fix: Update agg_kvbm_router.sh to run workers on two gpus (#4643)

262cce76 · Kris Hung · GitHub · 01a634d1 · 262cce76
Unverified Commit 262cce76 authored Dec 02, 2025 by Kris Hung Committed by GitHub Dec 02, 2025
Hide whitespace changes
Inline Side-by-side

Showing with 1 addition and 1 deletion

examples/backends/vllm/launch/agg_kvbm_router.sh examples/backends/vllm/launch/agg_kvbm_router.sh +1 -1

No files found.
--- a/examples/backends/vllm/launch/agg_kvbm_router.sh
+++ b/examples/backends/vllm/launch/agg_kvbm_router.sh
@@ -32,7 +32,7 @@ CUDA_VISIBLE_DEVICES=0 DYN_KVBM_CPU_CACHE_GB=2 \
 DYN_KVBM_LEADER_ZMQ_PUB_PORT=56003 \
 DYN_KVBM_LEADER_ZMQ_ACK_PORT=56004 \
 VLLM_NIXL_SIDE_CHANNEL_PORT=20097 \
-CUDA_VISIBLE_DEVICES=0 DYN_KVBM_CPU_CACHE_GB=2 \
+CUDA_VISIBLE_DEVICES=1 DYN_KVBM_CPU_CACHE_GB=2 \
    python3 -m dynamo.vllm \
    --model $MODEL \
    --enforce-eager \