Add 1k1k and adjust other configs concurrency

Author: kyleliang-nv

…5/1k1k/disagg-gb200-2p1d-dep4-dep16.yaml …5/1k1k/disagg-gb200-1p1d-dep4-dep16.yamlrecipes/vllm/kimi-k2.5/1k1k/disagg-gb200-2p1d-dep4-dep16.yaml renamed to recipes/vllm/kimi-k2.5/1k1k/disagg-gb200-1p1d-dep4-dep16.yaml recipes/vllm/kimi-k2.5/1k1k/disagg-gb200-2p1d-dep4-dep16.yaml renamed to recipes/vllm/kimi-k2.5/1k1k/disagg-gb200-1p1d-dep4-dep16.yaml

@@ -1,4 +1,4 @@
-name: "kimi-vllm-disagg-gb200-2p1d-dep4-dep16"
+name: "kimi-vllm-disagg-gb200-1p1d-dep4-dep16"
 
 model:
   path: "kimi-k2.5-nvfp4"
@@ -14,9 +14,9 @@ setup_script: vllm-container-deps.sh
 resources:
   gpu_type: "gb200"
   gpus_per_node: 4
-  prefill_nodes: 2
+  prefill_nodes: 1
   decode_nodes: 4
-  prefill_workers: 2
+  prefill_workers: 1
   decode_workers: 1
   gpus_per_prefill: 4
   gpus_per_decode: 16
@@ -54,7 +54,7 @@ backend:
       data-parallel-rpc-port: 13345
       enable-expert-parallel: true
       max-model-len: 3072
-      max-num-seqs: 1024
+      max-num-seqs: 4096
       enforce-eager: true
       compilation-config: '{"custom_ops":["+quant_fp8","+rms_norm","+rotary_embedding"],"pass_config":{"fuse_attn_quant":true,"fuse_allreduce_rms":true}}'
       max-num-batched-tokens: 16384
@@ -78,7 +78,7 @@ backend:
       data-parallel-rpc-port: 13345
       enable-expert-parallel: true
       max-model-len: 3072
-      max-num-seqs: 256
+      max-num-seqs: 4096
       max-num-batched-tokens: 10240
       safetensors-load-strategy: "prefetch"
       trust-remote-code: true
@@ -91,11 +91,11 @@ backend:
       compilation-config: '{"cudagraph_mode":"FULL_DECODE_ONLY","custom_ops":["+quant_fp8","+rms_norm","+rotary_embedding"],"pass_config":{"fuse_attn_quant":true,"fuse_allreduce_rms":true}}'
       gpu-memory-utilization: 0.9
       stream-interval: 50
-      max-cudagraph-capture-size: 256
+      max-cudagraph-capture-size: 512
 
 benchmark:
   type: "sa-bench"
   isl: 1024
   osl: 1024
-  concurrencies: "512x1024x2048"
+  concurrencies: "256x512x1024x2048x3072x4096"
   req_rate: "inf"

recipes/vllm/kimi-k2.5/1k1k/disagg-gb200-1p4d-dep4-tep4.yaml

@@ -94,5 +94,5 @@ benchmark:
   type: "sa-bench"
   isl: 1024
   osl: 1024
-  concurrencies: "4x8x16x32x64x128x256"
+  concurrencies: "4x8x16x32x64x128"
   req_rate: "inf"