From 678bb8bc82d19c4db5f70738dca338506655d853 Mon Sep 17 00:00:00 2001 From: thomawan Date: Fri, 21 Aug 2026 18:39:26 +0800 Subject: [PATCH] Update image for sglang mi35x v4 agentic workload --- .../single_node/agentic/dsv4_fp4_mi355x_sglang_mtp.sh | 1 + configs/amd-master.yaml | 2 +- perf-changelog.yaml | 11 ++++++++++- 3 files changed, 12 insertions(+), 2 deletions(-) diff --git a/benchmarks/single_node/agentic/dsv4_fp4_mi355x_sglang_mtp.sh b/benchmarks/single_node/agentic/dsv4_fp4_mi355x_sglang_mtp.sh index 61add3eafa..ada85bcdaa 100644 --- a/benchmarks/single_node/agentic/dsv4_fp4_mi355x_sglang_mtp.sh +++ b/benchmarks/single_node/agentic/dsv4_fp4_mi355x_sglang_mtp.sh @@ -73,6 +73,7 @@ export SGLANG_DSV4_REASONING_EFFORT=high export SGLANG_USE_ROCM700A=0 export SGLANG_HACK_FLASHMLA_BACKEND=unified_kv_triton export AITER_BF16_FP8_MOE_BOUND=0 +export SGLANG_OPT_USE_AITER_BATCHED_GEMM=true # Unified radix tree: per-component (full-attn / SWA) cache management for # hybrid-attention models, plus proactive release of out-of-window SWA KV diff --git a/configs/amd-master.yaml b/configs/amd-master.yaml index 67a0605398..e16afa27cc 100644 --- a/configs/amd-master.yaml +++ b/configs/amd-master.yaml @@ -1727,7 +1727,7 @@ glm5.2-fp4-mi355x-atom-agentic-mtp: # collective under long-context prefill, so no DEP arm ships until that path is # validated. conc 16 appears on both arms to isolate the host KV tier's gain. dsv4-fp4-mi355x-sglang-agentic-mtp: - image: lmsysorg/sglang-rocm:v0.5.17-rocm720-mi35x-20260813 + image: lmsysorg/sglang-rocm:v0.5.17-rocm720-mi35x-20260821 model: deepseek-ai/DeepSeek-V4-Pro model-prefix: dsv4 runner: cluster:mi355x-amds diff --git a/perf-changelog.yaml b/perf-changelog.yaml index d85a0f87d2..1e63212d00 100644 --- a/perf-changelog.yaml +++ b/perf-changelog.yaml @@ -6287,4 +6287,13 @@ - "Add concurrency 48 and 64 to the TP8/EP1 sweep, and 64 and 192 to the TP8/EP8 DP-attention sweep." - "Increase prefill mem_fraction_static from 0.72 to 0.85 and reduce HICACHE_RATIO from 4 to 3." pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/2643 - + +- config-keys: + - dsv4-fp4-mi355x-sglang-agentic-mtp + scenario-type: + - agentic-coding + description: + - "Bump image to lmsysorg/sglang-rocm:v0.5.17-rocm720-mi35x-20260821" + - "Enable SGLANG_OPT_USE_AITER_BATCHED_GEMM for mi355x sglang dsv4 agentic workload." + pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/2705 +