Skip to content

Pull requests: ROCm/mori

New pull request New
40 Open 490 Closed
40 Open 490 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

[AMD][DSV4] fix: recognize fp4_blockwise in the AUTO tuning config
#600 opened Aug 24, 2026 by karverma-amd Contributor Loading…
1 task
tune(ep/v2): give fp8 dispatch its own gfx1250 EP4 row
#598 opened Aug 24, 2026 by jhchouuu Collaborator Loading…
fix(EPv2): fix the hardcoded ep size
#597 opened Aug 24, 2026 by kawhil-amd Contributor Loading…
perf(umbp): cut per-range overhead in local BatchGetRanges
#596 opened Aug 24, 2026 by isytwu Collaborator Loading…
1 task done
@isytwu
feat(cco): add Triton device API bindings
#594 opened Aug 24, 2026 by yangyuhuiling Contributor • Draft
1 task
feat(EPv2): forward a per-token scale row with dispatch
#593 opened Aug 24, 2026 by jhchouuu Collaborator Loading…
1
bench(cco): add a one-operation P2P latency probe, rename xgmi_bw to ualoe_metric
#592 opened Aug 24, 2026 by jhchouuu Collaborator Loading…
feat(shmem): MORI_ENABLE_RAIL_ONLY to restrict RDMA QPs to same-rail peers
#591 opened Aug 22, 2026 by jhchouuu Collaborator Loading…
5 of 6 tasks
feat(umbp): peer-local multi-backend placement and logical tier management
#587 opened Aug 20, 2026 by wuyl1 Contributor Loading…
1
perf(ep): optimize IntraNode dispatch kernel for MI350X
#586 opened Aug 19, 2026 by kudomcho Contributor Loading…
6 tasks done
Don't merge: AI NIC perf collection
#585 opened Aug 19, 2026 by amirakb89 Contributor • Draft
1 task
MORI IO improvements 2nd try
#583 opened Aug 19, 2026 by pemeliya Contributor Loading…
1
feat(umbp): GPU Direct Storage (hipfile) SSD-to-GPU read path
#579 opened Aug 19, 2026 by isytwu Collaborator Loading…
5 tasks done
@isytwu
feat(ep/v2): TDM dispatch transport for the flydsl backend
#578 opened Aug 18, 2026 by XingerZhu Loading…
1 task
Io prepared transfers v3
#574 opened Aug 17, 2026 by amirakb89 Contributor Loading…
4 tasks
5
Io cpp bench default
#573 opened Aug 17, 2026 by amirakb89 Contributor Loading…
6
test bnxt ci
#570 opened Aug 17, 2026 by QizhouZhang97 Contributor Loading…
Feat/env check gpu memory
#566 opened Aug 14, 2026 by amirakb89 Contributor Loading…
6
EP support for rail isolated topology through Host based RDMA
#558 opened Aug 13, 2026 by itej89 Loading…
1 task
1
feat(umbp): support the UMBP tree connector on a distributed backend
#547 opened Aug 12, 2026 by maning00 Contributor Loading…
allocator: register mori as a torch SymmetricMemory backend
#544 opened Aug 11, 2026 by carlushuang Collaborator • Draft
@jhchouuu
refactor(umbp): make distributed mode backend- and transport-agnostic
#540 opened Aug 11, 2026 by TianDi101 Collaborator Loading…
perf(EP): tune EPv1 MI300X EP8 for DeepSeek-V4-Pro / Kimi-K3, key con…
#539 opened Aug 11, 2026 by kudomcho Contributor Loading…
1 task
perf(cco): preserve global address space in LSA pointer wrapper
#537 opened Aug 10, 2026 by yangyuhuiling Contributor Loading…
5 tasks done
chore(EPv2): split dispatch/combine tuning schedules into independent tables
#533 opened Aug 9, 2026 by kawhil-amd Contributor Loading…
ProTip! Filter pull requests by the default branch with base:main.