fw-ai / DeepGEMM Public

forked from deepseek-ai/DeepGEMM

Notifications You must be signed in to change notification settings
Fork 0
Star 0

Code
Pull requests 6
Actions
Projects
Security and quality
Insights

Additional navigation options

Code
Pull requests
Actions
Projects
Security and quality
Insights

Pull requests: fw-ai/DeepGEMM

Labels 9 Milestones 0

New pull request New

6 Open 10 Closed

Author

Filter by author

Uh oh!

There was an error while loading. Please reload this page.

Label

Filter by label

Uh oh!

There was an error while loading. Please reload this page.

Use alt + click/return to exclude labels

or ⇧ + click/return for logical OR

Projects

Filter by project

Uh oh!

There was an error while loading. Please reload this page.

Milestones

Filter by milestone

Uh oh!

There was an error while loading. Please reload this page.

Reviews

Filter by reviews

No reviews Review required Approved review Changes requested

Assignee

Filter by who’s assigned

Assigned to nobody

Uh oh!

There was an error while loading. Please reload this page.

Sort

Sort by

Newest Oldest Most commented Least commented Recently updated Least recently updated Best match

Most reactions

Pull requests list

mega_moe: reserve 2 SMs of headroom (grid = num_sms - 2) — original yingz/mega-sm-headroom branch

#14 opened Sep 9, 2026 by mayinghan

Loading…

feat(mega): add fused shared side-LoRA kernels

#9 opened Aug 13, 2026 by morgendave

Loading…

Add SM103 FP8 block128 MegaMoE training backend

#8 opened Jul 21, 2026 by renfeichen-fw • Draft

mega-MoE prefill: cycle-chunked dispatch pull (ring < num_min)

#4 opened Jul 6, 2026 by ipiszy • Draft

fix: cast fp32 block scales to UE8M0 before packing on SM100

#3 opened Jun 27, 2026 by ipiszy

Loading…

1 task

Packed MXFP4 & NVFP4 GEMM + mega-MoE (SM100, true 2-CTA multicast)

#2 opened Jun 26, 2026 by ipiszy • Draft

ProTip! Add no:assignee to see everything that’s not assigned.