Commit Graph

Select branches

Hide Pull Requests

dsv41-pd

main

Mono Color

• 104218d9ed ci: tag builds with -ci-<UTC datetime> suffix (naming rule: <branch>-<sha9>-<ci|local>-<UTC>) dsv41-pd minke.yu 2026-09-22 20:37:57 +08:00
• 92632a60ba ci: drop gha cache backend (runner cache server 404s on this instance) minke.yu 2026-09-22 18:49:20 +08:00
• 4b3b367b63 ci: hardcode Gitea registry (vars context falls back to docker.io on this instance) minke.yu 2026-09-22 18:23:35 +08:00
• a78da9b524 ci: Gitea Actions workflow to build sglang image on push to dsv41-pd minke.yu 2026-09-22 16:59:13 +08:00
• b081dd3d23 Merge branch 'main' into dsv41-pd minke.yu 2026-09-22 14:56:35 +08:00
• 8ac19cc19f [AMD][Kimi-K3] Fix deferred KDA gate projection and update DCP cookbook (#39066) main Kevin Mi 2026-09-21 23:35:12 -07:00
• 4c81cd1b09 [KDA] Fix missing beta sigmoid in PTX prefill (#40685) Mohammad Miadh AngkadandMohammad Angkad 2026-09-22 06:24:38 +00:00
• bc22e1de9e [DSpark] Fix draft CUDA graph stream explosion (#40658) Khoa Pham 2026-09-21 23:05:45 -07:00
• a0781f2714 [Docs] GLM-5.3/5.3-Flash cookbooks: enable reasoning/tool-call parsers by default via auto (#40497) Brayden ZhongandXinyuan Tong 2026-09-22 01:54:08 -04:00
• 04c0913434 [HiSparse] Add MHA hisparse support for MiniMax M3 (#31446) Guangda LiuandGuangda Liu 2026-09-22 13:28:03 +08:00
• 095e45100b [AMD] [GLM-5.3-Flash Day 0] Route mHC through AITER on gfx950 (#38545) 6 people 2026-09-21 22:24:29 -07:00
• 264da63319 [AMD] Update ROCm AITER pin to acf8fdf9 (#39965) kangwangamd 2026-09-22 13:01:17 +08:00
• b01961e295 [LFM2-VL] Add DSpark speculative decoding (#40651) Piotr Mazurek 2026-09-21 21:49:51 -07:00
• 9b59fc5db5 [ModelOpt][PP] Keep BF16 shared experts out of the NVFP4 fusion so TP1 pipeline stages can load (#40628) YAMY 2026-09-21 23:45:58 -05:00
• 2032f3a071 [Router] Abort the engine when a client disconnects mid-request (#39461) 5 people 2026-09-21 21:35:56 -07:00
• a9f02b0fa4 [NPU] Fix xgrammar apply_vocab_mask device dispatch to use torch.ops.npu (#36120) 3 people 2026-09-22 12:34:15 +08:00
• 877a293d6d [Benchmark] Optionally clear HiCache storage between cases (#40659) 3 people 2026-09-21 21:31:21 -07:00
• e1daf68304 [AMD] [GLM-5.3-Flash Day 0] Honor fused and per-expert names in quark exclude (#39317) 6 people 2026-09-21 21:26:42 -07:00
• 018b73c7a0 [PD] Pack draft KV head slices for DCP transfers (#40500) Khoa PhamandQiaolin Yu 2026-09-21 21:11:27 -07:00
• b44e248682 [AMD] [GLM-5.3-Flash Day 0] Enable FP8 and Quark MXFP4 MoE on gfx950 (#38546) 6 people 2026-09-21 21:07:37 -07:00
• 15ba54bd5d perf(engine): avoid timed waits for Engine responses (#39486) jthomson04 2026-09-21 20:42:03 -07:00
• 56fee88e23 fix(moe): support Llama4 NVFP4 router input weights on SM120 (#35504) Jan BernlöhrandPo-Han Huang 2026-09-22 05:18:11 +02:00
• 59a723ef1e [sgl-router] refactor - SLO ordering for bucket selection (#40292) Kan Wu 2026-09-21 20:16:10 -07:00
• 27f796ca6c [sgl-router] Fix readiness, IPv6 discovery, logging, and model validation (#40604) Kan Wu 2026-09-21 20:13:06 -07:00
• 1d025491f3 [Test] Set DP size in the mocked Metal profiler test (#40667) Cheng Wan 2026-09-21 19:57:52 -07:00
• 90cf471723 [AMD] [GLM-5.3-Flash Day 0] Support non-2048 top-k widths in the DSA page-table transform (#39340) 5 people 2026-09-22 10:50:17 +08:00
• 5f9c6b9eb0 [diffusion] fix: separate a use-scoped layerwise release from release_all (#40590) 3 people 2026-09-22 10:38:44 +08:00
• a1b2b976fe [diffusion] CI: restore public Qwen-Image 2.1 TP2 E2E coverage (#40507) MickandMick Qian 2026-09-22 10:38:00 +08:00
• 15eba3b464 Feat: Add TensorCast storage as a new HiCache backend (#27265) Yuhan Zhou 2026-09-22 10:19:04 +08:00
• bc30fa1759 [AMD][Fix] AgentX HIP TPOT regression when SGLANG_SIMULATE_ACC_LEN is set (#40598) jacky.cheng 2026-09-22 10:16:15 +08:00
• 9eda772a21 [Test] Handle tied top-k indices in graph-pool logprob regression (#40661) Cheng Wan 2026-09-21 19:02:43 -07:00
• e332e1b84e [Fix] Don't write conv state from the fused KDA verify kernel (#39524) Mohammad Miadh Angkadandmmangkad 2026-09-22 01:33:03 +00:00
• 35eb7cf8d6 [Intel][XPU][KVCanary] Enable KV Canary on Intel XPU (#33520) Dayananda VandClaude Opus 5 2026-09-22 06:49:01 +05:30
• 046cd6f4ea [XPU][ci]: disable XPU NIXL disaggregation test (#40540) Ma Mingfei 2026-09-22 09:02:22 +08:00
• c4d3770a68 [Kimi K3] Fix CUDA graph stream explosion (#40640) Khoa Pham 2026-09-21 17:49:50 -07:00
• 31b577bb08 [AMD] Pad QSA MQA decode Q-heads to 16 for ROCm MFMA (#38875) jacky.cheng 2026-09-22 08:43:46 +08:00
• 98c8dee23b Fix lint failure from draft-decode window test location (#40654) Cheng Wan 2026-09-21 17:41:00 -07:00
• 042b6a488f [AMD] [GLM-5.3-Flash Day 0] Enable zero-RoPE MHA prefill on ROCm (#39338) 3 people 2026-09-22 08:38:07 +08:00
• 582389cec5 [Fix] Keep diffusion encoder TP context bindings consistent (#40646) Cheng Wan 2026-09-21 17:23:15 -07:00
• c53cc8e1eb [NPU][BugFix] Avoid M-RoPE recompilation for variable sequence lengths (#40371) Tao LiandXiaoyu Zhang 2026-09-22 08:19:25 +08:00
• 61d0cf2074 [Spec] Windowed draft-decode attention for built-in EAGLE / MTP drafts (#32673) avalliappan-nvidia 2026-09-21 17:17:57 -07:00
• 9fdb71732a Avoid materializing GDN QKV tensors during target verification (#33778) Vedant V JhaveriandCopilot 2026-09-21 17:04:18 -07:00
• 506698761d [unified-memory] Hierarchical cache for every unified pool shape (#37507) Cheng Wan 2026-09-21 16:50:37 -07:00
• 22587fb15c [Fix] Run KV canary hooks for context-parallel prefill (#40642) Cheng Wan 2026-09-21 16:46:22 -07:00
• 00986c81be Support GLM-5.3-Flash hybrid attention CPU offload and PD index mapping (#40310) Yuxuan ZhangandXinyuan Tong 2026-09-22 07:03:03 +08:00
• 0229025127 [Spec][PP] Launch extend microbatches before the spec output exchange (#40499) YAMY 2026-09-21 17:47:03 -05:00
• 0c53fec476 [ROCm] fix: remove extra bf16 -> fp32 cast in jit grouped topk kernel path (#39775) RuibinCheung 2026-09-22 06:43:21 +08:00
• d47b8c454c [sgl-router] Release cancelled circuit-breaker probes (#40603) Kan Wu 2026-09-21 15:39:50 -07:00
• a5c2cc517c [CI] Split the CI control labels into four axes and resolve them live (#40527) Liangsheng Yin 2026-09-21 15:37:28 -07:00
• 66f19f5c46 [AMD] Enable HiCache for GLM-5.2 MI355X throughput recipe (#40570) Zhang, Jiejing 2026-09-21 15:28:27 -07:00
• 8bde82c0ad [AMD] [GLM-5.3-Flash Day 0] Build the fused DSA k-pool top-k JIT kernel on HIP (#39339) 5 people 2026-09-22 06:24:44 +08:00
• 2261c2e618 Add MiMo-V2.6 cookbook (#40622) 3 people 2026-09-22 05:58:53 +08:00
• acac4dd9d9 [Refactor] Clean up parallel runtime comments (#40632) Cheng Wan 2026-09-21 14:32:22 -07:00
• f532ad1f9a Fix GLM-5.3 forget-gate shape for nvCUTEDSL verify (#40607) William Hu 2026-09-21 16:52:26 -04:00
• e0c2e8dc4d [AMD] Tune Qwen3.5 TP4 GDN recurrent launch on gfx950 (#39987) jacky.cheng 2026-09-22 04:20:54 +08:00
• 1ed6822039 [Test] Anchor basic_perf thresholds to each metric's measured spread (#40617) Liangsheng Yin 2026-09-21 13:02:10 -07:00
• b18ca9ca44 [CI] Bump sgl-eval to 0.1.2 (#40620) Liangsheng Yin 2026-09-21 13:00:54 -07:00
• 11e661fd45 [Fix] Don't free the multi-CTAs KV counter the decode graphs captured (#39175) 4 people 2026-09-21 19:58:48 +00:00
• 44bdf225d8 Fix lint failure from MXFP8 reserved-slot test location (#40618) Cheng Wan 2026-09-21 12:31:11 -07:00
• bccf691b22 Bringing the parallel runtime up becomes a phase, not a side effect (#40345) Cheng Wan 2026-09-21 12:29:50 -07:00
• 1d3243d05f Take the parallel getters off the package's public surface (#40344) Cheng Wan 2026-09-21 12:27:50 -07:00
• 970e946e4f Retire the per-runner parallel record (#40343) Cheng Wan 2026-09-21 12:26:40 -07:00
• 73f071db52 Deprecate the parallel getters the context answers, and ratchet them shut (#40342) Cheng Wan 2026-09-21 12:25:32 -07:00
• 65be3fa71a A runner and the objects it builds freeze the placement they describe (#40341) Cheng Wan 2026-09-21 12:24:17 -07:00
• 2d0e94e3a3 Check the topology identities where the layout is written, and build at the published widths (#40340) Cheng Wan 2026-09-21 12:22:59 -07:00
• d5fdab7022 chore: add NIXL owners and CI access (#40602) ishandhananiandKangyan-Zhou 2026-09-21 12:21:30 -07:00
• 0db1a93adb State the draft's whole topology in its scope, and read the rest from the context (#40339) Cheng Wan 2026-09-21 12:19:38 -07:00
• ae7a516ba7 feat: use XGrammar V4.1 DSML parameter constraints (#39026) 4 people 2026-09-21 15:12:28 -04:00
• f0940fe3a6 Update DeepSeek-V4 Pro for B200 FP4 agentic PD disaggregation (#40610) Faradawn Yang 2026-09-21 12:10:03 -07:00
• 90b3f8544c [AMD] Use Triton softmax routing for Qwen3.5 on gfx950 (#39986) jacky.cheng 2026-09-22 03:02:29 +08:00
• f702a0be29 Revert "[Diffusion] migrate the whole _register_configs from registry.py to the model own config file" (#40611) ronnie_zheng 2026-09-21 21:05:19 +03:00
• 632919e498 [AMD] Fix DeepSeek-R1-MXFP4 accuracy with AITER FP8 (#37762) Bingxu Chen 2026-09-22 01:51:37 +08:00
• e6931ca889 [Diffusion] migrate the whole _register_configs from registry.py to the model own config file (#40475) ronnie_zheng 2026-09-21 20:45:53 +03:00
• 7a6191c4b9 Preallocate HiCache MHA staging before post-capture KV sizing (#40256) cctry 2026-09-21 10:44:29 -07:00
• 7ad55e4386 [HiCache] TMA-staged host<->device KV transfer kernel (sm_90+) (#40278) cctry 2026-09-21 10:38:23 -07:00
• 0cb37c018c [KDA] Enable ReplaySSM for GLM-5.3 Flash (#40517) William Hu 2026-09-21 13:34:08 -04:00
• 3c71bb018a [AMD] Enable GLM DSA prefill top-k to the v2 kernel (#37889) Eric.Chin.AMDandThomas Wang 2026-09-22 01:31:07 +08:00
• 5a6a1bb883 [mxfp8-kv] Skip writes to the reserved CUDA-graph padding slot (#35351) 4 people 2026-09-21 13:21:25 -04:00
• 008470abd8 [sgl-router] Bound streaming lifetimes and release guards on idle disconnect (#40391) Kan WuandShangming Cai 2026-09-21 10:07:07 -07:00
• 800613a74b [Test] Split the serving perf tests by topic into basic_perf/ and route their thresholds through a kit (#40505) Liangsheng Yin 2026-09-21 10:05:57 -07:00
• 14e9c40a72 [Observability] Expose python/rust frontend identity in /server_info (#39993) Sage 2026-09-21 18:32:52 +03:00
• 50ec9702d0 [diffusion] docs: update ComfyUI sections, trimmed examples, and the RTX 5090 DiT-resident recipe (1.42x) for Qwen-Image-2.1 cookbook (#40573) 3 people 2026-09-21 21:23:11 +08:00
• 69d1e5cfe0 [Docs][NPU] Add MiMo-V2.5-Pro FP4 DFlash best practice on Ascend NPU (#40577) iridiumine 2026-09-21 20:32:52 +08:00
• b410010087 [NPU] [DOC] Add kimi k3 cookbook for 950PR/DT Series (#40575) amote-i 2026-09-21 20:18:03 +08:00
• 0f6761b54f [sgl-router] Add SGLang-compatible DeepSeek V4.1 Flash rendering (#40532) Kan Wu 2026-09-21 03:35:09 -07:00
• 0abb251a20 [sgl-router] Match DeepSeek V4 rendering to SGLang (#40530) 3 people 2026-09-21 02:56:41 -07:00
• 2016f5e7a1 [sgl-router] Add Kimi-K3 rendering with SGLang parity (#40390) Kan WuandClaude Fable 5.1 2026-09-21 02:38:49 -07:00
• b86a30afba [AMD][DI][CI] Add a SPUR cluster profile to AMD DI CI (#40113) Zhaoyi Li 2026-09-21 03:56:46 -05:00
• 8faa2d6731 [NPU][Diffusion] Disable loading latency checks in Ascend fixtures (#40544) hanwlax 2026-09-21 16:56:35 +08:00
• 630b1ef322 [sgl-router] Fix reorg admission proxy test build after BucketResolver::new (#40537) 3 people 2026-09-21 16:42:29 +08:00
• 8d08dfdab7 [Fix] Add gigachat35 to the tool-call and reasoning parser name lists (#40554) Mohammad Miadh AngkadandMohammad Angkad 2026-09-21 08:34:43 +00:00
• a9871012ac [sgl-router] refactor - generalized admission policy definitions (#40271) Kan WuandClaude Fable 5.1 2026-09-21 01:26:30 -07:00
• 70b5b03e78 [NPU][CI] Fix paths-filter negation that makes every PR run the NPU tier (#40549) Shangming Cai 2026-09-21 16:07:27 +08:00
• c2f860af1c [ci][xpu] Record device time in the multimodal_gen perf lane (#39956) ashwini rathi 2026-09-21 13:09:29 +05:30
• b63f8416b3 [Feature] Gigachat 3.5 support (#29189) 6 people 2026-09-21 09:37:55 +03:00
• b54d5b7c7b disaggregation: Fix FakeKVSender queue accumulation (#28652) kpjeeja 2026-09-21 11:57:05 +05:30
• f5f3c38aad [Fix] Preserve YaRN scaling when extending rotary caches (#38786) skyler-apdx 2026-09-21 14:14:05 +08:00
• d20cd9d77f [XPU]Enable HiSparse hierarchical sparse KV cache on Intel XPU (#32792) AMRUTHA MandMa Mingfei 2026-09-21 11:37:49 +05:30
• fcb080bd40 [sgl-router] refactor - cache-aware policy (#40366) Kan WuandClaude Fable 5.1 2026-09-20 22:53:57 -07:00
• 11ecdbf39f Clean up startup logging and streamline log audits (#40526) Lianmin Zheng 2026-09-20 22:28:35 -07:00