 +65,7 @@ steps:
    # this runner has 2 GPUs available even though num_devices=2 is not set
    - pytest -v -s tests/compile/passes/distributed/test_fusion_all_reduce.py

- label: Fusion E2E Quick (H100)
+- label: ":nvidia: (H100) Fusion E2E Quick"
  key: fusion-e2e-quick-h100
  timeout_in_minutes: 25
  working_dir: "/vllm-workspace/"
@@ -84,7 +84,7 @@ steps:
    # Qwen/Deepseek requires +quant_fp8 as -quant_fp8 rms+quant fusion is not supported
    - pytest -v -s tests/compile/fusions_e2e/test_tp1_quant.py -k "inductor_partition and not +rms_norm and +quant_fp8 and (qwen3 or deepseek)"

- label: Fusion E2E Config Sweep (H100)
+- label: ":nvidia: (H100) Fusion E2E Config Sweep"
  key: fusion-e2e-config-sweep-h100
  timeout_in_minutes: 25
  working_dir: "/vllm-workspace/"
@@ -104,7 +104,7 @@ steps:
    # Run just llama3 (fp8) for all config combinations
    - pytest -v -s tests/compile/fusions_e2e/test_tp1_quant.py -k "llama-3"

- label: Fusion E2E Config Sweep (B200)
+- label: ":nvidia: (B200) Fusion E2E Config Sweep"
  key: fusion-e2e-config-sweep-b200
  timeout_in_minutes: 30
  working_dir: "/vllm-workspace/"
@@ -118,7 +118,7 @@ steps:
    # Run just llama3 (fp8 & fp4) for all config combinations (only inductor partition)
    - pytest -v -s tests/compile/fusions_e2e/test_tp1_quant.py -k "inductor_partition and (FLASHINFER and not +rms_norm and (not +quant_fp8 or +quant_fp8 and (qwen3 or deepseek)) or llama-3)"

- label: Fusion E2E TP2 Quick (H100)
+- label: ":nvidia: (H100) Fusion E2E TP2 Quick"
  key: fusion-e2e-tp2-quick-h100
  timeout_in_minutes: 35
  working_dir: "/vllm-workspace/"
@@ -136,7 +136,7 @@ steps:
    - pytest -v -s tests/compile/fusions_e2e/test_tp2_ar_rms.py -k "inductor_partition and not +rms_norm and (not +quant_fp8 or +quant_fp8 and (qwen3 or deepseek))"
    - pytest -v -s tests/compile/fusions_e2e/test_tp2_async_tp.py -k "inductor_partition and not +rms_norm and (not +quant_fp8 or +quant_fp8 and (qwen3 or deepseek))"

- label: Fusion E2E TP2 AR-RMS Config Sweep (H100)
+- label: ":nvidia: (H100) Fusion E2E TP2 AR-RMS Config Sweep"
  key: fusion-e2e-tp2-ar-rms-config-sweep-h100
  timeout_in_minutes: 30
  working_dir: "/vllm-workspace/"
@@ -156,7 +156,7 @@ steps:
    # Run just llama3 (fp8 & bf16) for all config combinations
    - pytest -v -s tests/compile/fusions_e2e/test_tp2_ar_rms.py -k "llama-3"

- label: Fusion E2E TP2 AsyncTP Config Sweep (H100)
+- label: ":nvidia: (H100) Fusion E2E TP2 AsyncTP Config Sweep"
  key: fusion-e2e-tp2-asynctp-config-sweep-h100
  timeout_in_minutes: 40
  working_dir: "/vllm-workspace/"
@@ -176,7 +176,7 @@ steps:
    # Run just llama3 (fp8 & bf16) for all config combinations
    - pytest -v -s tests/compile/fusions_e2e/test_tp2_async_tp.py -k "llama-3"

- label: Fusion E2E TP2 (B200)
+- label: ":nvidia: (B200) Fusion E2E TP2"
  key: fusion-e2e-tp2-b200
  timeout_in_minutes: 45
  working_dir: "/vllm-workspace/"
@@ -2,7 +2,7 @@ group: CUDA
 depends_on: 
  - image-build
 steps:
- label: Platform Tests
+- label: ":nvidia: (H200) CUDA Platform"
  key: platform-tests
  timeout_in_minutes: 20
  device: h200_18gb
@@ -18,7 +18,7 @@ steps:
    - pytest -v -s cuda/test_platform_no_cuda_init.py
    - pytest -v -s cuda/test_cuda_compatibility_path.py

- label: Cudagraph
+- label: ":nvidia: (H200) CUDAGraph"
  device: h200_35gb
  key: cudagraph
  timeout_in_minutes: 30
@@ -2,7 +2,7 @@ group: Disaggregated
 depends_on: 
  - image-build
 steps:
- label: Distributed NixlConnector PD accuracy (4 GPUs)
+- label: ":nvidia: (L4) Distributed NixlConnector PD accuracy"
  key: distributed-nixlconnector-pd-accuracy-4-gpus
  timeout_in_minutes: 55
  working_dir: "/vllm-workspace/tests"
@@ -15,6 +15,7 @@ steps:
    - bash v1/kv_connector/nixl_integration/config_sweep_accuracy_test.sh
  mirror:
    amd:
+      label: ":amd: (MI300) Distributed NixlConnector PD accuracy"
      dind: false
      device: mi300_4
      timeout_in_minutes: 60
@@ -28,7 +29,7 @@ steps:
        - 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: Distributed FlashInfer NixlConnector PD accuracy (4 GPUs)
+- label: ":nvidia: (L4) Distributed FlashInfer NixlConnector PD accuracy"
  key: distributed-flashinfer-nixlconnector-pd-accuracy-4-gpus
  timeout_in_minutes: 55
  working_dir: "/vllm-workspace/tests"
@@ -40,7 +41,7 @@ steps:
    - bash /vllm-workspace/.buildkite/scripts/install-kv-connectors.sh
    - FLASHINFER=1 bash v1/kv_connector/nixl_integration/config_sweep_accuracy_test.sh

- label: Push NixlConnector PP prefill PD accuracy (4 GPUs)
+- label: ":nvidia: (L4) Push NixlConnector PP prefill PD accuracy"
  key: push-nixlconnector-pp-prefill-pd-accuracy-4-gpus
  timeout_in_minutes: 30
  working_dir: "/vllm-workspace/tests"
@@ -53,7 +54,7 @@ steps:
    - bash /vllm-workspace/.buildkite/scripts/install-kv-connectors.sh
    - bash v1/kv_connector/nixl_push_integration/config_sweep_accuracy_test.sh

- label: DP EP Distributed NixlConnector PD accuracy tests (4 GPUs)
+- label: ":nvidia: (L4) DP EP Distributed NixlConnector PD accuracy"
  key: dp-ep-distributed-nixlconnector-pd-accuracy-tests-4-gpus
  timeout_in_minutes: 30
  working_dir: "/vllm-workspace/tests"
@@ -66,6 +67,7 @@ steps:
    - DP_EP=1 bash v1/kv_connector/nixl_integration/config_sweep_accuracy_test.sh
  mirror:
    amd:
+      label: ":amd: (MI300) DP EP Distributed NixlConnector PD accuracy"
      dind: false
      device: mi300_4
      timeout_in_minutes: 40
@@ -79,7 +81,7 @@ steps:
        - 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

- label: CrossLayer KV layout Distributed NixlConnector PD accuracy tests (4 GPUs)
+- label: ":nvidia: (L4) CrossLayer KV layout Distributed NixlConnector PD accuracy"
  key: crosslayer-kv-layout-distributed-nixlconnector-pd-accuracy-tests-4-gpus
  timeout_in_minutes: 55
  working_dir: "/vllm-workspace/tests"
@@ -92,6 +94,7 @@ steps:
    - CROSS_LAYERS_BLOCKS=True bash v1/kv_connector/nixl_integration/config_sweep_accuracy_test.sh
  mirror:
    amd:
+      label: ":amd: (MI300) CrossLayer KV layout Distributed NixlConnector PD accuracy"
      dind: false
      device: mi300_4
      timeout_in_minutes: 60
@@ -105,7 +108,7 @@ steps:
        - uv pip install --system -r /vllm-workspace/requirements/kv_connectors_rocm.txt
        - CROSS_LAYERS_BLOCKS=True ATTENTION_BACKEND=TRITON_ATTN bash v1/kv_connector/nixl_integration/config_sweep_accuracy_test.sh

- label: Hybrid SSM NixlConnector PD accuracy tests (4 GPUs)
+- label: ":nvidia: (L4) Hybrid SSM NixlConnector PD accuracy"
  key: hybrid-ssm-nixlconnector-pd-accuracy-tests-4-gpus
  timeout_in_minutes: 60
  working_dir: "/vllm-workspace/tests"
@@ -118,6 +121,7 @@ steps:
    - HYBRID_SSM=1 bash v1/kv_connector/nixl_integration/config_sweep_accuracy_test.sh
  mirror:
    amd:
+      label: ":amd: (MI300) Hybrid SSM NixlConnector PD accuracy"
      dind: false
      device: mi300_4
      timeout_in_minutes: 55
@@ -131,7 +135,7 @@ steps:
        - uv pip install --system -r /vllm-workspace/requirements/kv_connectors_rocm.txt
        - HYBRID_SSM=1 ATTENTION_BACKEND=TRITON_ATTN bash v1/kv_connector/nixl_integration/config_sweep_accuracy_test.sh

- label: NixlConnector PD edge case test (2 GPUs)
+- label: ":nvidia: (L4) NixlConnector PD edge case"
  key: nixlconnector-pd-edge-cases-2-gpus
  timeout_in_minutes: 40
  working_dir: "/vllm-workspace/tests"
@@ -147,7 +151,7 @@ steps:
    - bash /vllm-workspace/.buildkite/scripts/install-kv-connectors.sh
    - bash v1/kv_connector/nixl_integration/run_edge_case_test.sh

- label: Hybrid SSM NixlConnector PD prefix cache test (2 GPUs)
+- label: ":nvidia: (L4) Hybrid SSM NixlConnector PD prefix cache"
  key: hybrid-ssm-nixlconnector-pd-prefix-cache-2-gpus
  timeout_in_minutes: 25
  working_dir: "/vllm-workspace/tests"
@@ -161,7 +165,7 @@ steps:
    - bash /vllm-workspace/.buildkite/scripts/install-kv-connectors.sh
    - bash v1/kv_connector/nixl_integration/run_mamba_prefix_cache_test.sh

- label: MultiConnector (Nixl+Offloading) PD accuracy (2 GPUs)
+- label: ":nvidia: (L4) MultiConnector (Nixl+Offloading) PD accuracy"
  key: multiconnector-nixl-offloading-pd-accuracy-2-gpus
  timeout_in_minutes: 40
  working_dir: "/vllm-workspace/tests"
@@ -176,7 +180,7 @@ steps:
    - bash /vllm-workspace/.buildkite/scripts/install-kv-connectors.sh
    - bash v1/kv_connector/nixl_integration/run_multi_connector_accuracy_test.sh

- label: NixlConnector PD + Spec Decode acceptance (2 GPUs)
+- label: ":nvidia: (A100) NixlConnector PD + Spec Decode acceptance"
  key: nixlconnector-pd-spec-decode-acceptance-2-gpus
  timeout_in_minutes: 45
  device: a100
@@ -191,6 +195,7 @@ steps:
    - bash v1/kv_connector/nixl_integration/config_sweep_spec_decode_test.sh
  mirror:
    amd:
+      label: ":amd: (MI300) NixlConnector PD + Spec Decode acceptance"
      dind: false
      device: mi300_2
      timeout_in_minutes: 45
@@ -205,7 +210,7 @@ steps:
        - uv pip install --system -r /vllm-workspace/requirements/kv_connectors_rocm.txt
        - KV_CACHE_MEMORY_BYTES=8G ATTENTION_BACKEND=TRITON_ATTN bash v1/kv_connector/nixl_integration/config_sweep_spec_decode_test.sh

- label: MultiConnector (Nixl+Offloading) PD edge cases (2 GPUs)
+- label: ":nvidia: (L4) MultiConnector (Nixl+Offloading) PD edge cases"
  key: multiconnector-nixl-offloading-pd-edge-cases-2-gpus
  timeout_in_minutes: 25
  working_dir: "/vllm-workspace/tests"
@@ -221,7 +226,7 @@ steps:
    - bash v1/kv_connector/nixl_integration/run_multi_connector_edge_case_test.sh

 # P TP 4 - D DPEP 4 test case for DSv4-Flash
- label: DSv4-Flash Disaggregated DP EP
+- label: ":nvidia: (H200) DSv4-Flash Disaggregated DP EP"
  key: dsv4-flash-disaggregated
  timeout_in_minutes: 60
  device: h200
@@ -242,7 +247,7 @@ steps:
    - bash /vllm-workspace/.buildkite/scripts/install-kv-connectors.sh
    - bash v1/kv_connector/nixl_integration/run_accuracy_test.sh

- label: Kimi-Linear-48B-A3B Disaggregated DP EP
+- label: ":nvidia: (H200) Kimi-Linear-48B-A3B Disaggregated DP EP"
  key: kimi-linear-disaggregated
  timeout_in_minutes: 60
  device: h200
@@ -2,7 +2,7 @@ group: Disaggregated Mooncake
 depends_on:
  - image-build
 steps:
- label: Distributed MooncakeConnector PD accuracy (4 GPUs)
+- label: ":nvidia: (B200) Distributed MooncakeConnector PD accuracy"
  key: distributed-mooncakeconnector-pd-accuracy-4-gpus
  timeout_in_minutes: 10
  working_dir: "/vllm-workspace/tests"
@@ -2,7 +2,7 @@ group: Distributed
 depends_on: 
  - image-build
 steps:
- label: Distributed Comm Ops
+- label: ":nvidia: (L4) Distributed Comm Ops"
  key: distributed-comm-ops
  timeout_in_minutes: 25
  working_dir: "/vllm-workspace/tests"
@@ -17,7 +17,7 @@ steps:
  - pytest -v -s distributed/test_shm_buffer.py
  - pytest -v -s distributed/test_shm_storage.py

- label: Distributed DP Tests (2 GPUs)
+- label: ":nvidia: (L4) Distributed DP Basic"
  key: distributed-dp-tests-2-gpus
  timeout_in_minutes: 35
  working_dir: "/vllm-workspace/tests"
@@ -41,6 +41,7 @@ steps:
  - DP_SIZE=2 pytest -v -s entrypoints/openai/test_multi_api_servers.py
  mirror:
    amd:
+      label: ":amd: (MI300) Distributed DP Basic"
      dind: false
      device: mi300_2
      timeout_in_minutes: 45
@@ -58,7 +59,7 @@ steps:
      - tests/entrypoints/openai/test_multi_api_servers.py
      - vllm/platforms/rocm.py

- label: Distributed Compile + RPC Tests (2 GPUs)
+- label: ":nvidia: (L4) Distributed Compile + RPC"
  key: distributed-compile-rpc-tests-2-gpus
  timeout_in_minutes: 65
  working_dir: "/vllm-workspace/tests"
@@ -80,7 +81,7 @@ steps:
  - pytest -v -s entrypoints/llm/test_collective_rpc.py
  - pytest -v -s ./compile/test_wrapper.py

- label: Distributed Torchrun + Shutdown Tests (2 GPUs)
+- label: ":nvidia: (L4) Distributed Torchrun + Shutdown"
  key: distributed-torchrun-shutdown-tests-2-gpus
  timeout_in_minutes: 30
  working_dir: "/vllm-workspace/tests"
@@ -104,7 +105,7 @@ steps:
  - CUDA_VISIBLE_DEVICES=0,1 pytest -v -s v1/shutdown
  - pytest -v -s v1/worker/test_worker_memory_snapshot.py

- label: Distributed Torchrun + Examples (4 GPUs)
+- label: ":nvidia: (L4) Distributed Torchrun + Examples"
  key: distributed-torchrun-examples-4-gpus
  timeout_in_minutes: 30
  working_dir: "/vllm-workspace"
@@ -137,7 +138,7 @@ steps:
  - VLLM_ALLOW_INSECURE_SERIALIZATION=1 python3 examples/rl/rlhf_http_nccl.py
  - VLLM_ALLOW_INSECURE_SERIALIZATION=1 python3 examples/rl/rlhf_http_ipc.py

- label: Distributed DP Tests (4 GPUs)
+- label: ":nvidia: (L4) Distributed DP Extended"
  key: distributed-dp-tests-4-gpus
  timeout_in_minutes: 45
  working_dir: "/vllm-workspace/tests"
@@ -160,7 +161,7 @@ steps:
  - pytest -v -s v1/engine/test_engine_core_client.py::test_kv_cache_events_dp
  - pytest -v -s distributed/test_utils.py

- label: Distributed Compile + Comm (4 GPUs)
+- label: ":nvidia: (L4) Distributed Compile + Comm"
  key: distributed-compile-comm-4-gpus
  timeout_in_minutes: 70
  working_dir: "/vllm-workspace/tests"
@@ -181,7 +182,7 @@ steps:
  # test multi-node TP with multiproc executor (simulated on single node)
  - pytest -v -s distributed/test_multiproc_executor.py::test_multiproc_executor_multi_node

- label: Distributed Tests (8xH100)
+- label: ":nvidia: (H100) Distributed DP + EP"
  key: distributed-tests-8xh100
  timeout_in_minutes: 20
  device: h100
@@ -203,7 +204,7 @@ steps:
  # test with torchrun tp=2 and dp=4 with ep
  - torchrun --nproc-per-node=8 ../examples/features/torchrun/torchrun_dp_example_offline.py --tp-size=2 --pp-size=1 --dp-size=4 --enable-ep

- label: Distributed Tests (4xA100)
+- label: ":nvidia: (A100) Distributed"
  key: distributed-tests-4xa100
  device: a100
  optional: true
@@ -219,7 +220,7 @@ steps:
  - TARGET_TEST_SUITE=A100 pytest basic_correctness/ -v -s -m 'distributed(num_gpus=2)'
  - pytest -v -s -x lora/test_mixtral.py

- label: Distributed Tests (2xH100-2xMI300)
+- label: ":nvidia: (H100) Distributed Features"
  key: distributed-tests-2xh100-2xmi300
  timeout_in_minutes: 30
  device: h100
@@ -234,7 +235,7 @@ steps:
    - VLLM_ALLOW_INSECURE_SERIALIZATION=1 pytest -v -s tests/distributed/test_weight_transfer.py
    - pytest -v -s tests/distributed/test_packed_tensor.py

- label: Distributed Tests (2xB200)
+- label: ":nvidia: (B200) Distributed"
  key: distributed-tests-2xb200
  device: b200-k8s
  optional: true
@@ -249,7 +250,7 @@ steps:

    

- label: 2 Node Test (4 GPUs)
+- label: ":nvidia: (L4) Distributed 2-Node"
  key: 2-node-test-4-gpus
  timeout_in_minutes: 30
  working_dir: "/vllm-workspace/tests"
@@ -268,7 +269,7 @@ steps:
  commands:
    - ./.buildkite/scripts/run-multi-node-test.sh /vllm-workspace/tests 2 2 $IMAGE_TAG "VLLM_TEST_SAME_HOST=0 torchrun --nnodes 2 --nproc-per-node=2 --rdzv_backend=c10d --rdzv_endpoint=192.168.10.10 distributed/test_same_node.py | grep 'Same node test passed' && NUM_NODES=2 torchrun --nnodes 2 --nproc-per-node=2 --rdzv_backend=c10d --rdzv_endpoint=192.168.10.10 distributed/test_node_count.py | grep 'Node count test passed' && python3 ../examples/features/data_parallel/data_parallel_offline.py -dp=2 -tp=1 --dp-num-nodes=2 --dp-node-rank=0 --dp-master-addr=192.168.10.10 --dp-master-port=12345 --enforce-eager --trust-remote-code && VLLM_MULTI_NODE=1 pytest -v -s distributed/test_multi_node_assignment.py && VLLM_MULTI_NODE=1 pytest -v -s distributed/test_pipeline_parallel.py" "VLLM_TEST_SAME_HOST=0 torchrun --nnodes 2 --nproc-per-node=2 --rdzv_backend=c10d --rdzv_endpoint=192.168.10.10 distributed/test_same_node.py | grep 'Same node test passed' && NUM_NODES=2 torchrun --nnodes 2 --nproc-per-node=2 --rdzv_backend=c10d --rdzv_endpoint=192.168.10.10 distributed/test_node_count.py | grep 'Node count test passed' && python3 ../examples/features/data_parallel/data_parallel_offline.py -dp=2 -tp=1 --dp-num-nodes=2 --dp-node-rank=1 --dp-master-addr=192.168.10.10 --dp-master-port=12345 --enforce-eager --trust-remote-code"

- label: Pipeline + Context Parallelism (4 GPUs)
+- label: ":nvidia: (L4) Pipeline + Context Parallelism"
  key: pipeline-context-parallelism-4-gpus
  timeout_in_minutes: 55
  working_dir: "/vllm-workspace/tests"
@@ -284,7 +285,7 @@ steps:
  - pytest -v -s distributed/test_pp_cudagraph.py
  - pytest -v -s distributed/test_pipeline_parallel.py

- label: RayExecutorV2 (4 GPUs)
+- label: ":nvidia: (L4) RayExecutorV2"
  key: rayexecutorv2-4-gpus
  timeout_in_minutes: 45
  working_dir: "/vllm-workspace/tests"
@@ -2,7 +2,7 @@ group: Docker
 depends_on:
  - image-build-cpu
 steps:
- label: Docker Build Metadata
+- label: ":computer: (CPU) Docker Build Metadata"
  timeout_in_minutes: 20
  device: cpu-small
  source_file_dependencies:
@@ -2,7 +2,7 @@ group: E2E Integration
 depends_on: 
  - image-build
 steps:
- label: DeepSeek V2-Lite Sync EPLB Accuracy (4xH100)
+- label: ":nvidia: (H100) DeepSeek V2-Lite Sync EPLB Accuracy"
  key: deepseek-v2-lite-sync-eplb-accuracy-4xh100
  timeout_in_minutes: 25
  device: h100
@@ -12,7 +12,7 @@ steps:
  commands:
  - bash .buildkite/scripts/scheduled_integration_test/deepseek_v2_lite_ep_eplb.sh 0.25 200 8010

- label: Qwen3-30B-A3B-FP8-block Sync EPLB Accuracy (4xH100)
+- label: ":nvidia: (H100) Qwen3-30B-A3B-FP8-block Sync EPLB Accuracy"
  key: qwen3-30b-a3b-fp8-block-sync-eplb-accuracy-4xh100
  timeout_in_minutes: 25
  device: h100
@@ -22,7 +22,7 @@ steps:
  commands:
  - bash .buildkite/scripts/scheduled_integration_test/qwen30b_a3b_fp8_block_ep_eplb.sh 0.8 200 8020

- label: Qwen3-30B-A3B-FP8-block Sync EPLB Accuracy (2xB200)
+- label: ":nvidia: (B200) Qwen3-30B-A3B-FP8-block Sync EPLB Accuracy"
  key: qwen3-30b-a3b-fp8-block-sync-eplb-accuracy-2xb200
  timeout_in_minutes: 20
  device: b200-k8s
@@ -32,7 +32,7 @@ steps:
  commands:
  - bash .buildkite/scripts/scheduled_integration_test/qwen30b_a3b_fp8_block_ep_eplb.sh 0.8 200 8020 2 1

- label: Qwen3-30B-A3B-FP8 DP4 Async EPLB Accuracy
+- label: ":nvidia: (H100) Qwen3-30B-A3B-FP8 DP4 Async EPLB Accuracy"
  key: qwen3-30b-a3b-fp8-dp4-async-eplb-accuracy
  timeout_in_minutes: 25
  device: h100
@@ -42,7 +42,7 @@ steps:
  commands:
  - bash .buildkite/scripts/scheduled_integration_test/qwen30b_a3b_fp8_dp4_async_eplb.sh 0.8 200 8050

- label: DeepSeek V2-Lite Prefetch Offload Accuracy (H100)
+- label: ":nvidia: (H100) DeepSeek V2-Lite Prefetch Offload Accuracy"
  key: deepseek-v2-lite-prefetch-offload-accuracy-h100
  timeout_in_minutes: 20
  device: h100
@@ -2,7 +2,7 @@ group: Engine
 depends_on:
  - image-build
 steps:
- label: Engine
+- label: ":nvidia: (H200) Engine Core"
  key: engine
  timeout_in_minutes: 30
  device: h200_18gb
@@ -29,13 +29,14 @@ steps:
  - pytest -v -s engine test_sequence.py test_config.py test_logger.py test_vllm_port.py jit_monitor/test_hooks.py jit_monitor/test_hooks_gpu.py
  mirror:
    amd:
+      label: ":amd: (MI300) Engine Core"
      dind: false
      device: mi300_1
      timeout_in_minutes: 40
      depends_on:
      - image-build-amd

- label: Engine (1 GPU)
+- label: ":nvidia: (L4) V1 Engine"
  key: engine-1-gpu
  timeout_in_minutes: 45
  source_file_dependencies:
@@ -48,12 +49,13 @@ steps:
    - pytest -v -s v1/test_tensor_ipc_queue.py
  mirror:
    amd:
+      label: ":amd: (MI250) V1 Engine"
      device: mi250_1
      timeout_in_minutes: 45
      depends_on:
      - image-build-amd

- label: e2e Scheduling (1 GPU)
+- label: ":nvidia: (H200) E2E Scheduling"
  key: e2e-scheduling-1-gpu
  timeout_in_minutes: 53
  device: h200_18gb
@@ -64,12 +66,13 @@ steps:
    - pytest -v -s v1/e2e/general/test_async_scheduling.py
  mirror:
    amd:
+      label: ":amd: (MI250) E2E Scheduling"
      device: mi250_1
      timeout_in_minutes: 55
      depends_on:
      - image-build-amd

- label: e2e Core (1 GPU)
+- label: ":nvidia: (H200) E2E Core"
  device: h200_35gb
  key: e2e-core-1-gpu
  timeout_in_minutes: 40
@@ -80,6 +83,7 @@ steps:
    - pytest -v -s v1/e2e/general --ignore v1/e2e/general/test_async_scheduling.py
  mirror:
    amd:
+      label: ":amd: (MI250) E2E Core"
      device: mi250_1
      timeout_in_minutes: 50
      depends_on:
@@ -89,7 +93,7 @@ steps:
      - tests/v1/e2e/general/
      - vllm/platforms/rocm.py

- label: V1 e2e (2 GPUs)
+- label: ":nvidia: (L4) V1 E2E"
  key: v1-e2e-2-gpus
  timeout_in_minutes: 25 # TODO: Fix timeout after we have more confidence in the test stability
  optional: true
@@ -122,13 +126,14 @@ steps:
      v1/e2e/spec_decode/draft_model/test_draft_model.py::test_draft_model_engine_args_tensor_parallelism
  mirror:
    amd:
+      label: ":amd: (MI300) V1 E2E"
      dind: false
      device: mi300_2
      timeout_in_minutes: 30
      depends_on:
      - image-build-amd

- label: V1 e2e (4xH100)
+- label: ":nvidia: (H100) V1 E2E"
  key: v1-e2e-4xh100
  timeout_in_minutes: 35
  device: h100
@@ -2,7 +2,7 @@ group: Entrypoints
 depends_on: 
  - image-build
 steps:
- label: Entrypoints Unit Tests
+- label: ":nvidia: (H200) Entrypoints Unit"
  device: h200_35gb
  key: entrypoints-unit-tests
  timeout_in_minutes: 25
@@ -15,7 +15,7 @@ steps:
  - pytest -v -s entrypoints/unit_tests
  - pytest -v -s entrypoints/weight_transfer

- label: Entrypoints Integration (LLM)
+- label: ":nvidia: (H200) Entrypoints Integration (LLM)"
  device: h200_35gb
  key: entrypoints-integration-llm
  timeout_in_minutes: 60
@@ -31,13 +31,14 @@ steps:
  - pytest -v -s entrypoints/llm/offline_mode # Needs to avoid interference with other tests
  mirror:
    amd:
+      label: ":amd: (MI300) Entrypoints Integration (LLM)"
      dind: false
      device: mi300_1
      timeout_in_minutes: 55
      depends_on:
      - image-build-amd

- label: Entrypoints Integration (API Server)
+- label: ":nvidia: (H200) Entrypoints Integration (API Server)"
  key: entrypoints-integration-api-server
  device: h200_35gb
  timeout_in_minutes: 75
@@ -54,13 +55,14 @@ steps:
  - pytest -v -s entrypoints/scale_out
  mirror:
    amd:
+      label: ":amd: (MI300) Entrypoints Integration (API Server)"
      dind: false
      device: mi300_1
      timeout_in_minutes: 65
      depends_on:
      - image-build-amd

- label: Entrypoints Integration (API Server OpenAI - Part 1)
+- label: ":nvidia: (H200) Entrypoints Integration (API Server OpenAI - Part 1)"
  device: h200_35gb
  key: entrypoints-integration-api-server-openai-part-1
  timeout_in_minutes: 68
@@ -75,13 +77,14 @@ steps:
  - pytest -v -s entrypoints/openai --ignore=entrypoints/openai/completion --ignore=entrypoints/openai/chat_completion --ignore=entrypoints/openai/responses --ignore=entrypoints/openai/correctness
  mirror:
    amd:
+      label: ":amd: (MI300) Entrypoints Integration (API Server OpenAI - Part 1)"
      dind: false
      device: mi300_1
      timeout_in_minutes: 65
      depends_on:
      - image-build-amd

- label: Entrypoints Integration (API Server OpenAI - Part 2)
+- label: ":nvidia: (H200) Entrypoints Integration (API Server OpenAI - Part 2)"
  device: h200_35gb
  key: entrypoints-integration-api-server-openai-part-2
  timeout_in_minutes: 83
@@ -97,13 +100,14 @@ steps:
  - pytest -v -s entrypoints/openai/completion --ignore=entrypoints/openai/completion/test_tensorizer_entrypoint.py
  mirror:
    amd:
+      label: ":amd: (MI300) Entrypoints Integration (API Server OpenAI - Part 2)"
      dind: false
      device: mi300_1
      timeout_in_minutes: 70
      depends_on:
      - image-build-amd

- label: Entrypoints Integration (API Server Generate)
+- label: ":nvidia: (H200) Entrypoints Integration (API Server Generate)"
  device: h200_35gb
  key: entrypoints-integration-api-server-generate
  timeout_in_minutes: 50
@@ -122,13 +126,14 @@ steps:
  - pytest -v -s entrypoints/anthropic
  mirror:
    amd:
+      label: ":amd: (MI300) Entrypoints Integration (API Server Generate)"
      dind: false
      device: mi300_1
      timeout_in_minutes: 65
      depends_on:
      - image-build-amd

- label: Entrypoints Integration (Responses API)
+- label: ":nvidia: (H200) Entrypoints Integration (Responses API)"
  device: h200_35gb
  key: entrypoints-integration-responses-api
  timeout_in_minutes: 50
@@ -140,7 +145,7 @@ steps:
  commands:
  - pytest -v -s entrypoints/openai/responses

- label: Entrypoints Integration (Speech to Text)
+- label: ":nvidia: (H200) Entrypoints Integration (Speech to Text)"
  device: h200_35gb
  key: entrypoints-integration-speech_to_text
  timeout_in_minutes: 45
@@ -153,7 +158,7 @@ steps:
  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
  - pytest -v -s entrypoints/speech_to_text

- label: Entrypoints Integration (Multimodal)
+- label: ":nvidia: (H200) Entrypoints Integration (Multimodal)"
  device: h200_35gb
  key: entrypoints-integration-multimodal
  timeout_in_minutes: 45
@@ -166,7 +171,7 @@ steps:
  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
  - pytest -v -s entrypoints/multimodal

- label: Entrypoints Integration (Pooling)
+- label: ":nvidia: (H200) Entrypoints Integration (Pooling)"
  device: h200_35gb
  key: entrypoints-integration-pooling
  timeout_in_minutes: 75
@@ -179,7 +184,7 @@ steps:
  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
  - pytest -v -s entrypoints/pooling

- label: OpenAI API Correctness
+- label: ":nvidia: (H200) OpenAI API Correctness"
  key: openai-api-correctness
  timeout_in_minutes: 20
  device: h200_18gb
@@ -190,6 +195,7 @@ steps:
  - pytest -s entrypoints/openai/correctness/
  mirror:
    amd:
+      label: ":amd: (MI300) OpenAI API Correctness"
      dind: false
      device: mi300_1
      timeout_in_minutes: 30
@@ -2,7 +2,7 @@ group: Expert Parallelism
 depends_on: 
  - image-build
 steps:
- label: EPLB Algorithm
+- label: ":nvidia: (H200) EPLB Algorithm"
  key: eplb-algorithm
  timeout_in_minutes: 20
  device: h200_18gb
@@ -16,6 +16,7 @@ steps:
  - pytest -v -s distributed/test_eplb_utils.py
  mirror:
    amd:
+      label: ":amd: (MI300) EPLB Algorithm"
      dind: false
      device: mi300_1
      timeout_in_minutes: 30
@@ -27,7 +28,7 @@ steps:
      - tests/distributed/test_eplb_utils.py
      - vllm/platforms/rocm.py

- label: EPLB Execution # 17min
+- label: ":nvidia: (L4) EPLB Execution"
  key: eplb-execution
  timeout_in_minutes: 25
  working_dir: "/vllm-workspace/tests"
@@ -39,7 +40,7 @@ steps:
  - pytest -v -s distributed/test_eplb_execute.py
  - pytest -v -s distributed/test_eplb_spec_decode.py

- label: Elastic EP Scaling Test
+- label: ":nvidia: (H100) Elastic EP Scaling"
  key: elastic-ep-scaling-test
  timeout_in_minutes: 30
  device: h100
@@ -2,7 +2,7 @@ group: Fault Tolerance
 depends_on:
  - image-build
 steps:
- label: Fault Tolerance E2E (2xH100)
+- label: ":nvidia: (H100) Fault Tolerance E2E"
  key: fault-tolerance-e2e-2xh100
  timeout_in_minutes: 35
  device: h100
@@ -2,7 +2,7 @@ group: JIT Monitor
 depends_on:
  - image-build
 steps:
- label: No Runtime JITs e2e tests
+- label: ":nvidia: (H200) No Runtime JITs E2E"
  key: jit-monitor-no-runtime-jit
  device: h200_35gb
  timeout_in_minutes: 45
@@ -2,7 +2,7 @@ group: Kernels
 depends_on: 
  - image-build
 steps:
- label: vLLM IR Tests
+- label: ":nvidia: (H200) vLLM IR"
  key: vllm-ir-tests
  timeout_in_minutes: 35
  device: h200_18gb
@@ -14,7 +14,7 @@ steps:
    - pytest -v -s tests/ir
    - pytest -v -s tests/kernels/ir

- label: Kernels Core Operation Test
+- label: ":nvidia: (H200) Core Operation Kernels Shard %N"
  device: h200_35gb
  key: kernels-core-operation-test
  timeout_in_minutes: 120
@@ -27,7 +27,7 @@ steps:
    - pytest -v -s kernels/core --ignore=kernels/core/test_minimax_reduce_rms.py kernels/test_concat_mla_q.py kernels/test_fused_qk_norm_rope_gate.py --shard-id=$$BUILDKITE_PARALLEL_JOB --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT
  parallelism: 3

- label: Kernels MiniMax Reduce RMS Test (2 GPUs)
+- label: ":nvidia: (H100) MiniMax Reduce RMS Kernels"
  key: kernels-minimax-reduce-rms-test-2-gpus
  timeout_in_minutes: 20
  num_devices: 2
@@ -41,7 +41,7 @@ steps:
  commands:
    - pytest -v -s kernels/core/test_minimax_reduce_rms.py

- label: Deepseek V4 Kernel Test (H100)
+- label: ":nvidia: (H100) DeepSeek V4 Kernels"
  key: deepseek-v4-kernel-test-h100
  timeout_in_minutes: 30
  device: h100
@@ -54,7 +54,7 @@ steps:
    - pytest -v -s kernels/test_fused_deepseek_v4_*.py
    - pytest -v -s kernels/test_top_k_per_row.py

- label: Deepseek V4 Kernel Test (B200)
+- label: ":nvidia: (B200) DeepSeek V4 Kernels"
  key: deepseek-v4-kernel-test-b200
  timeout_in_minutes: 20
  device: b200-k8s
@@ -73,7 +73,7 @@ steps:
 # Files with dedicated jobs elsewhere in this file are excluded via --ignore
 # (test_kda, test_bf16x3_router_gemm_cutedsl and test_ll_bf16_gemm run in
 # their own jobs / Kernels (B200)).
- label: Kernels Root Misc Test (B200)
+- label: ":nvidia: (B200) Miscellaneous Kernels"
  key: kernels-root-misc-test-b200
  timeout_in_minutes: 45
  device: b200-k8s
@@ -101,7 +101,7 @@ steps:
    # BROKEN on main, pending kernel fixes (B200):
    #   test_shuffle_rows.py (1: test_shuffle_rows_edge_cases)

- label: Kernels Attention Test %N
+- label: ":nvidia: (L4) Attention Kernels Shard %N"
  key: kernels-attention-test
  timeout_in_minutes: 65
  source_file_dependencies:
@@ -116,6 +116,7 @@ steps:
  parallelism: 2
  mirror:
    amd:
+      label: ":amd: (MI300) Attention Kernels Shard %N"
      dind: false
      device: mi300_1
      timeout_in_minutes: 90
@@ -130,7 +131,7 @@ steps:
      - vllm/envs.py
      - vllm/platforms/rocm.py

- label: Kernels Attention DiffKV Test (H100)
+- label: ":nvidia: (H100) Attention DiffKV Kernels"
  key: kernels-attention-diffkv-test-h100
  timeout_in_minutes: 20
  device: h100
@@ -143,7 +144,7 @@ steps:
  commands:
    - pytest -v -s kernels/attention/test_triton_unified_attention_diffkv.py

- label: Kernels FlashMLA Test (H100)
+- label: ":nvidia: (H100) FlashMLA Kernels"
  key: kernels-flashmla-test-h100
  timeout_in_minutes: 25
  device: h100
@@ -161,7 +162,7 @@ steps:
    - pytest -v -s kernels/attention/test_flashmla_sparse.py
    - pytest -v -s kernels/attention/test_mla_cross_layer_kernel_equivalence.py

- label: Kernels Quantization Test %N
+- label: ":nvidia: (L4) Quantization Kernels Shard %N"
  key: kernels-quantization-test
  timeout_in_minutes: 60
  source_file_dependencies:
@@ -173,6 +174,7 @@ steps:
  parallelism: 2
  mirror:
    amd:
+      label: ":amd: (MI300) Quantization Kernels Shard %N"
      dind: false
      device: mi300_1
      timeout_in_minutes: 120
@@ -189,7 +191,7 @@ steps:
      depends_on:
      - image-build-amd

- label: Kernels MoE Test %N
+- label: ":nvidia: (L4) MoE Kernels Shard %N"
  key: kernels-moe-test
  timeout_in_minutes: 50
  source_file_dependencies:
@@ -206,6 +208,7 @@ steps:
  parallelism: 5
  mirror:
    amd:
+      label: ":amd: (MI300) MoE Kernels Shard %N"
      dind: false
      device: mi300_1
      timeout_in_minutes: 55
@@ -222,7 +225,7 @@ steps:
      depends_on:
      - image-build-amd

- label: Kernels Mamba Test
+- label: ":nvidia: (H200) Mamba Kernels"
  device: h200_35gb
  key: kernels-mamba-test
  timeout_in_minutes: 60
@@ -233,7 +236,7 @@ steps:
  commands:
    - pytest -v -s kernels/mamba

- label: Kernels DeepGEMM Test (H100)
+- label: ":nvidia: (H100) DeepGEMM Kernels"
  key: kernels-deepgemm-test-h100
  timeout_in_minutes: 35
  device: h100
@@ -260,7 +263,7 @@ steps:
    - pytest -v -s kernels/attention/test_deepgemm_attention.py
    - pytest -v -s quantization/test_cutlass_w4a16.py

- label: Kernels (B200)
+- label: ":nvidia: (B200) Kernels"
  key: kernels-b200
  timeout_in_minutes: 80
  working_dir: "/vllm-workspace/"
@@ -340,7 +343,7 @@ steps:
    # e2e
    - pytest -v -s tests/models/quantization/test_nvfp4.py

- label: B12X Linear Kernels (DGX Spark) Nightly
+- label: ":nvidia: (DGX) Spark B12X Linear Kernels Nightly"
  key: b12x-linear-kernels-dgx-spark-nightly
  timeout_in_minutes: 30
  device: dgx-spark
@@ -362,7 +365,7 @@ steps:
      model_executor/test_b12x_warmup.py
      kernels/quantization/test_block_fp8.py -k b12x

- label: Kernels Helion Test
+- label: ":nvidia: (H100) Helion Kernels Shard %N"
  key: kernels-helion-test
  timeout_in_minutes: 115
  device: h100
@@ -375,7 +378,7 @@ steps:
  parallelism: 2

 
- label: Kernels FP8 MoE Test (1xH100)
+- label: ":nvidia: (H100) FP8 MoE Kernels"
  key: kernels-fp8-moe-test-1xh100
  timeout_in_minutes: 40
  device: h100
@@ -392,7 +395,7 @@ steps:
    - pytest -v -s kernels/moe/test_triton_moe_no_act_mul.py
    - pytest -v -s kernels/moe/test_triton_moe_ptpc_fp8.py

- label: Kernels FP8 MoE Test (2xH100)
+- label: ":nvidia: (H100) DeepEP FP8 MoE Kernels"
  key: kernels-fp8-moe-test-2xh100
  timeout_in_minutes: 45
  device: h100
@@ -402,7 +405,7 @@ steps:
    - pytest -v -s kernels/moe/test_deepep_deepgemm_moe.py
    - pytest -v -s kernels/moe/test_deepep_moe.py

- label: Kernels Fp4 MoE Test (B200)
+- label: ":nvidia: (B200) FP4 MoE Kernels"
  key: kernels-fp4-moe-test-b200
  timeout_in_minutes: 25
  device: b200-k8s
@@ -415,7 +418,7 @@ steps:
    - pytest -v -s kernels/moe/test_ocp_mx_moe.py


- label: Kernels FusedMoE Layer Test (2 H100s)
+- label: ":nvidia: (H100) FusedMoE Layer Kernels"
  key: kernels-fusedmoe-layer-test-2-h100s
  timeout_in_minutes: 30
  device: h100
@@ -432,7 +435,7 @@ steps:
    - pytest -v -s kernels/moe/test_moe_layer.py


- label: Kernels FusedMoE Layer Test (2 B200s)
+- label: ":nvidia: (B200) FusedMoE Layer Kernels"
  key: kernels-fusedmoe-layer-test-2-b200s
  timeout_in_minutes: 90
  device: b200-k8s
@@ -2,7 +2,7 @@ group: LM Eval
 depends_on: 
  - image-build
 steps:
- label: LM Eval Small Models
+- label: ":nvidia: (H200) LM Eval Small Models"
  device: h200_35gb
  key: lm-eval-small-models
  timeout_in_minutes: 45
@@ -14,6 +14,7 @@ steps:
  - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=configs/models-small.txt
  mirror:
    amd:
+      label: ":amd: (MI300) LM Eval Small Models"
      dind: false
      device: mi300_1
      timeout_in_minutes: 45
@@ -42,7 +43,7 @@ steps:
 #   - export VLLM_WORKER_MULTIPROC_METHOD=spawn
 #   - pytest -s -v test_lm_eval_correctness.py --config-list-file=configs/models-large.txt --tp-size=4

- label: LM Eval Large Models (4xH100)
+- label: ":nvidia: (H100) LM Eval Large Models"
  key: lm-eval-large-models-4xh100
  device: h100
  optional: true
@@ -55,7 +56,7 @@ steps:
    - export VLLM_USE_DEEP_GEMM=0  # We found Triton is faster than DeepGEMM for H100
    - pytest -s -v test_lm_eval_correctness.py --config-list-file=configs/models-large-hopper.txt --tp-size=4

- label: LM Eval Small Models (1xB200)
+- label: ":nvidia: (B200) LM Eval Small Models"
  key: lm-eval-small-models-1xb200
  timeout_in_minutes: 50
  device: b200-k8s
@@ -66,7 +67,7 @@ steps:
  commands:
  - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=configs/models-blackwell.txt

- label: LM Eval Small Models Distributed (2xB200)
+- label: ":nvidia: (B200) LM Eval Small Models Distributed"
  key: lm-eval-small-models-distributed-2xb200
  timeout_in_minutes: 120
  device: b200-k8s
@@ -79,7 +80,7 @@ steps:
  commands:
  - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=configs/models-small-tp.txt

- label: LM Eval PCP (4xB200)
+- label: ":nvidia: (B200) LM Eval PCP"
  key: lm-eval-pcp-4xb200
  timeout_in_minutes: 360
  device: b200-k8s
@@ -101,7 +102,7 @@ steps:
  commands:
  - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=configs/models-pcp.txt

- label: LM Eval Spec Decode (4xB200)
+- label: ":nvidia: (B200) LM Eval Spec Decode"
  key: lm-eval-spec-decode-4xb200
  timeout_in_minutes: 120
  device: b200-k8s
@@ -121,7 +122,7 @@ steps:
  commands:
  - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=configs/models-spec-decode.txt

- label: LM Eval Large Models EP (2xB200)
+- label: ":nvidia: (B200) LM Eval Large Models EP"
  key: lm-eval-large-models-ep-2xb200
  timeout_in_minutes: 60
  device: b200-k8s
@@ -133,7 +134,7 @@ steps:
  commands:
  - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=configs/models-blackwell-ep.txt

- label: LM Eval Qwen3.5 Models (2xB200)
+- label: ":nvidia: (B200) LM Eval Qwen3.5 Models"
  key: lm-eval-qwen3-5-models-2xb200
  timeout_in_minutes: 45
  device: b200-k8s
@@ -150,7 +151,7 @@ steps:
  commands:
  - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=configs/models-qwen35-blackwell.txt

- label: LM Eval Large Models (8xH200)
+- label: ":nvidia: (H200) LM Eval Large Models"
  key: lm-eval-large-models-8xh200
  timeout_in_minutes: 50
  device: h200
@@ -160,6 +161,7 @@ steps:
    - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=configs/models-h200.txt
  mirror:
    amd:
+      label: ":amd: (MI300) LM Eval Large Models"
      dind: false
      device: mi300_8
      timeout_in_minutes: 40
@@ -170,7 +172,7 @@ steps:
      - export PYTORCH_ROCM_ARCH=gfx942 # Limit Quark compilation to save time
      - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=configs/models-mi3xx.txt

- label: MoE Refactor Integration Test (H100 - TEMPORARY)
+- label: ":nvidia: (H100) MoE Refactor Integration TEMPORARY"
  key: moe-refactor-integration-test-h100-temporary
  device: h100
  optional: true
@@ -178,7 +180,7 @@ steps:
  commands:
    - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=evals/gsm8k/configs/moe-refactor/config-h100.txt
  
- label: MoE Refactor Integration Test (B200 - TEMPORARY) %N
+- label: ":nvidia: (B200) MoE Refactor Integration TEMPORARY Shard %N"
  key: moe-refactor-integration-test-b200-temporary
  device: b200-k8s
  optional: true
@@ -187,7 +189,7 @@ steps:
  commands:
    - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=evals/gsm8k/configs/moe-refactor/config-b200-shard-$$BUILDKITE_PARALLEL_JOB.txt

- label: MoE Refactor Integration Test (B200 DP - TEMPORARY)
+- label: ":nvidia: (B200) MoE Refactor DP Integration TEMPORARY"
  key: moe-refactor-integration-test-b200-dp-temporary
  device: b200-k8s
  optional: true
@@ -195,7 +197,7 @@ steps:
  commands:
    - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=evals/gsm8k/configs/moe-refactor-dp-ep/config-b200.txt

- label: LM Eval Humming f16 (A100 - TEMPORARY) %N
+- label: ":nvidia: (A100) LM Eval Humming FP16 TEMPORARY Shard %N"
  key: lm-eval-humming-f16-a100
  timeout_in_minutes: 75
  device: a100
@@ -211,7 +213,7 @@ steps:
  commands:
  - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=evals/gsm8k/configs/humming/config-a100-shard-$$BUILDKITE_PARALLEL_JOB.txt

- label: LM Eval Humming Act int8 (A100 - TEMPORARY)
+- label: ":nvidia: (A100) LM Eval Humming Activation INT8 TEMPORARY"
  key: lm-eval-humming-act-a100
  timeout_in_minutes: 45
  device: a100
@@ -226,7 +228,7 @@ steps:
  commands:
  - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=evals/gsm8k/configs/humming/config-act-int8.txt

- label: LM Eval Humming f16 (H100 - TEMPORARY) %N
+- label: ":nvidia: (H100) LM Eval Humming FP16 TEMPORARY Shard %N"
  key: lm-eval-humming-f16-h100
  timeout_in_minutes: 70
  device: h100
@@ -242,7 +244,7 @@ steps:
  commands:
  - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=evals/gsm8k/configs/humming/config-h100-shard-$$BUILDKITE_PARALLEL_JOB.txt

- label: LM Eval Humming Act fp8/int8 (H100 - TEMPORARY)
+- label: ":nvidia: (H100) LM Eval Humming Activation FP8/INT8 TEMPORARY"
  key: lm-eval-humming-act-h100
  timeout_in_minutes: 70
  device: h100
@@ -258,7 +260,7 @@ steps:
  - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=evals/gsm8k/configs/humming/config-act-fp8.txt
  # - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=evals/gsm8k/configs/humming/config-act-int8.txt

- label: LM Eval Humming f16 (B200 - TEMPORARY)
+- label: ":nvidia: (B200) LM Eval Humming FP16 TEMPORARY"
  key: lm-eval-humming-f16-b200
  timeout_in_minutes: 50
  device: b200-k8s
@@ -273,7 +275,7 @@ steps:
  commands:
  - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=evals/gsm8k/configs/humming/config.txt

- label: LM Eval Humming Act fp8/int8 (B200 - TEMPORARY)
+- label: ":nvidia: (B200) LM Eval Humming Activation FP8/INT8 TEMPORARY"
  key: lm-eval-humming-act-b200
  timeout_in_minutes: 50
  device: b200-k8s
@@ -289,7 +291,7 @@ steps:
  - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=evals/gsm8k/configs/humming/config-act-fp8.txt
  - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=evals/gsm8k/configs/humming/config-act-int8.txt

- label: LM Eval TurboQuant KV Cache
+- label: ":nvidia: (H200) LM Eval TurboQuant KV Cache"
  key: lm-eval-turboquant-kv-cache
  timeout_in_minutes: 55
  device: h200_18gb
@@ -301,7 +303,7 @@ steps:
  commands:
  - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=evals/gsm8k/configs/models-turboquant.txt

- label: GPQA Eval (GPT-OSS) (2xH100)
+- label: ":nvidia: (H100) GPQA Eval (GPT-OSS)"
  key: gpqa-eval-gpt-oss-2xh100
  timeout_in_minutes: 35
  device: h100
@@ -315,7 +317,7 @@ steps:
    - uv pip install --system 'gpt-oss[eval]==0.0.5'
    - pytest -s -v evals/gpt_oss/test_gpqa_correctness.py --config-list-file=configs/models-h100.txt

- label: GPQA Eval (GPT-OSS) (2xB200)
+- label: ":nvidia: (B200) GPQA Eval (GPT-OSS)"
  key: gpqa-eval-gpt-oss-2xb200
  timeout_in_minutes: 30
  device: b200-k8s
@@ -329,7 +331,7 @@ steps:
    - uv pip install --system 'gpt-oss[eval]==0.0.5'
    - pytest -s -v evals/gpt_oss/test_gpqa_correctness.py --config-list-file=configs/models-b200.txt

- label: GPQA Eval (GPT-OSS) (DGX Spark)
+- label: ":nvidia: (DGX) Spark GPQA Eval (GPT-OSS)"
  key: gpqa-eval-gpt-oss-spark
  timeout_in_minutes: 35
  device: dgx-spark
@@ -345,7 +347,7 @@ steps:
    - uv pip install --system 'gpt-oss[eval]==0.0.5'
    - pytest -s -v evals/gpt_oss/test_gpqa_correctness.py --config-list-file=configs/models-spark.txt

- label: LM Eval KV-Offload (1xH200)
+- label: ":nvidia: (H200) LM Eval KV-Offload"
  key: kv-offload-small
  timeout_in_minutes: 30
  device: h200_35gb
@@ -358,7 +360,7 @@ steps:
  commands:
  - pytest -s -v evals/gsm8k/test_gsm8k_offloading.py -k "nemotron-h-8b or gemma-4-e4b-it"

- label: LM Eval KV-Offload (2xH100)
+- label: ":nvidia: (H100) LM Eval KV-Offload Medium"
  key: kv-offload-medium
  timeout_in_minutes: 45
  device: h100
@@ -372,7 +374,7 @@ steps:
  commands:
  - pytest -s -v evals/gsm8k/test_gsm8k_offloading.py -k "qwen3.5-35b or deepseek-v2-lite"

- label: LM Eval KV-Offload (4xH100)
+- label: ":nvidia: (H100) LM Eval KV-Offload Large"
  key: kv-offload-large
  timeout_in_minutes: 40
  device: h100
@@ -386,7 +388,7 @@ steps:
  commands:
  - pytest -s -v evals/gsm8k/test_gsm8k_offloading.py -k "deepseek-v4-flash"

- label: MRCR Eval Small Models
+- label: ":nvidia: (H200) MRCR Eval Small Models"
  device: h200_35gb
  timeout_in_minutes: 25
  source_file_dependencies:
@@ -2,7 +2,7 @@ group: LoRA
 depends_on: 
  - image-build
 steps:
- label: LoRA %N
+- label: ":nvidia: (H200) LoRA Shard %N"
  device: h200_35gb
  key: lora
  timeout_in_minutes: 40
@@ -14,6 +14,7 @@ steps:
  parallelism: 4
  mirror:
    amd:
+      label: ":amd: (MI300) LoRA Shard %N"
      dind: false
      device: mi300_1
      working_dir: "/vllm-workspace/tests"
@@ -26,7 +27,7 @@ steps:
      - image-build-amd


- label: LoRA TP (Distributed)
+- label: ":nvidia: (L4) LoRA TP (Distributed)"
  key: lora-tp-distributed
  timeout_in_minutes: 60
  num_devices: 4
@@ -2,7 +2,7 @@ group: Miscellaneous
 depends_on: 
  - image-build
 steps:
- label: V1 Sample + Logits
+- label: ":nvidia: (H200) V1 Sample + Logits"
  key: v1-sample-logits
  timeout_in_minutes: 83
  device: h200_18gb
@@ -33,13 +33,14 @@ steps:
    - pytest -v -s v1/test_outputs.py
  mirror:
    amd:
+      label: ":amd: (MI300) V1 Sample + Logits"
      dind: false
      device: mi300_1
      timeout_in_minutes: 70
      depends_on:
      - image-build-amd

- label: V1 Core + KV + Metrics
+- label: ":nvidia: (H200) V1 Core + KV + Metrics"
  device: h200_35gb
  key: v1-core-kv-metrics
  timeout_in_minutes: 80
@@ -90,13 +91,14 @@ steps:
    - pytest -v -s entrypoints/openai/correctness/test_lmeval.py::test_lm_eval_accuracy_v1_engine
  mirror:
    amd:
+      label: ":amd: (MI300) V1 Core + KV + Metrics"
      dind: false
      device: mi300_1
      timeout_in_minutes: 65
      depends_on:
      - image-build-amd

- label: V1 Others (CPU)
+- label: ":computer: (CPU) V1 Others"
  key: v1-others-cpu
  depends_on:
    - image-build-cpu
@@ -130,7 +132,7 @@ steps:
    - pytest -v -s -m 'cpu_test' v1/ec_connector/unit
    - pytest -v -s -m 'cpu_test' v1/metrics

- label: Extract Hidden States Integration
+- label: ":nvidia: (H200) Extract Hidden States Integration"
  key: extract-hidden-states-integration
  timeout_in_minutes: 20
  device: h200_18gb
@@ -144,7 +146,7 @@ steps:
    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
    - pytest -v -s v1/kv_connector/extract_hidden_states_integration

- label: Extract Hidden States Integration (2 GPUs)
+- label: ":nvidia: (L4) Extract Hidden States Integration"
  key: extract-hidden-states-integration-2-gpus
  timeout_in_minutes: 20
  num_devices: 2
@@ -158,7 +160,7 @@ steps:
    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
    - pytest -v -s -m 'distributed' v1/kv_connector/extract_hidden_states_integration

- label: Regression
+- label: ":nvidia: (H200) Regression"
  key: regression
  timeout_in_minutes: 30
  device: h200_18gb
@@ -181,7 +183,7 @@ steps:
  - pytest -v -s test_regression.py
  working_dir: "/vllm-workspace/tests" # optional

- label: Examples
+- label: ":nvidia: (H200) Examples"
  device: h200_35gb
  key: examples
  timeout_in_minutes: 40
@@ -215,6 +217,7 @@ steps:
    - 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
  mirror:
    amd:
+      label: ":amd: (MI300) Examples"
      dind: false
      device: mi300_1
      timeout_in_minutes: 75
@@ -226,7 +229,7 @@ steps:
      depends_on:
      - image-build-amd

- label: Metrics, Tracing (2 GPUs)
+- label: ":nvidia: (L4) Metrics, Tracing"
  key: metrics-tracing-2-gpus
  timeout_in_minutes: 25
  num_devices: 2
@@ -256,13 +259,14 @@ steps:
  - pytest -v -s tracing
  mirror:
    amd:
+      label: ":amd: (MI300) Metrics, Tracing"
      dind: false
      device: mi300_2
      timeout_in_minutes: 30
      depends_on:
      - image-build-amd

- label: Python-only Installation
+- label: ":nvidia: (H200) Python-only Installation"
  key: python-only-installation
  depends_on: ~
  optional: true
@@ -275,6 +279,7 @@ steps:
  - bash standalone_tests/python_only_compile.sh
  mirror:
    amd:
+      label: ":amd: (MI300) Python-only Installation"
      dind: false
      device: mi300_1
      timeout_in_minutes: 55
@@ -286,7 +291,7 @@ steps:
      - setup.py
      - vllm/platforms/rocm.py

- label: Async Engine, Inputs, Utils, Worker
+- label: ":nvidia: (H200) Async Engine, Inputs, Utils, Worker"
  device: h200_35gb
  key: async-engine-inputs-utils-worker
  timeout_in_minutes: 25
@@ -313,7 +318,7 @@ steps:
  - pytest -v -s -m 'not cpu_test' multimodal
  - pytest -v -s utils_

- label: Async Engine, Inputs, Utils, Worker, Config (CPU)
+- label: ":computer: (CPU) Async Engine, Inputs, Utils, Worker, Config"
  key: async-engine-inputs-utils-worker-config-cpu
  depends_on:
  - image-build-cpu
@@ -377,7 +382,7 @@ steps:
  - pytest -v -s transformers_utils
  - pytest -v -s config

- label: Batch Invariance (A100)
+- label: ":nvidia: (A100) Batch Invariance"
  key: batch-invariance-a100
  timeout_in_minutes: 60
  device: a100
@@ -391,7 +396,7 @@ steps:
    - pytest -v -s v1/determinism/test_batch_invariance.py
    - VLLM_TEST_MODEL=deepseek-ai/DeepSeek-V2-Lite-Chat pytest -v -s v1/determinism/test_batch_invariance.py::test_v1_generation_is_deterministic_across_batch_sizes_with_needle -k TRITON_MLA

- label: Batch Invariance (H100)
+- label: ":nvidia: (H100) Batch Invariance"
  key: batch-invariance-h100
  timeout_in_minutes: 60
  device: h100
@@ -407,7 +412,7 @@ steps:
    - VLLM_TEST_MODEL=deepseek-ai/DeepSeek-V2-Lite-Chat pytest -v -s v1/determinism/test_batch_invariance.py::test_v1_generation_is_deterministic_across_batch_sizes_with_needle -k TRITON_MLA
    - VLLM_TEST_MODEL=Qwen/Qwen3-30B-A3B-Thinking-2507-FP8 pytest -v -s v1/determinism/test_batch_invariance.py::test_v1_generation_is_deterministic_across_batch_sizes_with_needle -k FLASH_ATTN

- label: Batch Invariance (B200)
+- label: ":nvidia: (B200) Batch Invariance"
  key: batch-invariance-b200
  timeout_in_minutes: 45
  device: b200-k8s
@@ -427,7 +432,7 @@ steps:
    - pytest -v -s v1/determinism/test_cutlass_batch_invariance.py
    - pytest -v -s v1/determinism/test_online_batch_invariance.py

- label: Acceptance Length Test (Large Models) # optional
+- label: ":nvidia: (H200) Acceptance Length (Large Models)"
  device: h200_35gb
  key: acceptance-length-test-large-models
  timeout_in_minutes: 20
@@ -2,7 +2,7 @@ group: Model Executor
 depends_on: 
  - image-build
 steps:
- label: Model Executor
+- label: ":nvidia: (H200) Model Executor"
  device: h200_35gb
  key: model-executor
  timeout_in_minutes: 60
@@ -28,6 +28,7 @@ steps:
    - pytest -v -s entrypoints/openai/completion/test_tensorizer_entrypoint.py --timeout=900 --timeout-method=thread
  mirror:
    amd:
+      label: ":amd: (MI300) Model Executor"
      dind: false
      device: mi300_1
      timeout_in_minutes: 60
@@ -2,7 +2,7 @@ group: Model Runner V2
 depends_on:
  - image-build
 steps:
- label: Model Runner V2 Core Tests
+- label: ":nvidia: (H200) Model Runner V2 Core"
  device: h200_35gb
  key: model-runner-v2-core-tests
  timeout_in_minutes: 35
@@ -24,7 +24,7 @@ steps:
  # Temporary hack filter to exclude ngram spec decoding based tests.
  - pytest -v -s entrypoints/llm/test_struct_output_generate.py -k "xgrammar and not speculative_config6 and not speculative_config7 and not speculative_config8 and not speculative_config0"

- label: Model Runner V2 Examples
+- label: ":nvidia: (H200) Model Runner V2 Examples"
  device: h200_35gb
  key: model-runner-v2-examples
  timeout_in_minutes: 35
@@ -61,7 +61,7 @@ steps:
    # https://github.com/vllm-project/vllm/pull/26682 uses slightly more memory in PyTorch 2.9+ causing this test to OOM in 1xL4 GPU
    - 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

- label: Model Runner V2 Distributed (2 GPUs)
+- label: ":nvidia: (L4) Model Runner V2 Distributed"
  key: model-runner-v2-distributed-2-gpus
  timeout_in_minutes: 30
  working_dir: "/vllm-workspace/tests"
@@ -82,7 +82,7 @@ steps:
    - TP_SIZE=1 DP_SIZE=2 pytest -v -s v1/distributed/test_async_llm_dp.py -k "not ray"
    - TP_SIZE=1 DP_SIZE=2 pytest -v -s v1/distributed/test_eagle_dp.py

- label: Model Runner V2 Pipeline Parallelism (4 GPUs)
+- label: ":nvidia: (L4) Model Runner V2 Pipeline Parallelism"
  key: model-runner-v2-pipeline-parallelism-4-gpus
  timeout_in_minutes: 50
  working_dir: "/vllm-workspace/tests"
@@ -100,7 +100,7 @@ steps:
    - pytest -v -s distributed/test_pp_cudagraph.py -k "not ray"
    - pytest -v -s v1/distributed/test_pp_dp_v2.py

- label: Model Runner V2 Spec Decode
+- label: ":nvidia: (H200) Model Runner V2 Spec Decode"
  device: h200_35gb
  key: model-runner-v2-spec-decode
  timeout_in_minutes: 50
@@ -2,7 +2,7 @@ group: Models - Basic
 depends_on:
  - image-build
 steps:
- label: Basic Models Tests (Initialization)
+- label: ":nvidia: (H200) Basic Models (Initialization)"
  key: basic-models-tests-initialization
  timeout_in_minutes: 25
  device: h200_18gb
@@ -15,7 +15,7 @@ steps:
    # Run a subset of model initialization tests
    - pytest -v -s models/test_initialization.py::test_can_initialize_small_subset

- label: Basic Models Tests (Extra Initialization) %N
+- label: ":nvidia: (H200) Basic Models (Extra Initialization) Shard %N"
  device: h200_35gb
  key: basic-models-tests-extra-initialization
  timeout_in_minutes: 100
@@ -30,7 +30,7 @@ steps:
    - 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
  parallelism: 4

- label: Basic Models Tests (Other)
+- label: ":nvidia: (H200) Basic Models (Other)"
  device: h200_35gb
  key: basic-models-tests-other
  timeout_in_minutes: 35
@@ -44,13 +44,14 @@ steps:
    - pytest -v -s models/test_terratorch.py models/transformers/test_backend.py models/test_registry.py
  mirror:
    amd:
+      label: ":amd: (MI300) Basic Models (Other)"
      dind: false
      device: mi300_1
      timeout_in_minutes: 50
      depends_on:
      - image-build-amd

- label: Inkling Unit Tests (B200)
+- label: ":nvidia: (B200) Inkling"
  key: inkling-unit-tests-b200
  timeout_in_minutes: 40
  device: b200-k8s
@@ -63,7 +64,7 @@ steps:
    # FA4 kernel tests require SM100; the suite skips them elsewhere.
    - pytest -v -s models/inkling

- label: Kimi K3 Unit Tests (B200)
+- label: ":nvidia: (B200) Kimi K3"
  key: kimi-k3-unit-tests-b200
  timeout_in_minutes: 40
  device: b200-k8s
@@ -77,7 +78,7 @@ steps:
    # The native NVIDIA Kimi K3 kernels require the SM100 family.
    - pytest -v -s models/kimi_k3 kernels/attention/test_kimi_k3_mla_fused_epilogue.py kernels/test_bf16_skinny_gemm.py

- label: Basic Models Test (Other CPU) # 5min
+- label: ":computer: (CPU) Basic Models Other"
  key: basic-models-test-other-cpu
  depends_on:
  - image-build-cpu
@@ -2,7 +2,7 @@ group: Models - Distributed
 depends_on: 
  - image-build
 steps:
- label: Distributed Model Tests (2 GPUs)
+- label: ":nvidia: (L4) Distributed Models"
  key: distributed-model-tests-2-gpus
  timeout_in_minutes: 60
  working_dir: "/vllm-workspace/tests"
@@ -2,7 +2,7 @@ group: Models - Language
 depends_on:
  - image-build
 steps:
- label: Language Models Tests (Standard)
+- label: ":nvidia: (H200) Language Models (Standard)"
  key: language-models-tests-standard
  timeout_in_minutes: 30
  device: h200_18gb
@@ -16,13 +16,14 @@ steps:
    - pytest -v -s models/language -m 'core_model and (not slow_test)'
  mirror:
    amd:
+      label: ":amd: (MI300) Language Models (Standard)"
      dind: false
      device: mi300_1
      timeout_in_minutes: 45
      depends_on:
      - image-build-amd

- label: Language Models Tests (Extra Standard) %N
+- label: ":nvidia: (H200) Language Models (Extra Standard) Shard %N"
  device: h200_35gb
  key: language-models-tests-extra-standard
  timeout_in_minutes: 40
@@ -39,6 +40,7 @@ steps:
  parallelism: 2
  mirror:
    amd:
+      label: ":amd: (MI300) Language Models (Extra Standard) Shard %N"
      dind: false
      device: mi300_1
      timeout_in_minutes: 40
@@ -55,7 +57,7 @@ steps:
      - tests/models/language/pooling/test_classification.py
      - vllm/_aiter_ops.py
      - vllm/platforms/rocm.py
- label: Language Models Tests (Hybrid) %N
+- label: ":nvidia: (H200) Language Models (Hybrid) Shard %N"
  device: h200_35gb
  key: language-models-tests-hybrid
  timeout_in_minutes: 65
@@ -72,6 +74,7 @@ steps:
  parallelism: 2
  mirror:
    amd:
+      label: ":amd: (MI300) Language Models (Hybrid) Shard %N"
      dind: false
      device: mi300_1
      timeout_in_minutes: 60
@@ -85,7 +88,7 @@ steps:
 # Granite 4 hybrid generation is sensitive to hardware-specific Triton SSD
 # autotuning (https://github.com/vllm-project/vllm/issues/25194). Keep this one
 # correctness test on L4 until its H200 output matches the Transformers reference.
- label: Language Models Tests (Granite L4 Compatibility)
+- label: ":nvidia: (L4) Granite Language Model Compatibility"
  key: language-models-tests-granite-l4-compatibility
  timeout_in_minutes: 65
  source_file_dependencies:
@@ -97,7 +100,7 @@ steps:
    - CAUSAL_CONV1D_FORCE_BUILD=TRUE uv pip install --system --no-build-isolation 'git+https://github.com/Dao-AILab/causal-conv1d@v1.6.0'
    - pytest -v -s models/language/generation -m hybrid_model -k 'granite-4.0-tiny-preview'

- label: Language Models Test (Extended Generation) # 80min
+- label: ":nvidia: (H200) Language Models (Extended Generation)"
  device: h200_35gb
  key: language-models-test-extended-generation
  timeout_in_minutes: 80
@@ -112,7 +115,7 @@ steps:
    - CAUSAL_CONV1D_FORCE_BUILD=TRUE uv pip install --system --no-build-isolation 'git+https://github.com/Dao-AILab/causal-conv1d@v1.6.0'
    - pytest -v -s models/language/generation -m '(not core_model) and (not hybrid_model)'

- label: Language Models Test (PPL)
+- label: ":nvidia: (H200) Language Models (PPL)"
  key: language-models-test-ppl
  timeout_in_minutes: 30
  device: h200_18gb
@@ -124,7 +127,7 @@ steps:
  commands:
    - pytest -v -s models/language/generation_ppl_test

- label: Language Models Test (Extended Pooling) %N
+- label: ":nvidia: (H200) Language Models (Extended Pooling) Shard %N"
  device: h200_35gb
  key: language-models-test-extended-pooling
  timeout_in_minutes: 120
@@ -138,6 +141,7 @@ steps:
    - pytest -v -s models/language/pooling -m 'not core_model' --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --shard-id=$$BUILDKITE_PARALLEL_JOB
  mirror:
    amd:
+      label: ":amd: (MI300) Language Models (Extended Pooling) Shard %N"
      dind: false
      device: mi300_1
      timeout_in_minutes: 95
@@ -147,7 +151,7 @@ steps:
      commands:
      - pytest -v -s models/language/pooling -m 'not core_model' --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --shard-id=$$BUILDKITE_PARALLEL_JOB

- label: Language Models Test (MTEB)
+- label: ":nvidia: (H200) Language Models (MTEB)"
  key: language-models-test-mteb
  timeout_in_minutes: 68
  device: h200_18gb
@@ -2,7 +2,7 @@ group: Models - Multimodal
 depends_on: 
  - image-build
 steps:
- label: "Multi-Modal Models (Standard) 1: qwen2"
+- label: ":nvidia: (H200) Multimodal Models (Standard) 1: qwen2"
  key: multi-modal-models-standard-1-qwen2
  timeout_in_minutes: 68
  device: h200_18gb
@@ -15,13 +15,14 @@ steps:
    - pytest -v -s models/multimodal/generation/test_ultravox.py -m core_model
  mirror:
    amd:
+      label: ":amd: (MI300) Multimodal Models (Standard) 1: qwen2"
      dind: false
      device: mi300_1
      timeout_in_minutes: 65
      depends_on:
      - image-build-amd

- label: "Multi-Modal Models (Standard) 2: qwen3 + gemma"
+- label: ":nvidia: (H200) Multimodal Models (Standard) 2: qwen3 + gemma"
  key: multi-modal-models-standard-2-qwen3-gemma
  timeout_in_minutes: 75
  device: h200_18gb
@@ -35,13 +36,14 @@ steps:
    - pytest -v -s models/multimodal/generation/test_qwen2_5_vl.py -m core_model
  mirror:
    amd:
+      label: ":amd: (MI300) Multimodal Models (Standard) 2: qwen3 + gemma"
      dind: false
      device: mi300_1
      timeout_in_minutes: 55
      depends_on:
      - image-build-amd

- label: "Multi-Modal Models (Standard) 3: llava + qwen2_vl"
+- label: ":nvidia: (H200) Multimodal Models (Standard) 3: llava + qwen2_vl"
  device: h200_35gb
  key: multi-modal-models-standard-3-llava-qwen2-vl
  timeout_in_minutes: 40
@@ -54,12 +56,13 @@ steps:
    - pytest -v -s models/multimodal/generation/test_qwen2_vl.py -m core_model
  mirror:
    amd:
+      label: ":amd: (MI250) Multimodal Models (Standard) 3: llava + qwen2_vl"
      device: mi250_1
      timeout_in_minutes: 55
      depends_on:
      - image-build-amd

- label: "Multi-Modal Models (Standard) 4: other + whisper"
+- label: ":nvidia: (H200) Multimodal Models (Standard) 4: other + whisper"
  device: h200_35gb
  key: multi-modal-models-standard-4-other-whisper
  timeout_in_minutes: 75
@@ -74,13 +77,14 @@ steps:
    - cd .. && VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s tests/models/multimodal/generation/test_whisper.py -m core_model  # Otherwise, mp_method="spawn" doesn't work
  mirror:
    amd:
+      label: ":amd: (MI300) Multimodal Models (Standard) 4: other + whisper"
      dind: false
      device: mi300_1
      timeout_in_minutes: 50
      depends_on:
      - image-build-amd

- label: Multi-Modal Processor (CPU) %N
+- label: ":computer: (CPU) Multimodal Processor Shard %N"
  key: multi-modal-processor-cpu
  depends_on:
  - image-build-cpu
@@ -95,7 +99,7 @@ steps:
    - pytest -v -s models/multimodal/processing --ignore models/multimodal/processing/test_tensor_schema.py --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --shard-id=$$BUILDKITE_PARALLEL_JOB
  parallelism: 4

- label: Multi-Modal Processor # 44min
+- label: ":nvidia: (H200) Multimodal Processor"
  key: multi-modal-processor
  timeout_in_minutes: 98
  device: h200_18gb
@@ -107,7 +111,7 @@ steps:
  commands:
    - pytest -v -s models/multimodal/processing/test_tensor_schema.py

- label: Multi-Modal Accuracy Eval (Small Models) # 50min
+- label: ":nvidia: (H200) Multimodal Accuracy Eval (Small Models)"
  device: h200_35gb
  key: multi-modal-accuracy-eval-small-models
  timeout_in_minutes: 30
@@ -120,6 +124,7 @@ steps:
  - pytest -s -v test_lm_eval_correctness.py --config-list-file=configs/models-mm-small.txt --tp-size=1
  mirror:
    amd:
+      label: ":amd: (MI300) Multimodal Accuracy Eval (Small Models)"
      dind: false
      device: mi300_1
      timeout_in_minutes: 35
@@ -132,7 +137,7 @@ steps:
      - vllm/platforms/rocm.py
      - vllm/model_executor/model_loader/

- label: Multi-Modal Models (Extended Generation 1)
+- label: ":nvidia: (H200) Multimodal Models (Extended Generation 1)"
  device: h200_35gb
  key: multi-modal-models-extended-generation-1
  optional: true
@@ -146,13 +151,14 @@ steps:
    - pytest -v -s models/multimodal/test_mapping.py
  mirror:
    amd:
+      label: ":amd: (MI300) Multimodal Models (Extended Generation 1)"
      dind: false
      device: mi300_1
      timeout_in_minutes: 90
      depends_on:
      - image-build-amd

- label: Multi-Modal Models (PPL)
+- label: ":nvidia: (H200) Multimodal Models (PPL)"
  device: h200_35gb
  key: multi-modal-models-extended-ppl
  optional: true
@@ -163,13 +169,14 @@ steps:
    - pytest -v -s models/multimodal/generation_ppl_test/
  mirror:
    amd:
+      label: ":amd: (MI300) Multimodal Models (PPL)"
      dind: false
      device: mi300_1
      timeout_in_minutes: 90
      depends_on:
      - image-build-amd

- label: Multi-Modal Models (Extended Generation 2) %N
+- label: ":nvidia: (H200) Multimodal Models (Extended Generation 2) Shard %N"
  device: h200_35gb
  key: multi-modal-models-extended-generation-2
  parallelism: 4
@@ -181,7 +188,7 @@ steps:
  commands:
    - pytest -v -s models/multimodal/generation/test_common.py -m 'split(group=0) and not core_model' --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --shard-id=$$BUILDKITE_PARALLEL_JOB

- label: Multi-Modal Models (Extended Generation 3)
+- label: ":nvidia: (H200) Multimodal Models (Extended Generation 3)"
  device: h200_35gb
  key: multi-modal-models-extended-generation-3
  optional: true
@@ -192,7 +199,7 @@ steps:
  commands:
    - pytest -v -s models/multimodal/generation/test_common.py -m 'split(group=1) and not core_model'

- label: Multi-Modal Models (Extended Pooling)
+- label: ":nvidia: (H200) Multimodal Models (Extended Pooling)"
  key: multi-modal-models-extended-pooling
  optional: true
  device: h200_18gb
@@ -204,6 +211,7 @@ steps:
    - pytest -v -s models/multimodal/pooling -m 'not core_model'
  mirror:
    amd:
+      label: ":amd: (MI300) Multimodal Models (Extended Pooling)"
      dind: false
      device: mi300_1
      timeout_in_minutes: 60
@@ -2,7 +2,7 @@ group: Plugins
 depends_on: 
  - image-build
 steps:
- label: Plugin Tests (2 GPUs)
+- label: ":nvidia: (L4) Plugin Integration"
  key: plugin-tests-2-gpus
  timeout_in_minutes: 35
  working_dir: "/vllm-workspace/tests"
@@ -51,7 +51,7 @@ steps:
  - pytest -v -s plugins_tests/lora_resolvers # unit tests for in-tree lora resolver plugins


- label: GGUF Plugin
+- label: ":nvidia: (H200) GGUF Plugin"
  key: gguf-plugin
  device: h200_18gb
  timeout_in_minutes: 30
@@ -64,7 +64,7 @@ steps:
    - pip install "vllm-gguf-plugin >= 0.0.2"
    - pytest -v -s plugins_tests/gguf

- label: BitsAndBytes Plugin
+- label: ":nvidia: (H200) BitsAndBytes Plugin"
  key: bitsandbytes-plugin
  device: h200_18gb
  timeout_in_minutes: 30
@@ -81,7 +81,7 @@ steps:
    - pip install "vllm-bnb-plugin >= 0.0.1"
    - pytest -v -s plugins_tests/bitsandbytes -m 'not distributed'

- label: BitsAndBytes Plugin (2 GPUs)
+- label: ":nvidia: (L4) BitsAndBytes Plugin"
  key: bitsandbytes-plugin-2-gpus
  timeout_in_minutes: 15
  working_dir: "/vllm-workspace/tests"
@@ -2,7 +2,7 @@ group: PyTorch
 depends_on: 
  - image-build
 steps:
- label: PyTorch Compilation Unit Tests
+- label: ":nvidia: (H200) PyTorch Compilation"
  device: h200_35gb
  key: pytorch-compilation-unit-tests
  timeout_in_minutes: 60
@@ -43,7 +43,7 @@ steps:
  # (using -0 for proper path handling)
  - "find compile/ -maxdepth 1 -name 'test_*.py' -print0 | xargs -0 -n1 -I{} pytest -s -v '{}'"

- label: PyTorch Compilation Unit Tests (H100)
+- label: ":nvidia: (H200) PyTorch Compilation H100 Cases"
  key: pytorch-compilation-unit-tests-h100
  timeout_in_minutes: 30
  device: h200_18gb
@@ -77,7 +77,7 @@ steps:
  commands:
  - "find compile/h100/ -name 'test_*.py' -print0 | xargs -0 -n1 -I{} pytest -s -v '{}'"

- label: PyTorch Compilation Passes Unit Tests
+- label: ":nvidia: (L4) PyTorch Compilation Passes"
  key: pytorch-compilation-passes-unit-tests
  timeout_in_minutes: 45
  source_file_dependencies:
@@ -110,7 +110,7 @@ steps:
  commands:
  - pytest -s -v compile/passes --ignore compile/passes/distributed

- label: PyTorch Fullgraph Test
+- label: ":nvidia: (H200) PyTorch Fullgraph"
  device: h200_35gb
  key: pytorch-fullgraph-test
  timeout_in_minutes: 90
@@ -151,7 +151,7 @@ steps:
 # Hopper-only DeepSeek-V2-Lite cases in this file require two 29.3-GiB model
 # instances and cannot fit a 35GB MIG slice. L4 retains the original coverage:
 # those SM90 cases skip while the architecture-compatible cases still run.
- label: PyTorch Fullgraph CUDAGraph (L4 Compatibility)
+- label: ":nvidia: (L4) PyTorch Fullgraph CUDAGraph Compatibility"
  key: pytorch-fullgraph-cudagraph-l4-compatibility
  timeout_in_minutes: 60
  source_file_dependencies:
@@ -184,7 +184,7 @@ steps:
  commands:
  - pytest -s -v compile/fullgraph/test_full_cudagraph.py

- label: Pytorch Nightly Dependency Override Check # 2min
+- label: ":nvidia: (H200) PyTorch Nightly Dependency Override Check"
  key: pytorch-nightly-dependency-override-check
  # if this test fails, it means the nightly torch version is not compatible with some
  # of the dependencies. Please check the error message and add the package to whitelist
@@ -197,6 +197,7 @@ steps:
  - bash standalone_tests/pytorch_nightly_dependency.sh
  mirror:
    amd:
+      label: ":amd: (MI300) PyTorch Nightly Dependency Override Check"
      dind: false
      device: mi300_1
      timeout_in_minutes: 30
@@ -2,7 +2,7 @@ group: Quantization
 depends_on:
  - image-build
 steps:
- label: Quantization %N
+- label: ":nvidia: (H200) Quantization Shard %N"
  device: h200_35gb
  key: quantization
  timeout_in_minutes: 40
@@ -21,7 +21,7 @@ steps:
  # parameter. It was not exercised by the previous L4 job.
  - VLLM_TEST_FORCE_LOAD_FORMAT=auto pytest -v -s quantization/ --ignore quantization/test_blackwell_moe.py -k 'not test_compressed_tensors_w4a8_fp8' --shard-id=$$BUILDKITE_PARALLEL_JOB --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT

- label: Quantized Fusions
+- label: ":nvidia: (H200) Quantized Fusions"
  device: h200_35gb
  key: quantized-fusions
  timeout_in_minutes: 20
@@ -34,7 +34,7 @@ steps:
  commands:
    - pytest -v -s fusion/

- label: Quantized MoE Test (B200)
+- label: ":nvidia: (B200) Quantized MoE"
  key: quantized-moe-test-b200
  timeout_in_minutes: 120
  working_dir: "/vllm-workspace/"
@@ -52,7 +52,7 @@ steps:
  commands:
    - pytest -s -v tests/quantization/test_blackwell_moe.py

- label: Quantized Models Test
+- label: ":nvidia: (H200) Quantized Models"
  device: h200_35gb
  key: quantized-models-test
  timeout_in_minutes: 65
@@ -2,7 +2,7 @@ group: Ray Compatibility
 depends_on:
  - image-build
 steps:
- label: Ray Dependency Compatibility Check
+- label: ":nvidia: (H200) Ray Dependency Compatibility Check"
  key: ray-dependency-compatibility-check
  # Informational only — does not block the pipeline.
  # If this fails, it means the PR introduces a dependency that
@@ -2,7 +2,7 @@ group: Rust Frontend E2E
 depends_on:
  - image-build
 steps:
- label: Rust Frontend OpenAI Coverage
+- label: ":nvidia: (H200) Rust Frontend OpenAI Coverage"
  timeout_in_minutes: 30
  device: h200_18gb
  working_dir: "/vllm-workspace/tests"
@@ -37,7 +37,7 @@ steps:
  - pytest -v -s entrypoints/openai/test_uds.py
  - pytest -v -s v1/sample/test_logprobs_e2e.py -k "test_prompt_logprobs_e2e_server"

- label: Rust Frontend Serve/Admin Coverage
+- label: ":nvidia: (H200) Rust Frontend Serve/Admin Coverage"
  timeout_in_minutes: 25
  device: h200_18gb
  working_dir: "/vllm-workspace/tests"
@@ -66,7 +66,7 @@ steps:
  # /tokenizer_info is not implemented in the Rust frontend (the CLI flag is accepted as a no-op).
  - pytest -v -s entrypoints/serve/tokenize/test_tokenization.py -k "not tokenizer_info"

- label: Rust Frontend Core Correctness
+- label: ":nvidia: (H200) Rust Frontend Core Correctness"
  timeout_in_minutes: 20
  device: h200_18gb
  working_dir: "/vllm-workspace/tests"
@@ -80,7 +80,7 @@ steps:
  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
  - pytest -s entrypoints/openai/correctness/test_lmeval.py::test_lm_eval_accuracy_v1_engine

- label: Rust Frontend Tool Use
+- label: ":nvidia: (H200) Rust Frontend Tool Use"
  device: h200_35gb
  timeout_in_minutes: 25
  working_dir: "/vllm-workspace/tests"
@@ -95,7 +95,7 @@ steps:
  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
  - pytest -v -s tool_use --ignore=tool_use/mistral --models llama3.2 -k "not test_response_format_with_tool_choice_required and not test_parallel_tool_calls_false and not test_tool_call_and_choice"

- label: Rust Frontend Distributed
+- label: ":nvidia: (L4) Rust Frontend Distributed"
  timeout_in_minutes: 25
  num_devices: 4
  working_dir: "/vllm-workspace/tests"
@@ -1,7 +1,7 @@
 group: Rust Frontend Cargo
 depends_on: []
 steps:
- label: Rust Frontend Cargo Style + Clippy
+- label: ":computer: (CPU) Rust Frontend Cargo Style + Clippy"
  key: rust-frontend-cargo-style-clippy
  depends_on: []
  timeout_in_minutes: 20
@@ -15,7 +15,7 @@ steps:
  commands:
  - .buildkite/scripts/run-rust-frontend-cargo-ci.sh style-clippy

- label: Rust Frontend Cargo Tests
+- label: ":computer: (CPU) Rust Frontend Cargo"
  key: rust-frontend-cargo-tests
  depends_on: []
  timeout_in_minutes: 20
@@ -2,7 +2,7 @@ group: Samplers
 depends_on: 
  - image-build
 steps:
- label: Samplers Test
+- label: ":nvidia: (H200) Samplers"
  device: h200_35gb
  key: samplers-test
  timeout_in_minutes: 40
@@ -19,6 +19,7 @@ steps:
    - VLLM_USE_FLASHINFER_SAMPLER=1 pytest -v -s samplers
  mirror:
    amd:
+      label: ":amd: (MI250) Samplers"
      device: mi250_1
      timeout_in_minutes: 40
      depends_on:
@@ -2,7 +2,7 @@ group: Spec Decode
 depends_on:
  - image-build
 steps:
- label: V1 Spec Decode
+- label: ":nvidia: (H200) V1 Spec Decode"
  device: h200_35gb
  key: v1-spec-decode
  timeout_in_minutes: 40
@@ -23,13 +23,14 @@ steps:
    - pytest -v -s -m 'not slow_test' v1/spec_decode
  mirror:
    amd:
+      label: ":amd: (MI300) V1 Spec Decode"
      dind: false
      device: mi300_1
      timeout_in_minutes: 50
      depends_on:
      - image-build-amd

- label: "Spec Decode Eagle 1: DeepSeek + Qwen"
+- label: ":nvidia: (H200) Spec Decode Eagle 1: DeepSeek + Qwen"
  key: spec-decode-eagle-1-deepseek-qwen
  timeout_in_minutes: 25
  device: h200_35gb
@@ -41,6 +42,7 @@ steps:
    - pytest -v -s v1/e2e/spec_decode/eagle/ -k "deepseek_eagle or qwen3_eagle3"
  mirror:
    amd:
+      label: ":amd: (MI300) Spec Decode Eagle 1: DeepSeek + Qwen"
      dind: false
      device: mi300_1
      timeout_in_minutes: 25
@@ -55,7 +57,7 @@ steps:
      - tests/v1/e2e/spec_decode/
      - vllm/platforms/rocm.py

- label: "Spec Decode Eagle 2: Llama 3 + Qwen VL + Other"
+- label: ":nvidia: (H200) Spec Decode Eagle 2: Llama 3 + Qwen VL + Other"
  key: spec-decode-eagle-2-llama3-qwen-vl-other
  timeout_in_minutes: 25
  device: h200_35gb
@@ -67,6 +69,7 @@ steps:
    - pytest -v -s v1/e2e/spec_decode/eagle/ -k "not deepseek_eagle and not qwen3_eagle3"
  mirror:
    amd:
+      label: ":amd: (MI300) Spec Decode Eagle 2: Llama 3 + Qwen VL + Other"
      dind: false
      device: mi300_1
      timeout_in_minutes: 25
@@ -81,7 +84,7 @@ steps:
      - tests/v1/e2e/spec_decode/
      - vllm/platforms/rocm.py

- label: Spec Decode Eagle Nightly (B200)
+- label: ":nvidia: (B200) Spec Decode Eagle Nightly"
  key: spec-decode-eagle-nightly-b200
  timeout_in_minutes: 25
  device: b200-k8s
@@ -93,7 +96,7 @@ steps:
  commands:
    - pytest -v -s v1/e2e/spec_decode/eagle/

- label: Spec Decode Speculators + MTP
+- label: ":nvidia: (H200) Spec Decode Speculators + MTP"
  key: spec-decode-speculators-mtp
  timeout_in_minutes: 50
  device: h200_35gb
@@ -108,6 +111,7 @@ steps:
    - pytest -v -s v1/e2e/spec_decode/mtp/
  mirror:
    amd:
+      label: ":amd: (MI300) Spec Decode Speculators + MTP"
      dind: false
      device: mi300_1
      timeout_in_minutes: 75
@@ -123,7 +127,7 @@ steps:
      - tests/v1/e2e/spec_decode/
      - vllm/platforms/rocm.py

- label: Spec Decode Speculators + MTP Nightly (B200)
+- label: ":nvidia: (B200) Spec Decode Speculators + MTP Nightly"
  key: spec-decode-speculators-mtp-nightly-b200
  timeout_in_minutes: 30
  device: b200-k8s
@@ -137,7 +141,7 @@ steps:
    - pytest -v -s v1/e2e/spec_decode/speculators/
    - pytest -v -s v1/e2e/spec_decode/mtp/

- label: Spec Decode Ngram + Suffix
+- label: ":nvidia: (H200) Spec Decode N-Gram + Suffix"
  key: spec-decode-ngram-suffix
  timeout_in_minutes: 20
  device: h200_35gb
@@ -151,6 +155,7 @@ steps:
    - python3 spec_decode/test_custom_proposer.py
  mirror:
    amd:
+      label: ":amd: (MI300) Spec Decode N-Gram + Suffix"
      dind: false
      device: mi300_1
      timeout_in_minutes: 35
@@ -165,7 +170,7 @@ steps:
      - tests/v1/e2e/spec_decode/
      - vllm/platforms/rocm.py

- label: Spec Decode Draft Model
+- label: ":nvidia: (H200) Spec Decode Draft Model"
  key: spec-decode-draft-model
  timeout_in_minutes: 45
  device: h200_18gb
@@ -177,6 +182,7 @@ steps:
    - pytest -v -s v1/e2e/spec_decode/draft_model/
  mirror:
    amd:
+      label: ":amd: (MI300) Spec Decode Draft Model"
      dind: false
      device: mi300_1
      timeout_in_minutes: 55
@@ -191,7 +197,7 @@ steps:
      - tests/v1/e2e/spec_decode/
      - vllm/platforms/rocm.py

- label: Spec Decode Draft Model Nightly (B200)
+- label: ":nvidia: (B200) Spec Decode Draft Model Nightly"
  key: spec-decode-draft-model-nightly-b200
  timeout_in_minutes: 40
  device: b200-k8s
@@ -203,7 +209,7 @@ steps:
  commands:
    - pytest -v -s v1/e2e/spec_decode/draft_model/

- label: Speculators Correctness Nightly (H100)
+- label: ":nvidia: (H100) Speculators Correctness Nightly"
  key: speculators-correctness
  timeout_in_minutes: 30
  device: h100
@@ -218,7 +224,7 @@ steps:
    - export VLLM_ALLOW_INSECURE_SERIALIZATION=1
    - pytest -v -s v1/spec_decode/test_speculators_correctness.py -m slow_test

- label: Spec Decode DeepSeek MTP Parallel Load (2xB200-2xMI300)
+- label: ":nvidia: (B200) Spec Decode DeepSeek MTP Parallel Load"
  key: spec-decode-deepseek-mtp-parallel-load-2xb200-2xmi300
  timeout_in_minutes: 30
  device: b200-k8s
@@ -235,6 +241,7 @@ steps:
    - pytest -v -s v1/e2e/spec_decode/test_mtp_parallel_load.py
  mirror:
    amd:
+      label: ":amd: (MI300) Spec Decode DeepSeek MTP Parallel Load"
      dind: false
      device: mi300_2
      timeout_in_minutes: 45
@@ -254,7 +261,7 @@ steps:
      - tests/v1/e2e/spec_decode/test_mtp_parallel_load.py
      - vllm/platforms/rocm.py

- label: Spec Decode AL DFlash Nightly
+- label: ":nvidia: (H200) Spec Decode AL DFlash Nightly"
  key: spec-decode-dflash-nightly
  timeout_in_minutes: 90
  device: h200_35gb
@@ -269,6 +276,7 @@ steps:
    - pytest -v -s v1/e2e/spec_decode/acceptance_rates/dflash/
  mirror:
    amd:
+      label: ":amd: (MI300) Spec Decode AL DFlash Nightly"
      dind: false
      device: mi300_1
      timeout_in_minutes: 120
@@ -289,7 +297,7 @@ steps:
      - tests/v1/e2e/spec_decode/
      - vllm/platforms/rocm.py

- label: Spec Decode AL DSpark Nightly
+- label: ":nvidia: (H200) Spec Decode AL DSpark Nightly"
  key: spec-decode-dspark-nightly
  timeout_in_minutes: 60
  device: h200_35gb
@@ -304,6 +312,7 @@ steps:
    - pytest -v -s v1/e2e/spec_decode/acceptance_rates/dspark/
  mirror:
    amd:
+      label: ":amd: (MI300) Spec Decode AL DSpark Nightly"
      dind: false
      device: mi300_1
      timeout_in_minutes: 90
@@ -328,7 +337,7 @@ steps:
      - tests/v1/e2e/spec_decode/
      - vllm/platforms/rocm.py

- label: Spec Decode AL MTP + Other Acceptance Nightly
+- label: ":nvidia: (H200) Spec Decode AL MTP + Other Acceptance Nightly"
  key: spec-decode-mtp-other-acceptance-nightly
  timeout_in_minutes: 60
  device: h200_35gb
@@ -2,7 +2,7 @@ group: Torch ABI
 depends_on:
  - image-build
 steps:
- label: Torch Stable ABI Audit
+- label: ":nvidia: (L4) Torch Stable ABI Audit"
  key: torch-stable-abi-audit
  timeout_in_minutes: 5
  source_file_dependencies:
@@ -2,7 +2,7 @@ group: Weight Loading
 depends_on: 
  - image-build
 steps:
- label: Weight Loading Multiple GPU  # 33min
+- label: ":nvidia: (L4) Weight Loading Multi-GPU"
  key: weight-loading-multiple-gpu
  timeout_in_minutes: 50
  working_dir: "/vllm-workspace/tests"
@@ -16,6 +16,7 @@ steps:
    - bash weight_loading/run_model_weight_loading_test.sh -c weight_loading/models.txt
  mirror:
    amd:
+      label: ":amd: (MI300) Weight Loading Multi-GPU"
      dind: false
      device: mi300_2
      timeout_in_minutes: 35