 Cheng Wan
|
6fdcb9934c
|
fix(runner): size eager static buffers for prefill budget and MLP-sync autotune (#28677)
|
2026-06-19 13:23:26 -07:00 |
|
|
Cheng Wan
|
2aa7b58aa7
|
refactor(runner): reuse a prepared static buffer for every dummy run (#28740)
|
2026-06-19 13:16:07 -07:00 |
|
|
Cheng WanandClaude Opus 4.8
|
856b0dc74b
|
refactor(runner): move kernel warmup into the shared runner lifecycle (warmup()) (#28739)
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
2026-06-19 13:13:13 -07:00 |
|
|
Cheng WanandClaude Opus 4.8
|
d705a91de1
|
refactor(runner): add EagerRunner, own the eager path, polymorphic dispatch (#28386)
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
2026-06-19 13:05:43 -07:00 |
|
|
|
ab0714d0ee
|
ci: run GB300 nightly suite in the standard Nvidia nightly workflow (#28536)
Co-authored-by: Baizhou Zhang <sobereddiezhang@gmail.com>
Co-authored-by: Mohammad Miadh Angkad <176301910+mmangkad@users.noreply.github.com>
|
2026-06-19 12:24:38 -07:00 |
|
|
Michael
|
7c505c2927
|
[AMD] fix(jit): port kv_canary write/verify/plan kernels to ROCm (#28357)
|
2026-06-19 12:19:37 -07:00 |
|
|
|
c436a8161a
|
[AMD] Enable HiSparse on ROCm (#26639)
Co-authored-by: clintg6 <7388379+clintg6@users.noreply.github.com>
Co-authored-by: HAI <hixiao@gmail.com>
|
2026-06-19 11:59:45 -07:00 |
|
|
Jaybe
|
ca88b7f1d2
|
fix: remove manual rope parameters injection in PretrainedConfig (#23910)
|
2026-06-19 17:41:51 +00:00 |
|
|
Mohammad Miadh Angkad
|
88c261c3f3
|
Fix IndexCache PP topk handoff (#28532)
|
2026-06-19 23:39:09 +08:00 |
|
|
Oguz Ulgen
|
3af991fb3e
|
[AMD] Make breakable CUDA graph run on ROCm/HIP (#28173)
|
2026-06-19 07:16:00 -07:00 |
|
|
Liangsheng Yin
|
9bb9d17e1a
|
[Spec] Unify speculative grammar token-accept path in decode processing (#28682)
|
2026-06-19 02:49:40 -07:00 |
|
|
shuwenn
|
941a11ada3
|
[HiCache] refactor: remove unused transfer buffer (#26003)
|
2026-06-19 09:46:04 +00:00 |
|
|
Cheng Wan
|
a6db86d535
|
refactor(runner): split BaseRunner (shared) from BaseCudaGraphRunner (#28385)
|
2026-06-19 02:04:15 -07:00 |
|
|
Cheng Wan
|
1c6331cbd6
|
refactor(runner): rename runner replay/load/can_run for the shared surface (#28384)
|
2026-06-19 01:45:04 -07:00 |
|
|
Xiaoyu Zhang
|
31c0a98066
|
[codex] Update diffusion skills for latest main (#28711)
|
2026-06-19 16:37:59 +08:00 |
|
|
Mick
|
af2ec2a0dd
|
[diffusion] perf: merge LTX-2 stage-1 distilled LoRA into the base in original mode (#28594)
|
2026-06-19 15:41:47 +08:00 |
|
|
|
59eb142ec2
|
[NPU] [Diffusion] Performance Optimization for LTX-2 Model (#22445)
Co-authored-by: gengjinsong <gengjinsong@huawei.com>
Co-authored-by: gengjinsong <904939979@qq.com>
Co-authored-by: ronnie_zheng <zl19940307@163.com>
|
2026-06-19 08:58:45 +03:00 |
|
|
Xiaoyu Zhang
|
3b417d3999
|
Revert "[Diffusion] FLUX: fuse FeedForward GELU into up-proj GEMM (cublasLt epilogue)" (#28708)
|
2026-06-19 13:44:57 +08:00 |
|
|
Xiaoyu ZhangandClaude Opus 4.8
|
7da92e1112
|
[diffusion] Sana: pack self-attn q/k/v and cross-attn k/v into single GEMMs (#28393)
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
|
2026-06-19 13:25:52 +08:00 |
|
|
billishyahao
|
24d15dd92e
|
[AMD][DSV4] fix nonetype issue when enabling hicache (#28541)
|
2026-06-18 21:26:48 -07:00 |
|
|
Mohammad Miadh Angkad
|
c7397de571
|
Use Marlin for SM120 MXFP4 MoE (#28231)
|
2026-06-18 19:19:41 -07:00 |
|
|
giang_ng_tr
|
b36360dc5b
|
[AMD][Perf] Split-KV flash-decode attention for EAGLE target-verify (Triton backend) (#27382)
|
2026-06-18 19:11:10 -07:00 |
|
|
Lucia Fang
|
1c8551169d
|
Add opt-in CUDA-graph capture-trace export (#28551)
|
2026-06-18 18:51:14 -07:00 |
|
|
huangtingweiandhzh0425
|
6b7ecca663
|
[HiCache]Support hybrid pool staged H2D kernel (#28434)
Co-authored-by: hzh0425 <hzh0425@apache.org>
|
2026-06-19 09:48:03 +08:00 |
|
|
Brayden ZhongandBrayden Zhong
|
05ee93c44f
|
Remove redundant cast and copy in calling trtllm_fp8_block_scale_moe (#28555)
Co-authored-by: Brayden Zhong <brayden@radixark.ai>
|
2026-06-18 18:11:29 -07:00 |
|
|
Rain Jiang
|
ef01618dfb
|
support MPServer and embedded server for granian to enable muti tokenizer worker (#28573)
|
2026-06-18 17:59:12 -07:00 |
|
|
Brayden ZhongandBrayden Zhong
|
ea407df4b0
|
Use Flashinfer allreduce fusion for MNNVL allreduce for Nemotron (#28346)
Co-authored-by: Brayden Zhong <brayden@radixark.ai>
|
2026-06-18 17:30:12 -07:00 |
|
|
giang_ng_tr
|
c1067f88d6
|
[AMD][Perf] Tune extend attention block sizes for gfx950 (head_dim > 128) (#27793)
|
2026-06-18 15:16:52 -07:00 |
|
|
Baizhou Zhang
|
e3026ef016
|
[3/N][CP] Implement zigzag CP strategy (#28421)
|
2026-06-18 15:10:30 -07:00 |
|
|
|
9fc9d37f6d
|
Fix spec decoding with grammar in disagg (#24082)
Co-authored-by: jimmy.shong <jimmy.shong@radixark.ai>
Co-authored-by: Jimmy Shong <69131491+Jiminator@users.noreply.github.com>
Co-authored-by: Xinyuan Tong <xinyuantong.cs@gmail.com>
Co-authored-by: Xinyuan Tong <115166877+JustinTong0323@users.noreply.github.com>
Co-authored-by: Codex <codex@example.com>
|
2026-06-18 14:58:32 -07:00 |
|
|
|
bea282cede
|
[DeepSeek-V4] Fuse UE8M0 scale rounding into FP8 group quantization (#26766)
Co-authored-by: liqichao <liqichao@baidu.com>
Co-authored-by: yhyang201 <yhyang201@gmail.com>
|
2026-06-18 14:41:56 -07:00 |
|
|
Cheng Wan
|
f83e4d5968
|
refactor(runner): unify eager-forward DP/MLP-sync padding into one helper (#28383)
|
2026-06-18 13:41:30 -07:00 |
|
|
Qiaolin Yu
|
2411737244
|
[spec decoding] fully overlap spec decoding for hybrid linear attention backend (#28579)
|
2026-06-18 13:31:12 -07:00 |
|
|
DovLin
|
cf0afe3223
|
[FA]Add lost params in fa varlen func (#28642)
|
2026-06-18 13:30:20 -07:00 |
|
|
Yi ZhangandKe Bao
|
66a7fd5c0b
|
refactor: mamba radix cache server args initialize (#28151)
Co-authored-by: Ke Bao <ispobaoke@gmail.com>
|
2026-06-18 13:02:36 -07:00 |
|
|
Liangsheng Yin
|
8f6d9ef9a5
|
[misc] Drop redundant req_pool_indices_cpu guards; fold hisparse into GLM-5.1 e2e (#28607)
|
2026-06-18 12:51:23 -07:00 |
|
|
Lawrence Wu
|
792cb3a5d0
|
fix: add missing guard for use_jit_ep_activation (#27377)
|
2026-06-18 12:51:09 -07:00 |
|
|
Jimmy Shong
|
97e3b8998d
|
Pass quant_config to attention gate projection (#28649)
|
2026-06-18 20:04:25 +08:00 |
|
|
syy-hw
|
bb9d31f22d
|
[NPU] Add head_dim=256 to _can_use_tnd whitelist (#28635)
|
2026-06-18 17:55:50 +08:00 |
|
|
Estrella-xx
|
2a9cce5d27
|
[NPU] Add MTP support for GLM-4.7-Flash (#28516)
|
2026-06-18 17:54:06 +08:00 |
|
|
Xiaoyu Zhang
|
b7d7dfb4ed
|
[Bugfix] Fix Intern-S1 FP8 expert count lookup (#28629)
|
2026-06-18 17:49:59 +08:00 |
|
|
Cheng Wan
|
67db2ac3e7
|
refactor(runner): unify pp_proxy_tensors forward kwarg into one helper (#28382)
|
2026-06-18 02:23:47 -07:00 |
|
|
cctry
|
fa71064147
|
fix: speculative draft worker clobbering target attention backend (#28559)
|
2026-06-18 01:21:14 -07:00 |
|
|
Mohammad Miadh Angkadandshuwenn
|
59001267c3
|
Fix bench serving base-url-only runs (#28617)
Co-authored-by: shuwenn <47200617+alphabetc1@users.noreply.github.com>
|
2026-06-18 00:34:00 -07:00 |
|
|
|
9b10821c8e
|
[NPU] Add Ascend NPU support for DeepSeek-V4 (#25144)
Co-authored-by: khalil2ji3mp6 <khalilzhk@gmail.com>
Co-authored-by: randgun <kelonlu@163.com>
Co-authored-by: t00937989 <tanlei33@huawei.com>
|
2026-06-18 15:30:25 +08:00 |
|
|
Vladislav NosivskoyandZhangheng
|
b7ae7149e8
|
[HiCache] Fix SWA L3 cache miss due to a prefetch/hit len mismatch (#27291)
Co-authored-by: Zhangheng <hzh0425@apache.org>
|
2026-06-18 14:55:00 +08:00 |
|
|
kangwangamdandkangwangamd
|
6309fb9abb
|
[AMD] Fix Always mask padded topk_ids on HIP to prevent garbage MoE routing (DeepSeek-R1-MXFP4 accuracy regression) (#28378)
Co-authored-by: kangwangamd <kangwangamd@users.noreply.github.com>
|
2026-06-17 23:39:44 -07:00 |
|
|
Mick
|
3b61dc32c9
|
[diffusion] fix: fix fp8 fused tp scale loading (#28546)
|
2026-06-18 14:38:38 +08:00 |
|
|
cctry
|
7976928c57
|
Abort during chunked prefill + PD peer-liveness abort (#28086)
|
2026-06-17 23:13:28 -07:00 |
|
|
Jimmy Shong
|
d2539980b6
|
[Fix] don't force hybrid-SWA when sliding_window is disabled (#28604)
|
2026-06-17 22:11:27 -07:00 |
|