31 Commits
Author SHA1 Message Date
Hanming Luandhanminglu 6e1d9c1464 [profiler] Ignore PREBUILT batches in profile-by-stage (#40098)
Co-authored-by: hanminglu <hanminglu@fb.com>
2026-09-17 22:33:55 -07:00
Hanming Lu 45715e7f20 [Fix] Disable NCCL graph buffer registration for the TP LM-head all-to-all (pure-DP decode hang under request bursts) (#38936) 2026-09-11 13:10:48 -07:00
Hanming Lu e3eeabbbfa [Profiler] Add SGLANG_PROFILE_BY_STAGE_DECODE_MIN_BS to defer the decode-stage capture (#38067) 2026-09-04 15:12:55 -07:00
Hanming Lu 526af15845 [Metrics] Discount queued prefill load by recent cache hits when waiting-queue matching is off (#35248) 2026-08-18 11:27:21 -07:00
Hanming Lu 9c1517df4a Raise PD zmq per-context socket cap via SGLANG_DISAGGREGATION_ZMQ_MAX_SOCKETS (#34450) 2026-08-11 15:27:32 -07:00
Hanming LuandLianmin Zheng b53a39c5e4 Limit prefill delayer debug logs to rank zero (#34020)
Co-authored-by: Lianmin Zheng <lianminzheng@gmail.com>
2026-08-07 13:41:19 -07:00
Hanming Lu bc7e1a07c3 Bound prefill delayer all-branch delay and decay the max_prefill_bs high-watermark (#32880) 2026-08-03 14:43:02 -07:00
Hanming Lu 297f9336c8 [Test] Enable mamba lazy extra buffer alloc edge case kl test in CI (#31991) 2026-07-21 20:45:43 -07:00
Hanming Lu d03c8cee80 Negotiate PrefillDelayer only after KV-budget admission checks (#31835) 2026-07-21 12:07:30 -07:00
Hanming Lu 927979e127 Fix pad-row top-k masking with custom_routing_function under DP attention (#31838) 2026-07-21 11:49:17 -07:00
Hanming Lu a10eee3d80 [Tokenizer] Fix abort racing server crash when large amount of aborts (#28341) 2026-06-16 14:49:35 -07:00
Hanming Lu 81dcb00673 [Spec v2] Use decode kernel for TRT-LLM MHA draft extend (#28241) 2026-06-15 16:57:08 -07:00
Hanming Lu 163fafba51 fix test_qwen3_next_models flaky (#27419) 2026-06-06 14:26:32 +08:00
Hanming LuandYAMY e0b692600f [Mamba] extra buffer lazy support (#27118)
Co-authored-by: YAMY <74099316+YAMY1234@users.noreply.github.com>
2026-06-03 12:42:11 -07:00
Hanming Lu b603f08c0c [DP] Fix FlashInfer dispatcher workspace sizing and set_dp_buffer_len (#26643) 2026-06-02 13:25:20 -07:00
Hanming Lu a5a64a311a [Spec] trtllm mha supports overlap plan stream (#25925) 2026-05-23 03:25:57 -07:00
Hanming Lu ddf3817924 Revert "[AMD]fix: use CUDA event for targeted draft-to-verify sync in… (#25917) 2026-05-20 18:49:01 -07:00
Hanming Lu 371b6c9ea0 [Scheduler] fix chunked prefill not always being full (#25741) 2026-05-20 15:03:11 -07:00
Hanming Lu 862d39e06c [Mamba] Fix extra_buffer overlap schedule races (#24954) 2026-05-19 12:13:20 +08:00
Hanming Lu afc7c9f7f3 [TRTLLM/SWA/Spec] fix trtllm mha + swa + spec accept length drop (#25103) 2026-05-15 17:15:31 -07:00
Hanming Lu 92f281f856 [Spec][trtllm] use decode kernel for draft extend (#24566) 2026-05-07 02:25:26 -07:00
Hanming Lu 9d5f16d456 [SWA] fix swa radix cache match_len_since_tombstone update when hits swa_tombstone (#17061) 2026-01-15 11:28:33 +08:00
Hanming Lu 5dcff94791 [nemetron/mtp] fix nemotron mtp (#16275) 2026-01-09 23:22:21 +08:00
Hanming Lu 41609b52fe support page size large than 64 for mamba radix cache with fix (#16768) 2026-01-09 13:28:37 +08:00
Hanming Lu e61dabf5e4 [Qwen3-next] support mamba radix cache for overlap scheduler (#14792) 2025-12-14 18:54:16 -08:00
Hanming Lu e592ee6545 [Qwen3-next] remove heuristics and add radix cache kl test (#14520) 2025-12-06 12:11:40 -08:00
Hanming Lu a0dde90af6 [CODEOWNER] update codeowner for qwen3-next related (#14522) 2025-12-05 16:15:10 -08:00
Hanming Luandyizhang2077 48d6bea1ea [GDN/SWA] mamba and swa radix cache edge case fix (#12111)
Co-authored-by: yizhang2077 <1109276519@qq.com>
2025-11-04 11:03:37 +08:00
Hanming Lu 66fb9b1307 [ServerArgs] allow --mamba-ssm-dtype extend (#12481) 2025-11-02 11:50:04 -08:00
Hanming Lu 97d857c096 [Mamba] Increase default mamba_full_memory_ratio to 0.9 (#11679) 2025-10-16 09:56:43 +08:00
Hanming LuandYing Sheng 9379da77de SWA Prefix Cache (#7367)
Co-authored-by: Ying Sheng <sqy1415@gmail.com>
2025-07-13 12:31:07 -07:00