Commit Graph

Select branches

Hide Pull Requests

dsv41-pd

dsv41-pd-visioncp

main

Mono Color

• 8b596c10b0 [PD] Diversify fake-prefill handoff tokens (#37302) 3 people 2026-09-01 18:34:38 -07:00
• db017e3490 Build Rust extensions on hosted runners in parallel (#37258) Ke Bao 2026-09-02 09:00:25 +08:00
• 9cc43ca4e1 [NPU] Update sgl-kernel-npu version to 2026.9.0 and move memfabric deps into pyproject (#37399) Even Zhou 2026-09-02 08:57:59 +08:00
• 73a2414289 [CI] Double the base-b-test-1-gpu-large timeout (#37452) DarkSharpnessandClaude Opus 5 2026-09-02 08:47:51 +08:00
• f4c17fed07 [Diffusion] Fuse FLUX.2 NVFP4 FC1, SwiGLU, and FC2 quantization (#37096) Xiaoyu ZhangandCursor 2026-09-02 08:21:14 +08:00
• c593527f33 [Kernel] Add KDA NVFP4 GEMM for Qwen3.x on SM120 (#36865) 3 people 2026-09-02 08:16:54 +08:00
• 6d34a4d3ce [Cookbook] Verify DeepSeek-V4 Flash Vision on GB300 (#37492) zijiexia 2026-09-01 17:11:41 -07:00
• 33428d3dae Fix GPU kernel ordering and MXFP8 quantization dispatch (#37331) 3 people 2026-09-01 17:09:44 -07:00
• 221a6273ce [Kernel] Avoid zero-bias allocation in fused softmax routing (#36811) YAMY 2026-09-01 16:29:57 -07:00
• ed82bea146 [Cookbook] DeepSeek-V4: add DGX Spark (2x GB10) Flash Official FP4 recipe (#37479) Jimmy ShongandClaude Fable 5.1 2026-09-01 15:42:39 -07:00
• 9978aaec8b [ROCm][Bugfix] Cap the DSA MQA-logits budget at AITER's buffer_store limit (#36960) Yikai Zhangandamdpilot-upstream-sync 2026-09-01 15:40:12 -07:00
• f8618714c5 add reindex_device_id to device OOT plugin (#36220) 3 people 2026-09-01 15:26:47 -07:00
• fb8d7eedda Fix dummy initialization of inverse weight scales (#35491) 3 people 2026-09-02 05:19:14 +08:00
• e57e934bcc [Bugfix] Accept int64 top-k IDs in FlashInfer routed MoE packer (#32882) Po-Han Huang (NVIDIA) 2026-09-02 05:17:46 +08:00
• 83a9b5dd88 [mem_cache] Drop the torch.unique sync from the SWA page expansion (#37463) Liangsheng Yin 2026-09-01 14:16:47 -07:00
• b24c8f10e7 [FlashInfer] Avoid D2H sync for sliding-window lengths (#32218) 4 people 2026-09-02 05:08:22 +08:00
• 0f18d389b4 [Cookbook] Verify DeepSeek-V4 Flash Vision balanced and high-throughput on B200 (#37468) zijiexiaandClaude Fable 5 2026-09-01 13:09:14 -07:00
• 442c7c1e29 [Docs] GLM-5.3-Flash cookbook: add NVFP4 FP8+TRT-LLM benchmark rows (follow-up to #37109) (#37412) Xinyuan Tong 2026-09-02 03:56:22 +08:00
• 0b1ce3d140 [Feature] Unified memory: support decode context parallelism for Kimi-Linear (#36890) Cheng Wan 2026-09-01 12:44:26 -07:00
• 3315356cc0 docs(cookbook): enable FlashInfer GDN for Qwen3.5 B200 (#37360) Ankur Singh 2026-09-01 11:48:41 -07:00
• a2b8681d1d [CI][MLX] Restore the mamba_branching_seqlen attribute the MLX runner reads off a request (#37453) Shuwen Wang 2026-09-02 02:16:45 +08:00
• 9a05b470fa [Memory] Size the CUDA graph pool from warmup measurements and fix graph-pool borrowing (#36911) cctryandcctry 2026-09-01 09:32:38 -07:00
• c34f378342 fix(nixl): make FILE path-mode devId globally unique (#34362) 3 people 2026-09-02 00:11:59 +08:00
• bb3e3cbceb [AMD] Fix v4 topk issue (#37439) Thomas Wang 2026-09-01 23:37:30 +08:00
• 44a92e54b9 [AMD] fix aiter cannot get heuristic kernel regression (#37438) billishyahao 2026-09-01 23:35:31 +08:00
• cb6dd58fbe [Kernel] Replace dsv3_router_gemm with the unified tiny GEMM (#34693) DarkSharpnessandClaude Opus 5 2026-09-01 23:00:21 +08:00
• ee462b5899 [Kernel] Add tuned LFM2.5 Triton MoE configs on B300 (#37158) 3 people 2026-09-01 22:55:38 +08:00
• 5993f91f84 [Kernel] Register merged diffusion agent kernels with KDA backend (#37385) Xiaoyu Zhang 2026-09-01 22:51:10 +08:00
• 4c7ff0d906 [CI] Double JIT kernel unit test timeout (#37435) Xiaoyu Zhang 2026-09-01 22:37:37 +08:00
• b6c06e1efb [DSA] Drop the redundant 512 from the top-k transform entry-point names (#36831) DarkSharpnessandClaude Opus 5 2026-09-01 22:37:03 +08:00
• 00689c0c94 [CI] Add entrypoint to hc_combine test for standalone execution (#37406) Even Zhou 2026-09-01 22:08:00 +08:00
• 3ae54c6ca2 test(npu): add DSV4-Flash / GLM-5.2 / Kimi-K3 gpqa accuracy cases (#37431) 3 people 2026-09-01 22:00:04 +08:00
• 4c2c169e6b [NPU]Strip padding before FIA kernel for vision encoder padded sequences (#36329) Estrella-xx 2026-09-01 19:28:19 +08:00
• 3103bc7462 [AMD][CI] Add daily ROCm 10 and ROCm 7.2 test coverage (#37409) YC Yen-Ching TsengandChen 2026-09-01 13:23:19 +02:00
• ce7e79b32c [AMD] Fix nightly ROCm 7.0 image build: patch missing <optional> include in AITER topk kernel (#36216) 5 people 2026-09-01 04:06:32 -07:00
• ed122ea984 [AMD] Enable topk v2 GLM ROCm (#36851) Eric.Chin.AMDandKingRei 2026-09-01 18:48:50 +08:00
• b425897366 [AMD] Gate the aiter memory-reserve exemption behind an env var (#37242) YC Yen-Ching Tseng 2026-09-01 12:46:15 +02:00
• 49db27528a fix(test): deflake zmq load-snapshot round-trip tests (#35787) datdo-msftandShangming Cai 2026-09-01 03:14:55 -07:00
• a77283fb02 [Rust] Rename mem-cache to sglang-radix-tree (#37290) Jialin Ouyang 2026-09-01 03:03:44 -07:00
• c16a8fc899 [NPU] [bugfix] Fix NPU MLA HiCache backup accessing missing data_ptrs. (#36813) chx96642264 2026-09-01 17:31:11 +08:00
• dc1ae02684 [Cookbook] Add the DFlash2 speculative option to GLM-5.3 (#37392) zijiexiaandClaude Opus 5 2026-09-01 02:08:28 -07:00
• b68702be99 [DSV4] hc-prenorm: fuse the combine step into a Triton kernel (#35118) eeecho 2026-09-01 17:04:05 +08:00
• 6c72b49a57 Revert "[AMD] Add GLM-5.3-Flash recipes for MI300X, MI325X, and MI355X (#36608)" (#37380) zijiexia 2026-09-01 01:25:13 -07:00
• 5edcd0a445 [FlashInfer V0.6.18] feat(dsv4): support --dsa-topk-backend flashinfer with fused top-k (#33237) Ziang Li 2026-09-01 01:18:10 -07:00
• 3484f7f836 [mem_cache] Add free_kv_row to release a request's kv row by row range (#36721) Liangsheng Yin 2026-09-01 01:14:43 -07:00
• 1c3ad92438 [Diffusion] Fuse FLUX.2 ModelOpt FP8 producers and QKV packing (#37162) Xiaoyu ZhangandCursor 2026-09-01 16:14:29 +08:00
• 379e33d87e [Cookbook] Add NVFP4 options for DeepSeek-V4 Flash Official (0731) and Pro Official (0813) (#37351) zijiexiaandClaude Fable 5 2026-09-01 01:03:39 -07:00
• 03b33cbe5d [CI] Fix hybrid wrapper test fake missing kv_index_translator (#37374) Mohammad Miadh Angkad 2026-09-01 15:55:05 +08:00
• 8a191554e3 [AMD] Enable 12-head MLA aiter fp8 Gluon decode (batched bh16bn128). (#34647) chien-an-chen 2026-09-01 14:47:45 +08:00
• 5e79110122 [Fix][CPU] fix xeon ci failure by test_qwen35_flashinfer_fusion (#37338) Ma Mingfei 2026-09-01 14:17:35 +08:00
• 60548501bb [Docs] Add NVFP4 section to GLM-5.3-Flash cookbook (#37109) Xinyuan Tong 2026-09-01 14:14:30 +08:00
• 959ca033eb refactor(hicache): simplify decode offload state bookkeeping (#37299) Liangsheng Yin 2026-08-31 23:09:41 -07:00
• ae2bd5728b [vlm] fix: contain multimodal feature transport failures (#37047) Mick 2026-09-01 13:46:38 +08:00
• 33ed29a0ee test: update hybrid attention runner fixtures (#37345) YAMY 2026-08-31 22:00:50 -07:00
• 458987b5ac [AMD][MORI] Bump MoRI to 7c51d18 for ionic RoCE dmabuf fix (#509) (#37286) Zhaoyi Li 2026-08-31 23:37:05 -05:00
• b21000aef1 [Unified Cache][4/N]: Add Mooncake backend for external linker (#37205) huangtingweiandhzh0425 2026-09-01 11:40:06 +08:00
• 3b14f37b74 [Fix] Use real ReqKvInfo in unit-test req mocks (#37339) Liangsheng Yin 2026-08-31 20:25:51 -07:00
• 60f881b40c [CI/NPU] Isolate multi-node tests by run_id to prevent concurrent-run… (#35500) hhhh1252023 2026-09-01 11:08:27 +08:00
• 5b04408784 [MoE] Add FlashInfer SM90 MXFP4 W4A8 CUTLASS MoE (#34967) Yuan Luoandluoyuan.luo 2026-09-01 11:04:41 +08:00
• 22337e9c56 fix(unified-memory): forward the KV-index translator through every wrapper backend (#37307) 3 people 2026-08-31 19:55:21 -07:00
• e6f21cdadc [Cohere Command-A-Plus] Optimize decode and BCG capture on SM10X (#36624) Mohammad Miadh Angkad 2026-09-01 10:37:47 +08:00
• 1591dcd91a [diffusion] fix: fix loading a block-FP8 quantized MiniMax-H3 DiT (#35703) Yiqi Yang 2026-09-01 10:33:58 +08:00
• 71cee04ebe [Diffusion] Optimize Qwen-Image TP collectives and attention (#36680) 3 people 2026-09-01 10:28:37 +08:00
• 562b661e0e [Feature] Megatron LayerNorm sequence parallelism (--enable-layernorm-sp) (#30915) Martin Hua 2026-08-31 19:27:28 -07:00
• 783af667fb xpu: record per-model metrics to jsonl for nightly dashboard (#36699) ashwini rathiandarathi-hlab 2026-09-01 07:50:56 +05:30
• f50b4ad7ae [DCP] Support decode context parallelism on the trtllm_mla decode path (#33926) 3 people 2026-09-01 09:19:44 +07:00
• 97744189b8 [Kernel] Raise shape limits in shared FLA and MoE kernels (ported from #36507) (#37317) Khoa Pham 2026-08-31 18:53:46 -07:00
• 9a85473a89 [FlashInfer v0.6.18] add FlashInfer CuTe DSL NVFP4 W4A16 mode (#35120) Ziang Li 2026-08-31 18:47:30 -07:00
• 175973d834 [Diffusion] Fuse Qwen-Image residual norm and NVFP4 quantization (#37129) Xiaoyu ZhangandCursor 2026-09-01 08:54:10 +08:00
• 079afaffb1 [Diffusion] Fuse FLUX.2 gated residual normalization on Blackwell (#37112) Xiaoyu ZhangandCursor 2026-09-01 08:38:33 +08:00
• 6715debb2a [Diffusion] Fuse Qwen-Image FP8 QKV projection and Blackwell epilogue (#37123) Xiaoyu ZhangandCursor 2026-09-01 08:34:41 +08:00
• 07c8f7294d Bump sgl-deep-gemm to 0.1.7 (#37279) Baizhou Zhang 2026-08-31 17:16:34 -07:00
• 455232de6e [Cookbook] Enable DSpark on the DeepSeek-V4 Flash Vision low-latency recipes (#37301) zijiexia 2026-08-31 16:43:54 -07:00
• ef9e58fd6d feat(unified-memory): three sub-pools for mamba + hybrid-SWA models (#35177) 4 people 2026-08-31 15:10:12 -07:00
• 98cb3535b7 feat(unified-memory): byte-budget sizing, feasibility floor, and a conservation verifier (#35158) 4 people 2026-08-31 15:09:28 -07:00
• 961beee9e5 fix(unified-memory): four boot/correctness fixes on the hybrid model paths (#35154) 4 people 2026-08-31 15:08:43 -07:00
• 88cf5c9541 [Cookbook] Add DeepSeek-V4-Flash-Vision-Exp to the DeepSeek-V4 page (#37293) zijiexia 2026-08-31 14:52:28 -07:00
• 2530204502 [mem_cache] Make release, row-reuse asserts, and presence checks read the KV record (#37167) Liangsheng Yin 2026-08-31 12:46:15 -07:00
• 95f0f41021 [CI] Move tests onto the right CI stages (#34074) Liangsheng YinandBaizhou Zhang 2026-08-31 12:39:59 -07:00
• 579270d459 [Rust] Simplify request defaults and document batch header ABI (#37226) Lianmin Zheng 2026-08-31 12:31:35 -07:00
• 2138494272 [Rust] Keep sampling and scheduler wire schemas in sync (#37222) Lianmin Zheng 2026-08-31 12:31:07 -07:00
• 48098b5f23 [Rust] Derive server address and accept signed env values (#37221) Lianmin Zheng 2026-08-31 12:30:40 -07:00
• 1da86b9801 [Rust] Split and rename embedded server components (#37220) Lianmin Zheng 2026-08-31 12:28:43 -07:00
• cf51650335 fix(config): retain pre-engine resolution declarations (#37195) ishandhanani 2026-08-31 14:24:29 -05:00
• 9e9d26a4af Fix Mooncake serving benchmark trace rows (#37201) Shuwen Wang 2026-09-01 02:43:59 +08:00
• 549166b819 fix(gateway): bump wfaas to 1.0.2 so ContinueNextStep unblocks dependents (#37249) Juhyun-Kim-Memphis 2026-09-01 03:30:34 +09:00
• 07d84ebd6d [2/N][Mixed] Mixed chunk prefill with spec enabled (#36933) Yuwei AnandClaude Fable 5 2026-08-31 10:48:07 -07:00
• 9cf157c252 [Radix Cache] Add Rust TreeCore backend with shared parity tests (#32710) 3 people 2026-08-31 09:26:20 -07:00
• 52e1c24744 [Diffusion] Fuse FLUX.2 token concatenation and NVFP4 quantization (#37141) Xiaoyu Zhang 2026-08-31 21:33:02 +08:00
• d60d658f5f [Kernel] Add GB300 Triton MoE configs for GLM-4.5 FP8 (#37159) Xiaoyu ZhangandSong Bian 2026-08-31 21:31:04 +08:00
• 771e613d96 [Diffusion] Fuse Qwen-Image final adaptive LayerNorm (#37144) Xiaoyu Zhang 2026-08-31 18:22:20 +08:00
• 1ed9bfac2c [Diffusion] Fuse Qwen-Image FP8 norm and activation quantization (#37156) Xiaoyu ZhangandCursor 2026-08-31 18:19:42 +08:00
• a874b83c2c test: re-enable DSV4-Flash W8A8 8p nightly perf cases (#37214) pllimax 2026-08-31 17:20:46 +08:00
• 63b2adbeac [NPU] Fix evalscope accuracy parsing and add glm5_1 aime26 request timeout (#36459) pllimax 2026-08-31 17:19:03 +08:00
• a53718e88c fix(ci): update attention backend test fixtures (#37219) YAMY 2026-08-31 02:15:17 -07:00
• 0674be736c [AMD] build gfx1250 release image from main (#37225) YC Yen-Ching Tseng 2026-08-31 11:04:04 +02:00
• 6580d5cd9a weight cache: key daemon paths by GPU UUID (#36101) 3 people 2026-08-31 15:44:45 +07:00
• 3865efc9f7 [AMD] support gfx1250 on ROCM 10 (#36871) +111 people 2026-08-31 10:19:11 +02:00
• 712a720c8a [KDA] Fused-accept state advance for FlashInfer KDA MTP verify (#33722) Yuan Luoandluoyuan.luo 2026-08-31 16:11:49 +08:00
• f61bb7b40a [unified-memory] Drop the vacated 'dense' qualifier and the restating comments (#37170) Cheng Wan 2026-08-31 00:54:13 -07:00