[ROCm][CI] Adding qwen3 dp4 eplb (#47480)

Signed-off-by: Andreas Karatzas <akaratza@amd.com>

.buildkite/scripts/scheduled_integration_test/qwen30b_a3b_fp8_dp4_async_eplb.sh

@@ -18,6 +18,10 @@ wait_for_server() {
 
 MODEL="Qwen/Qwen3-30B-A3B-FP8"
 BACK="allgather_reducescatter"
+if command -v rocm-smi &> /dev/null || [[ -d /opt/rocm ]] || [[ -n "${ROCM_PATH:-}" ]]; then
+  # Disable MOE padding for ROCm since it is causing eplb to fail.
+  export VLLM_ROCM_MOE_PADDING=0
+fi
 
 cleanup() {
   if [[ -n "${SERVER_PID:-}" ]] && kill -0 "${SERVER_PID}" 2>/dev/null; then

.buildkite/test-amd.yaml

@@ -1159,6 +1159,27 @@ steps:
   commands:
   - bash .buildkite/scripts/scheduled_integration_test/qwen30b_a3b_fp8_block_ep_eplb.sh 0.8 200 8020
 
+- label: Qwen3-30B-A3B-FP8 DP4 Async EPLB Accuracy (4xH100-4xMI300) # TBD
+  timeout_in_minutes: 180
+  mirror_hardwares: [amdexperimental, amdproduction, amdgfx942nightly, amdmi300]
+  agent_pool: mi300_4
+  num_gpus: 4
+  optional: true
+  working_dir: "/vllm-workspace"
+  source_file_dependencies:
+  - vllm/model_executor/models/
+  - vllm/model_executor/model_loader/
+  - vllm/model_executor/layers/quantization/
+  - vllm/distributed/eplb
+  - vllm/model_executor/layers/fused_moe/
+  - vllm/v1/attention/backends/
+  - vllm/v1/attention/selector.py
+  - .buildkite/scripts/scheduled_integration_test/
+  - vllm/_aiter_ops.py
+  - vllm/platforms/rocm.py
+  commands:
+  - bash .buildkite/scripts/scheduled_integration_test/qwen30b_a3b_fp8_dp4_async_eplb.sh 0.8 200 8050
+
 - label: Qwen3-Next-80B-A3B-Instruct MTP Async EPLB Accuracy # TBD
   timeout_in_minutes: 180
   mirror_hardwares: [amdexperimental, amdproduction, amdgfx942nightly, amdmi300]