diff --git a/configs/amd-master.yaml b/configs/amd-master.yaml index d26796a52f..3069f78dae 100644 --- a/configs/amd-master.yaml +++ b/configs/amd-master.yaml @@ -307,7 +307,7 @@ qwen3.5-fp8-mi355x-sglang-disagg: qwen3.5-fp4-mi355x-sglang: image: lmsysorg/sglang-rocm:v0.5.16-rocm720-mi35x-20260730 - model: amd/Qwen3.5-397B-A17B-MXFP4 + model: amd/Qwen3.5-397B-A17B-MXFP4-AttnFP8-V2 model-prefix: qwen3.5 runner: mi355x precision: fp4 diff --git a/perf-changelog.yaml b/perf-changelog.yaml index 64f87b4479..d29f72add4 100644 --- a/perf-changelog.yaml +++ b/perf-changelog.yaml @@ -5770,3 +5770,10 @@ - "models.yaml: add mtp_flags (--speculative-algorithm EAGLE --speculative-eagle-topk 1) to DeepSeek-V4-Pro-AgentX; num-steps/num-draft-tokens are derived from DECODE_MTP_SIZE in server_sglang.sh. DeepSeek-V4's MTP head is a native NextN/EAGLE draft shipped with the model, so no --speculative-draft-model-path is needed." - "Two search-space arms: TP8/EP1 no-DP at conc-list [2,4,8,16,32], and TP8/EP8/DPA (ep=8, dp-attn=true) at conc-list [64,96,128], both on image lmsysorg/sglang-rocm:v0.5.15.post1-rocm720-mi35x-20260719." pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/2309 + +- config-keys: + - qwen3.5-fp4-mi355x-sglang + description: + - "Update the Qwen3.5 FP4 MI355X SGLang checkpoint to amd/Qwen3.5-397B-A17B-MXFP4-AttnFP8-V2" + - "Quantize attention to PTPC FP8 and shared experts to MXFP4" + pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/2465