[CI] Remove the retired A100 queue's steps; keep their live AMD mirrors standalone

.buildkite/test_areas/distributed.yaml

@@ -479,45 +479,34 @@ steps:
       - export TORCH_CPP_LOG_LEVEL=INFO
       - 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: ":nvidia: (A100) [Nightly] Distributed"
-  key: distributed-tests-4xa100
-  device: a100
+# Key kept from the former A100 step's mirror for job history and retries.
+- label: ':amd: (MI355) Distributed'
+  key: amd-distributed-tests-4xa100
+  dind: false
+  device: mi355_4
   optional: true
   num_devices: 4
+  timeout_in_minutes: 40
+  working_dir: /vllm-workspace/tests
+  depends_on:
+  - image-build-amd
   source_file_dependencies:
   - vllm/
   - "!vllm/distributed/kv_transfer/"
+  - tests/models/test_vision.py
+  - vllm/platforms/rocm.py
+  - tests/distributed/
+  - tests/basic_correctness/
+  - tests/lora/test_mixtral.py
   commands:
   # NOTE: don't test llama model here, it seems hf implementation is buggy
   # see https://github.com/vllm-project/vllm/pull/5689 for details
   - pytest -v -s distributed/test_custom_all_reduce.py
   - torchrun --nproc_per_node=2 distributed/test_ca_buffer_sharing.py
   - VLLM_ALLOW_INSECURE_SERIALIZATION=1 pytest -v -s distributed/test_weight_transfer.py -k multiple_sources
-  - TARGET_TEST_SUITE=A100 pytest basic_correctness/ -v -s -m 'distributed(num_gpus=2)'
+  - TARGET_TEST_SUITE=MI355 pytest basic_correctness/ -v -s -m 'distributed(num_gpus=2)'
   - pytest -v -s -x lora/test_mixtral.py
-  mirror:
-    amd:
-      label: ':amd: (MI355) Distributed'
-      dind: false
-      device: mi355_4
-      timeout_in_minutes: 40
-      working_dir: /vllm-workspace/tests
-      depends_on:
-      - image-build-amd
-      source_file_dependencies:
-      - vllm/
-      - tests/models/test_vision.py
-      - vllm/platforms/rocm.py
-      - tests/distributed/
-      - tests/basic_correctness/
-      - tests/lora/test_mixtral.py
-      commands:
-      - pytest -v -s distributed/test_custom_all_reduce.py
-      - torchrun --nproc_per_node=2 distributed/test_ca_buffer_sharing.py
-      - VLLM_ALLOW_INSECURE_SERIALIZATION=1 pytest -v -s distributed/test_weight_transfer.py -k multiple_sources
-      - TARGET_TEST_SUITE=MI355 pytest basic_correctness/ -v -s -m 'distributed(num_gpus=2)'
-      - pytest -v -s -x lora/test_mixtral.py
-      - pytest -v -s models/test_vision.py -m 'distributed(num_gpus=4)'
+  - pytest -v -s models/test_vision.py -m 'distributed(num_gpus=4)'
 
 - label: ":nvidia: (H100) [Nightly] Distributed Features"
   key: distributed-tests-2xh100-2xmi300

.buildkite/test_areas/lm_eval.yaml

@@ -94,19 +94,6 @@ steps:
       commands:
       - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=configs/models-watermark.txt
 
-# - label: LM Eval Large Models (4xA100)
-#   key: lm-eval-large-models-4xa100
-#   device: a100
-#   optional: true
-#   num_devices: 4
-#   working_dir: "/vllm-workspace/.buildkite/lm-eval-harness"
-#   source_file_dependencies:
-#   - csrc/
-#   - vllm/model_executor/layers/quantization
-#   commands:
-#   - 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: ":nvidia: (H100) [Nightly] LM Eval Large Models"
   key: lm-eval-large-models-4xh100
   automatic_shard: true
@@ -865,38 +852,6 @@ steps:
   commands:
   - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=evals/gsm8k/configs/humming/config-l4.txt
 
-- label: ":nvidia: (A100) [Nightly] LM Eval Humming Default TEMPORARY Shard %N"
-  key: lm-eval-humming-a100-default
-  optional: true
-  timeout_in_minutes: 75
-  device: a100
-  num_devices: 1
-  parallelism: 3
-  source_file_dependencies:
-  - vllm/model_executor/layers/quantization/humming.py
-  - vllm/model_executor/layers/quantization/utils/humming/
-  - vllm/model_executor/layers/fused_moe/experts/fused_humming_moe.py
-  - vllm/model_executor/layers/fused_moe/oracle/
-  - vllm/model_executor/kernels/linear/
-  commands:
-  - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=evals/gsm8k/configs/humming/config-a100-shard-$$BUILDKITE_PARALLEL_JOB.txt
-
-- label: ":nvidia: (A100) [Nightly] LM Eval Humming Activation INT8 TEMPORARY Shard %N"
-  key: lm-eval-humming-a100-act
-  optional: true
-  timeout_in_minutes: 45
-  device: a100
-  num_devices: 1
-  parallelism: 2
-  source_file_dependencies:
-  - vllm/model_executor/layers/quantization/humming.py
-  - vllm/model_executor/layers/quantization/utils/humming/
-  - vllm/model_executor/layers/fused_moe/experts/fused_humming_moe.py
-  - vllm/model_executor/layers/fused_moe/oracle/
-  - vllm/model_executor/kernels/linear/
-  commands:
-  - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=evals/gsm8k/configs/humming/config-a100-act-int8-shard-$$BUILDKITE_PARALLEL_JOB.txt
-
 - label: ":nvidia: (H100) [Nightly] LM Eval Humming Default TEMPORARY"
   key: lm-eval-humming-h100-default
   automatic_shard: true

.buildkite/test_areas/misc.yaml

@@ -881,39 +881,30 @@ steps:
   commands:
   - pytest -v -s tool_parsers
 
-- label: ":nvidia: (A100) Batch Invariance"
-  key: batch-invariance-a100
+# Key kept from the former A100 step's mirror for job history and retries.
+- label: ":amd: (MI250) Batch Invariance"
+  key: amd-batch-invariance-a100
+  dind: false
+  device: mi250_1
   timeout_in_minutes: 60
-  device: a100
+  env:
+    VLLM_GPU_MEMORY_UTILIZATION: "0.8"
+    VLLM_TEST_CLEAN_GPU_MEMORY: "1"
+  depends_on:
+  - image-build-amd
   source_file_dependencies:
-    - vllm/v1/attention
-    - vllm/model_executor/layers
-    - vllm/model_executor/kernels/linear
-    - vllm/model_executor/determinism
-    - tests/v1/determinism/
+  - vllm/v1/attention
+  - vllm/model_executor/layers
+  - vllm/model_executor/kernels/linear
+  - vllm/model_executor/determinism
+  - tests/v1/determinism/
+  - vllm/_aiter_ops.py
+  - vllm/platforms/rocm.py
+  - docker/Dockerfile.rocm_base
   commands:
     - export VLLM_WORKER_MULTIPROC_METHOD=spawn
     - 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
-  mirror:
-    amd:
-      label: ":amd: (MI250) Batch Invariance"
-      dind: false
-      device: mi250_1
-      timeout_in_minutes: 60
-      env:
-        VLLM_GPU_MEMORY_UTILIZATION: "0.8"
-        VLLM_TEST_CLEAN_GPU_MEMORY: "1"
-      depends_on:
-      - image-build-amd
-      source_file_dependencies:
-      - vllm/v1/attention
-      - vllm/model_executor/layers
-      - vllm/model_executor/kernels/linear
-      - vllm/model_executor/determinism
-      - tests/v1/determinism/
-      - vllm/_aiter_ops.py
-      - vllm/platforms/rocm.py
 
 - label: ":nvidia: (H100) Batch Invariance"
   key: batch-invariance-h100