Skip to content

Commit a7afb20

Browse files
committed
agentx: adjust kimik2.7 agentic sweep top conc 48 -> 42
1 parent 416fc3d commit a7afb20

2 files changed

Lines changed: 3 additions & 3 deletions

File tree

configs/amd-master.yaml

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -820,8 +820,8 @@ kimik2.7-fp4-mi355x-vllm-agentic:
820820
agentic-coding:
821821
- dram-utilization: 0.80
822822
search-space:
823-
- { tp: 4, kv-offloading: none, conc-list: [1, 4, 8, 16, 32, 48] }
824-
- { tp: 4, kv-offloading: dram, kv-offload-backend: lmcache, conc-list: [1, 4, 8, 16, 32, 48] }
823+
- { tp: 4, kv-offloading: none, conc-list: [1, 4, 8, 16, 32, 42] }
824+
- { tp: 4, kv-offloading: dram, kv-offload-backend: lmcache, conc-list: [1, 4, 8, 16, 32, 42] }
825825

826826

827827
kimik2.5-fp4-mi355x-atom:

perf-changelog.yaml

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -4639,6 +4639,6 @@
46394639
- kimik2.7-fp4-mi355x-vllm-agentic
46404640
description:
46414641
- "Add Kimi-K2.7-Code-MXFP4 agentic-coding config on MI355X (cluster:gbt350docker->mi355x-amds) via vLLM v0.24.0, TP4"
4642-
- "Sweep KV-offloading none vs dram/LMCache at conc [1,4,8,16,32,48]; LMCache MP server + LMCacheMPConnector, dram-utilization 0.80"
4642+
- "Sweep KV-offloading none vs dram/LMCache at conc [1,4,8,16,32,42]; LMCache MP server + LMCacheMPConnector, dram-utilization 0.80"
46434643
- "Recipe benchmarks/single_node/agentic/kimik2.7_fp4_mi355x.sh clones LMCache to container-local /opt/lmcache-src (pinned) so CI checkout never trips over root-owned build artifacts"
46444644
pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/PLACEHOLDER

0 commit comments

Comments
 (0)