From 27044a35d682a402ccdc75487016f069d583c9b3 Mon Sep 17 00:00:00 2001
From: Oseltamivir <58582368+Oseltamivir@users.noreply.github.com>
Date: Fri, 16 Oct 2026 21:24:36 +0800
Subject: [PATCH 1/2] feat: add HiCache offload arm to H100 Muse Spark at c2-c8

---
 .../musespark/sglang/h100-fp4-mtp/agentic.yaml   | 16 ++++++++++++++++
 inferencex-e2e/configs/nvidia-master.yaml        |  4 +++-
 inferencex-e2e/perf-changelog.yaml               |  9 +++++++++
 3 files changed, 28 insertions(+), 1 deletion(-)

diff --git a/inferencex-e2e/benchmarks/single_node/srt-slurm-recipes/musespark/sglang/h100-fp4-mtp/agentic.yaml b/inferencex-e2e/benchmarks/single_node/srt-slurm-recipes/musespark/sglang/h100-fp4-mtp/agentic.yaml
index 860efad7e9..3c55839fa3 100644
--- a/inferencex-e2e/benchmarks/single_node/srt-slurm-recipes/musespark/sglang/h100-fp4-mtp/agentic.yaml
+++ b/inferencex-e2e/benchmarks/single_node/srt-slurm-recipes/musespark/sglang/h100-fp4-mtp/agentic.yaml
@@ -77,3 +77,19 @@ zip_override_tp8:
   benchmark:
     env:
       CONC: ['1', '2', '4', '8', '16']
+      KV_OFFLOADING: none
+
+# The GPU KV pool is a few GB per rank, so agentic prefixes evict between turns;
+# HiCache keeps them in host DRAM instead of recomputing them.
+zip_override_tp8_hicache:
+  roles:
+    agg:
+      args:
+        max-running-requests: [4, 8, 8]
+        cuda-graph-max-bs-decode: [4, 8, 8]
+        enable-hierarchical-cache: true
+        hicache-size: 160
+  benchmark:
+    env:
+      CONC: ['2', '4', '8']
+      KV_OFFLOADING: dram
diff --git a/inferencex-e2e/configs/nvidia-master.yaml b/inferencex-e2e/configs/nvidia-master.yaml
index bc36fab26d..1fc173c6a4 100644
--- a/inferencex-e2e/configs/nvidia-master.yaml
+++ b/inferencex-e2e/configs/nvidia-master.yaml
@@ -9708,5 +9708,7 @@ musespark-fp4-h100-sglang-agentic-dflash:
   srt-recipe-dir: musespark/sglang/h100-fp4-mtp
   scenarios:
     agentic-coding:
-    - search-space:
+    - dram-utilization: 0.80
+      search-space:
       - { tp: 8, ep: 8, kv-offloading: none, spec-decoding: draft_model, conc-list: [1, 2, 4, 8, 16], srt-recipe: agentic.yaml }
+      - { tp: 8, ep: 8, kv-offloading: dram, kv-offload-backend: { name: hicache }, spec-decoding: draft_model, conc-list: [2, 4, 8], srt-recipe: agentic.yaml }
diff --git a/inferencex-e2e/perf-changelog.yaml b/inferencex-e2e/perf-changelog.yaml
index 2ea1838d04..ec2296cc23 100644
--- a/inferencex-e2e/perf-changelog.yaml
+++ b/inferencex-e2e/perf-changelog.yaml
@@ -10004,3 +10004,12 @@
     - "Synthetic acceptance follows the committed musespark_dflash golden curve (#4138, #4139): AL 4.05 at 7 draft tokens, thinking_on."
     - "The server loads the checkpoint and draft from the pre-seeded shared HF cache with HF_HUB_OFFLINE=1, so concurrent points make no Hub API calls."
   pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/4127
+
+- config-keys:
+    - musespark-fp4-h100-sglang-agentic-dflash
+  scenario-type:
+    - agentic-coding
+  description:
+    - "Add a HiCache host-DRAM KV offload arm at concurrency 2, 4 and 8 with 160 GB per rank, alongside the GPU-resident sweep; the few-GB GPU KV pool left by 67 GB of weights per H100 evicts agentic prefixes between turns."
+    - "Keep the batch-8 cap on running requests and decode graphs from the GPU-resident sweep."
+  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/XXXX

From e36cc330a7cba3762746dc9ad1b5e28916dd289d Mon Sep 17 00:00:00 2001
From: Oseltamivir <58582368+Oseltamivir@users.noreply.github.com>
Date: Fri, 16 Oct 2026 21:24:44 +0800
Subject: [PATCH 2/2] chore: link H100 HiCache entry to PR 4181

---
 inferencex-e2e/perf-changelog.yaml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/inferencex-e2e/perf-changelog.yaml b/inferencex-e2e/perf-changelog.yaml
index ec2296cc23..8ec368892e 100644
--- a/inferencex-e2e/perf-changelog.yaml
+++ b/inferencex-e2e/perf-changelog.yaml
@@ -10012,4 +10012,4 @@
   description:
     - "Add a HiCache host-DRAM KV offload arm at concurrency 2, 4 and 8 with 160 GB per rank, alongside the GPU-resident sweep; the few-GB GPU KV pool left by 67 GB of weights per H100 evicts agentic prefixes between turns."
     - "Keep the batch-8 cap on running requests and decode graphs from the GPU-resident sweep."
-  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/XXXX
+  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/4181