Skip to content

Pull requests: ROCm/ATOM

New pull request New
240 Open 1,667 Closed
240 Open 1,667 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

[Scheduler] Shortest-job-first prefill admission to cut p90 TTFT
#2158 opened Sep 7, 2026 by ganyi1996ppo Contributor Loading…
1 task
4
feat: add GLM-5.3-Flash MTP support
#2157 opened Sep 7, 2026 by cubezhang Contributor Loading…
1
bench: set ATOM_GC_THRESHOLD for the EPLB MegaMoE c=4096 case
#2156 opened Sep 7, 2026 by JiaoliangYu Contributor • Draft
1
[CI] upload AIPerf Perfetto traces from benchmark workflows
#2155 opened Sep 7, 2026 by MengqingCao Contributor • Draft
6 tasks
1
feat(kimi-k3): support P/D disaggregation and add the AgentX benchmark CI ci:atom
#2154 opened Sep 7, 2026 by wanzhenchn Contributor Loading…
1
[ATOM][Doc][PD]: add GLM-5.2 AgentX PD recipe
#2153 opened Sep 7, 2026 by MengqingCao Contributor Loading…
1 task
1
[CI] Add opt-in paired performance check for pull requests (ci:perf) ci:perf Run the paired PR performance check (merge-base vs head)
#2152 opened Sep 7, 2026 by zufayu Contributor • Draft
2
feat(dspark): capture the Kimi-K3 draft block into a CUDA graph ci:full
#2151 opened Sep 7, 2026 by whx-sjtu Contributor Loading…
1
docker: ROCm 10 base (pip SDK) + wheel release flow
#2150 opened Sep 7, 2026 by zufayu Contributor • Draft
4 tasks
1
feat(v4): DeepSeek-V4-Flash-Vision support ci:atom
#2149 opened Sep 7, 2026 by HaonanWang98 Contributor Loading…
1 task
1
[MTP] Fix negative pad-row KV length in draft decode metadata
#2148 opened Sep 7, 2026 by whx-sjtu Contributor Loading…
2
[MiniMax-M3][LMCache] Drive ATOM's byte-level KV offload from the vLLM plugin
#2146 opened Sep 6, 2026 by XiaobingSuper Contributor Loading…
6
[atom-vllm] LMCache offload for GLM-5.2
#2142 opened Sep 4, 2026 by kliuae Contributor Loading…
1 task
1
feat: add GLM-5.3-Flash multimodal support
#2141 opened Sep 4, 2026 by cubezhang Contributor Loading…
1
build(rust): commit Cargo.lock for atom/mesh — unblocks all image builds broken by tinyvec 1.13.0
#2138 opened Sep 3, 2026 by sunway513 Collaborator Loading…
3
[atom-vllm]: enable DSpark speculative decoding for DeepSeek-V4-Flash-0731
#2137 opened Sep 3, 2026 by peizhang56 Loading…
1 task done
1
2
[ATOMesh][CI][benchmark] add GLM-5.2 CPP4+DCP4 ATOMesh coverage
#2134 opened Sep 3, 2026 by MengqingCao Contributor Loading…
1 task
1
[NOT READY] refactor K3 LMCache
#2132 opened Sep 3, 2026 by zejunchen-zejun Collaborator • Draft
1
(recipe) add Kimi-K3 AgentX recipe
#2131 opened Sep 3, 2026 by qichu-yun Contributor Loading…
1 task
1
perf(kimi-k3): make the KDA temporal state dtype configurable
#2130 opened Sep 3, 2026 by XiaobingSuper Contributor Loading…
9
[ATOM Plugin CI] Fix the K3 DSpark DCP8 MLA head-width, capture-layout and cudagraph-break defects ci:vllm
#2129 opened Sep 3, 2026 by Phi-C Contributor Loading…
2
fix kimi k3 piecewise cudagraph ci:atom
#2126 opened Sep 3, 2026 by ganyi1996ppo Contributor Loading…
1 task
11
[PCP] Shard shared_experts on the folded TP grid, and sync sampled ids outside the speculative branch
#2125 opened Sep 3, 2026 by yitingw1 Contributor • Draft
1 task
1
fix(minimax-m3): unblock fp8 KV cache and EAGLE3 spec decode on the vLLM plugin ci:vllm
#2122 opened Sep 3, 2026 by PerryZhang01 Contributor Loading…
1 task
3
[ATOM][PD][DCP] Support MLA/Indexer cache transfer in P-CPP/D-DCP scenario ci:atom
#2121 opened Sep 3, 2026 by MengqingCao Contributor Loading…
1 task
7
ProTip! Filter pull requests by the default branch with base:main.