From 75846b89436db220c5569b89a7c6610377daf867 Mon Sep 17 00:00:00 2001 From: functionstackx <47992694+functionstackx@users.noreply.github.com> Date: Mon, 7 Sep 2026 20:33:39 -0400 Subject: [PATCH] [Klaud Cold] Update qwen3.5-fp8-h100-sglang-agentic-mtp SGLang image to nightly-dev-cu13-20260907-30705c00 Co-Authored-By: Claude Fable 5.1 --- configs/nvidia-master.yaml | 2 +- perf-changelog.yaml | 8 ++++++++ 2 files changed, 9 insertions(+), 1 deletion(-) diff --git a/configs/nvidia-master.yaml b/configs/nvidia-master.yaml index 0a6c86df6d..3c56c721e4 100644 --- a/configs/nvidia-master.yaml +++ b/configs/nvidia-master.yaml @@ -7439,7 +7439,7 @@ qwen3.5-fp8-h200-sglang-agentic-hicache-mtp: qwen3.5-fp8-h100-sglang-agentic-mtp: - image: lmsysorg/sglang:v0.5.16-cu130 + image: lmsysorg/sglang:nightly-dev-cu13-20260907-30705c00 model: Qwen/Qwen3.5-397B-A17B-FP8 model-prefix: qwen3.5 runner: cluster:h100-dgxc diff --git a/perf-changelog.yaml b/perf-changelog.yaml index 6b249ea15e..e9ef43bf3e 100644 --- a/perf-changelog.yaml +++ b/perf-changelog.yaml @@ -6921,3 +6921,11 @@ - "Pick up the latest automatic ROCm DeepSeek-V4 optimizations, including fused mHC post/pre plus RMSNorm, gfx950 C4A top-k dispatch, fused C4 compressor GEMMs, fused SWA q/kv RMSNorm plus q FP8 quantization, and medium-batch cooperative top-k tuning." - "Keep the existing VLLM_ROCM_USE_AITER=1, VLLM_ROCM_USE_AITER_MOE=1, and --moe-backend aiter settings, and explicitly add VLLM_ROCM_USE_AITER_FUSION_SHARED_EXPERTS=1 plus VLLM_ROCM_QUICK_REDUCE_QUANTIZATION=INT4 to both STP and MTP paths. The current checkpoint's shared-expert path does not satisfy the latest vLLM fusion conditions, so that fusion flag self-disables while preserving recipe parity." pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/2792 + +- config-keys: + - qwen3.5-fp8-h100-sglang-agentic-mtp + scenario-type: + - agentic-coding + description: + - "Update SGLang image from lmsysorg/sglang:v0.5.16-cu130 (v0.5.16 release) to lmsysorg/sglang:nightly-dev-cu13-20260907-30705c00 (2026-09-07 cu13 dev nightly, digest sha256:19b8fa1223cc339c1eae7a5b703f1a8c2543b5b119155bf3d7efaef18f77f007, tag commit sgl-project/sglang@30705c00; Docker Hub last pushed 2026-09-07T01:43:42Z). benchmarks/single_node/agentic/qwen3.5_fp8_h100_mtp.sh is unchanged: SGLANG_ENABLE_SPEC_V2 EAGLE MTP at 3 steps, golden acceptance length 3.39, flashinfer attention (sm_90) with allreduce fusion, fp8 quantization and fp8_e4m3 KV, HiCache kernel IO / page_first layout; the TP8/EP8 GPU-resident [1, 4, 8, 12, 16] and HiCache [4, 8, 12, 16] grids are unchanged. Same tag the Qwen3.5 SGLang AgentX recipes on B200 (#2861/#2862) and H200 (#2868) moved to. The non-MTP sibling qwen3.5-fp8-h100-sglang-agentic is not bumped: its agentic script was removed in #2561 and the arm is slated for retirement under MODELS.md." + pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/2869