From f2b5fabb23a62c03dfb809bcd4b678ea2e87ce5a Mon Sep 17 00:00:00 2001
From: peizhang56 <pei.zhang@amd.com>
Date: Mon, 29 Jun 2026 18:59:08 -0700
Subject: [PATCH] [ROCm][CI] Move LM Eval Large Models (8 GPUs) to mi300 pool
 (#47094)

Signed-off-by: pei.zhang <pei.zhang@amd.com>
Co-authored-by: Andreas Karatzas <akaratza@amd.com>
Co-authored-by: Claude <noreply@anthropic.com>
---
 .buildkite/test-amd.yaml | 42 ++++++++++++++++++++--------------------
 1 file changed, 21 insertions(+), 21 deletions(-)

diff --git a/.buildkite/test-amd.yaml b/.buildkite/test-amd.yaml
index ea76ae1c37f..410f90768d0 100644
--- a/.buildkite/test-amd.yaml
+++ b/.buildkite/test-amd.yaml
@@ -1198,6 +1198,27 @@ steps:
   - export VLLM_WORKER_MULTIPROC_METHOD=spawn
   - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=configs/models-mi3xx.txt
 
+- label: ROCm LM Eval Large Models (8 GPUs) # TBD
+  timeout_in_minutes: 180
+  mirror_hardwares: [amdexperimental, amdproduction, amdgfx942nightly, amdmi300]
+  agent_pool: mi300_8
+  optional: true
+  num_gpus: 8
+  working_dir: "/vllm-workspace/.buildkite/lm-eval-harness"
+  source_file_dependencies:
+  - vllm/model_executor/models/
+  - vllm/model_executor/model_loader/
+  - vllm/model_executor/layers/quantization/
+  - vllm/v1/attention/backends/
+  - vllm/v1/attention/selector.py
+  - vllm/model_executor/layers/layernorm.py
+  - csrc/
+  - vllm/_aiter_ops.py
+  - vllm/platforms/rocm.py
+  commands:
+  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+  - pytest -s -v test_lm_eval_correctness.py --config-list-file=configs/models-large-rocm.txt --tp-size=8
+
 #---------------------------------------------------------  mi300 · examples  ----------------------------------------------------------#
 
 - label: Examples # TBD
@@ -2392,27 +2413,6 @@ steps:
   - export VLLM_USE_DEEP_GEMM=0
   - pytest -s -v test_lm_eval_correctness.py --config-list-file=configs/models-large-rocm-fp8.txt --tp-size=4
 
-- label: ROCm LM Eval Large Models (8 GPUs) # TBD
-  timeout_in_minutes: 180
-  mirror_hardwares: [amdexperimental, amdproduction, amdgfx942nightly, amdmi325]
-  agent_pool: mi325_8
-  optional: true
-  num_gpus: 8
-  working_dir: "/vllm-workspace/.buildkite/lm-eval-harness"
-  source_file_dependencies:
-  - vllm/model_executor/models/
-  - vllm/model_executor/model_loader/
-  - vllm/model_executor/layers/quantization/
-  - vllm/v1/attention/backends/
-  - vllm/v1/attention/selector.py
-  - vllm/model_executor/layers/layernorm.py
-  - csrc/
-  - vllm/_aiter_ops.py
-  - vllm/platforms/rocm.py
-  commands:
-  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-  - pytest -s -v test_lm_eval_correctness.py --config-list-file=configs/models-large-rocm.txt --tp-size=8
-
 #-----------------------------------------------------  mi325 · models / language  -----------------------------------------------------#
 
 - label: Language Models Test (Extended Generation) # TBD