A4 & A4X TRTLLM GKE single-host inference benchmarking recipes

inference/a4/single-host-serving/sglang/README.md → inference/a4/single-host-serving/sglang/deepseek-r1-671b/README.md

@@ -126,7 +126,7 @@ First, you'll configure your local environment. These steps are required once be
 git clone https://github.com/ai-hypercomputer/gpu-recipes.git
 cd gpu-recipes
 export REPO_ROOT=$(pwd)
-export RECIPE_ROOT=$REPO_ROOT/inference/a4/single-host-serving/sglang
+export RECIPE_ROOT=$REPO_ROOT/inference/a4/single-host-serving/sglang/deepseek-r1-671b
 ```
 
 <a name="configure-vars"></a>
@@ -450,4 +450,4 @@ To avoid incurring further charges, clean up the resources you created.
 3.  (Optional) Delete the built Docker image from Artifact Registry if no longer needed.
 4.  (Optional) Delete Cloud Build logs.
 5.  (Optional) Clean up files in your GCS bucket if benchmarking was performed.
-6.  (Optional) Delete the [test environment](#test-environment) provisioned including GKE cluster.
+6.  (Optional) Delete the [test environment](#test-environment) provisioned including GKE cluster.

inference/a4/single-host-serving/sglang/values.yaml → inference/a4/single-host-serving/sglang/deepseek-r1-671b/values.yaml

@@ -59,4 +59,4 @@ network:
   gibVersion: us-docker.pkg.dev/gce-ai-infra/gpudirect-gib/nccl-plugin-gib:v1.0.5
   ncclSettings:
     - name: NCCL_DEBUG
-      value: "WARN"
+      value: "WARN"