[ROCm][CI] Stage F gating (#55252)

Signed-off-by: Andreas Karatzas <akaratza@amd.com>
Signed-off-by: Andreas Karatzas <Andreas.Karatzas@amd.com>
Co-authored-by: OpenAI Codex <codex@openai.com>

.buildkite/test_areas/disaggregated.yaml

@@ -172,6 +172,23 @@ steps:
   commands:
     - bash /vllm-workspace/.buildkite/scripts/install-kv-connectors.sh
     - bash v1/kv_connector/nixl_integration/run_mamba_prefix_cache_test.sh
+  mirror:
+    amd:
+      label: ":amd: (MI355) Hybrid SSM NixlConnector PD prefix cache"
+      dind: false
+      device: mi355_2
+      timeout_in_minutes: 25
+      depends_on:
+      - image-build-amd
+      source_file_dependencies:
+      - vllm/distributed/kv_transfer/kv_connector/v1/nixl/
+      - vllm/v1/core/sched/
+      - vllm/v1/core/kv_cache_coordinator.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/run_mamba_prefix_cache_test.sh
 
 - label: ":nvidia: (L4) MultiConnector (Nixl+Offloading) PD accuracy"
   key: multiconnector-nixl-offloading-pd-accuracy-2-gpus
@@ -234,6 +251,24 @@ steps:
   commands:
     - bash /vllm-workspace/.buildkite/scripts/install-kv-connectors.sh
     - bash v1/kv_connector/nixl_integration/run_multi_connector_edge_case_test.sh
+  mirror:
+    amd:
+      label: ":amd: (MI355) MultiConnector (Nixl+Offloading) PD edge cases"
+      dind: false
+      device: mi355_2
+      timeout_in_minutes: 25
+      depends_on:
+      - image-build-amd
+      source_file_dependencies:
+      - vllm/distributed/kv_transfer/kv_connector/v1/nixl/
+      - vllm/distributed/kv_transfer/kv_connector/v1/multi_connector.py
+      - vllm/distributed/kv_transfer/kv_connector/v1/offloading_connector.py
+      - vllm/distributed/kv_transfer/kv_connector/v1/offloading/
+      - 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/run_multi_connector_edge_case_test.sh
 
 # P TP 4 - D DPEP 4 test case for DSv4-Flash
 - label: ":nvidia: (H200) DSv4-Flash Disaggregated DP EP"

.buildkite/test_areas/distributed.yaml

@@ -105,6 +105,26 @@ steps:
   - export NCCL_CUMEM_HOST_ENABLE=0
   - pytest -v -s entrypoints/llm/test_collective_rpc.py
   - pytest -v -s ./compile/test_wrapper.py
+  mirror:
+    amd:
+      label: ":amd: (MI355) Distributed Compile + RPC"
+      dind: false
+      device: mi355_2
+      timeout_in_minutes: 65
+      depends_on:
+      - image-build-amd
+      source_file_dependencies:
+      - vllm/compilation/
+      - vllm/distributed/
+      - "!vllm/distributed/kv_transfer/"
+      - vllm/engine/
+      - vllm/executor/
+      - vllm/worker/worker_base.py
+      - vllm/v1/engine/
+      - vllm/v1/worker/
+      - tests/compile/test_wrapper.py
+      - tests/entrypoints/llm/test_collective_rpc.py
+      - vllm/platforms/rocm.py
 
 - label: ":nvidia: (L4) Distributed Torchrun + Shutdown"
   key: distributed-torchrun-shutdown-tests-2-gpus
@@ -335,6 +355,25 @@ steps:
   - pytest -v -s distributed/test_pp_cudagraph.py
   - pytest -v -s distributed/test_pipeline_parallel.py
   - pytest -v -s v1/distributed/test_pp_dp_v2.py
+  mirror:
+    amd:
+      label: ":amd: (MI250) Pipeline + Context Parallelism"
+      device: mi250_4
+      timeout_in_minutes: 90
+      depends_on:
+      - image-build-amd
+      source_file_dependencies:
+      - vllm/distributed/
+      - "!vllm/distributed/kv_transfer/"
+      - vllm/engine/
+      - vllm/executor/
+      - vllm/model_executor/models/
+      - vllm/model_executor/layers/
+      - vllm/v1/attention/backends/
+      - vllm/v1/attention/selector.py
+      - tests/distributed/
+      - vllm/_aiter_ops.py
+      - vllm/platforms/rocm.py
 
 - label: ":nvidia: (L4) RayExecutorV2"
   key: rayexecutorv2-4-gpus

.buildkite/test_areas/lm_eval.yaml

@@ -79,6 +79,26 @@ steps:
   autorun_on_main: true
   commands:
   - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=configs/models-small-tp.txt
+  mirror:
+    amd:
+      label: ":amd: (MI355) LM Eval Small Models Distributed"
+      dind: false
+      device: mi355_2
+      timeout_in_minutes: 45
+      depends_on:
+      - image-build-amd
+      source_file_dependencies:
+      - csrc/
+      - vllm/model_executor/layers/quantization
+      - vllm/model_executor/models/
+      - vllm/model_executor/model_loader/
+      - vllm/v1/attention/backends/
+      - vllm/v1/attention/selector.py
+      - tests/evals/gsm8k/configs/models-mi3xx-fp8-and-mixed.txt
+      - vllm/_aiter_ops.py
+      - vllm/platforms/rocm.py
+      commands:
+      - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=configs/models-mi3xx-fp8-and-mixed.txt
 
 - label: ":nvidia: (B200) LM Eval PCP"
   key: lm-eval-pcp-4xb200
@@ -375,6 +395,28 @@ steps:
   commands:
     - 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
+  mirror:
+    amd:
+      label: ":amd: (MI355) GPQA Eval (GPT-OSS)"
+      dind: false
+      device: mi355_2
+      timeout_in_minutes: 40
+      depends_on:
+      - image-build-amd
+      source_file_dependencies:
+      - csrc/
+      - vllm/model_executor/layers/quantization
+      - vllm/model_executor/models/
+      - vllm/model_executor/model_loader/
+      - vllm/v1/attention/backends/
+      - vllm/v1/attention/selector.py
+      - vllm/model_executor/layers/fused_moe/
+      - tests/evals/gpt_oss/
+      - vllm/_aiter_ops.py
+      - vllm/platforms/rocm.py
+      commands:
+      - 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-gfx950.txt
 
 - label: ":nvidia: (B200) Qwen3.8-Flash-Next-FP8 Accuracy Eval"
   key: accuracy-eval-qwen3-8-flash-next-fp8-b200

.buildkite/test_areas/misc.yaml

@@ -251,6 +251,35 @@ steps:
   commands:
     - export VLLM_WORKER_MULTIPROC_METHOD=spawn
     - pytest -v -s -m 'distributed' tests/v1/kv_connector/extract_hidden_states_integration
+  mirror:
+    amd:
+      label: ":amd: (MI355) Extract Hidden States Integration"
+      dind: false
+      device: mi355_2
+      timeout_in_minutes: 20
+      depends_on:
+      - image-build-amd
+      source_file_dependencies:
+      - vllm/config/speculative.py
+      - vllm/distributed/kv_transfer/kv_connector/
+      - vllm/model_executor/layers/attention/
+      - vllm/model_executor/layers/mamba/
+      - vllm/model_executor/model_loader/
+      - vllm/model_executor/models/extract_hidden_states.py
+      - vllm/model_executor/models/llama.py
+      - vllm/model_executor/models/qwen3_5.py
+      - vllm/model_executor/models/qwen3_next.py
+      - vllm/model_executor/models/registry.py
+      - vllm/transformers_utils/configs/extract_hidden_states.py
+      - vllm/transformers_utils/configs/qwen3_5.py
+      - vllm/v1/attention/backends/
+      - vllm/v1/attention/selector.py
+      - vllm/v1/kv_cache_interface.py
+      - vllm/v1/spec_decode/extract_hidden_states.py
+      - vllm/v1/worker/gpu_model_runner.py
+      - vllm/_aiter_ops.py
+      - tests/v1/kv_connector/extract_hidden_states_integration/
+      - vllm/platforms/rocm.py
 
 - label: ":nvidia: (H200 MIG 18GB) Regression"
   key: regression
@@ -537,6 +566,24 @@ steps:
     - pytest -v -s v1/determinism/test_rms_norm_batch_invariant.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
     - 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
+  mirror:
+    amd:
+      label: ":amd: (MI250) Batch Invariance"
+      device: mi250_1
+      timeout_in_minutes: 30
+      depends_on:
+      - image-build-amd
+      source_file_dependencies:
+      - vllm/v1/attention
+      - vllm/model_executor/layers
+      - tests/v1/determinism/
+      - vllm/_aiter_ops.py
+      - vllm/platforms/rocm.py
+      commands:
+      - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+      - pip install pytest-timeout pytest-forked
+      - pytest -v -s v1/determinism/test_batch_invariance.py
+      - pytest -v -s v1/determinism/test_rms_norm_batch_invariant.py
 
 - label: ":nvidia: (B200) Batch Invariance"
   key: batch-invariance-b200

.buildkite/test_areas/models_basic.yaml

@@ -156,3 +156,12 @@ steps:
   device: cpu-small
   commands:
     - pytest -v -s models/test_utils.py models/test_vision.py models/test_adapters.py models/test_qwen3_5_mtp_config.py models/transformers/fusers/
+  mirror:
+    amd:
+      label: ":amd: (MI250) Basic Models Other"
+      device: mi250_1
+      no_gpu: true
+      working_dir: "/vllm-workspace/tests"
+      timeout_in_minutes: 25
+      depends_on:
+      - image-build-amd

.buildkite/test_areas/models_multimodal.yaml

@@ -98,6 +98,15 @@ steps:
   commands:
     - 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
+  mirror:
+    amd:
+      label: ":amd: (MI250) Multimodal Processor Shard %N"
+      device: mi250_1
+      no_gpu: true
+      working_dir: "/vllm-workspace/tests"
+      timeout_in_minutes: 80
+      depends_on:
+      - image-build-amd
 
 - label: ":nvidia: (H200 MIG 18GB) Multimodal Processor"
   key: multi-modal-processor

.buildkite/test_areas/pytorch.yaml

@@ -160,6 +160,14 @@ steps:
   # Use `find` to launch multiple instances of pytest so that
   # they do not suffer from https://github.com/vllm-project/vllm/issues/28965
   - "find compile/fullgraph/ -name 'test_*.py' -not -name 'test_full_cudagraph.py' -print0 | xargs -0 -n1 -I{} pytest -s -v '{}'"
+  mirror:
+    amd:
+      label: ":amd: (MI355) PyTorch Fullgraph"
+      dind: false
+      device: mi355_1
+      timeout_in_minutes: 90
+      depends_on:
+      - image-build-amd
 
 # 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:

.buildkite/test_areas/rust_frontend.yaml

@@ -131,6 +131,20 @@ steps:
   - export VLLM_USE_RUST_FRONTEND=1
   - export VLLM_WORKER_MULTIPROC_METHOD=spawn
   - pytest -s entrypoints/openai/correctness/test_lmeval.py::test_lm_eval_accuracy_v1_engine
+  mirror:
+    amd:
+      label: ":amd: (MI355) Rust Frontend Core Correctness"
+      dind: false
+      device: mi355_1
+      timeout_in_minutes: 20
+      depends_on:
+      - image-build-amd
+      source_file_dependencies:
+      - rust/
+      - vllm/entrypoints/openai/
+      - tests/utils.py
+      - tests/entrypoints/openai/correctness/test_lmeval.py
+      - vllm/platforms/rocm.py
 
 - label: ":nvidia: (H200 MIG 35GB) Rust Frontend Tool Use"
   key: rust-frontend-tool-use
@@ -147,6 +161,21 @@ steps:
   - export VLLM_USE_RUST_FRONTEND=1
   - 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"
+  mirror:
+    amd:
+      label: ":amd: (MI355) Rust Frontend Tool Use"
+      dind: false
+      device: mi355_1
+      timeout_in_minutes: 25
+      depends_on:
+      - image-build-amd
+      source_file_dependencies:
+      - rust/
+      - vllm/entrypoints/openai/
+      - vllm/tool_parsers/
+      - tests/utils.py
+      - tests/tool_use/
+      - vllm/platforms/rocm.py
 
 - label: ":nvidia: (L4) Rust Frontend Distributed"
   key: rust-frontend-distributed

.buildkite/test_areas/rust_frontend_cargo.yaml

@@ -14,6 +14,15 @@ steps:
   - .buildkite/scripts/run-rust-frontend-cargo-ci.sh
   commands:
   - .buildkite/scripts/run-rust-frontend-cargo-ci.sh style-clippy
+  mirror:
+    amd:
+      label: ":amd: (MI250) Rust Frontend Cargo Style + Clippy"
+      device: mi250_1
+      no_gpu: true
+      working_dir: "/vllm-workspace"
+      timeout_in_minutes: 35
+      depends_on:
+      - image-build-amd
 
 - label: ":computer: (CPU) Rust Frontend Cargo"
   key: rust-frontend-cargo-tests
@@ -28,3 +37,12 @@ steps:
   - .buildkite/scripts/run-rust-frontend-cargo-ci.sh
   commands:
   - .buildkite/scripts/run-rust-frontend-cargo-ci.sh test
+  mirror:
+    amd:
+      label: ":amd: (MI250) Rust Frontend Cargo"
+      device: mi250_1
+      no_gpu: true
+      working_dir: "/vllm-workspace"
+      timeout_in_minutes: 35
+      depends_on:
+      - image-build-amd