fix

Author: Wei Zhao (Engrg-Hardware 1)

benchmarks/multi_node/srt-slurm-recipes/vllm/deepseek-v4/agentic/agg-gb200-tp8-agentic.yaml

@@ -130,7 +130,12 @@ benchmark:
     PORT: "8000"
     # One aggregate worker owns all eight GPUs. Use the aggregate accounting
     # path instead of summing independent prefill and decode allocations.
+    # Aggregate (IS_MULTINODE=false) uses the single-node topology labels below
+    # directly (num_gpus = TP), so they are set here rather than CI-injected.
     IS_MULTINODE: "false"
+    TP: "8"
+    EP_SIZE: "1"
+    DP_ATTENTION: "false"
     AIPERF_USE_DYNAMO_CONV_AWARE_ROUTING: "1"
     AIPERF_DYNAMO_SESSION_TIMEOUT_SECONDS: "14400"
     AIPERF_DATASET_MMAP_CACHE_DIR: "/aiperf_mmap_cache"