Skip to content
Navigation Menu
Sign in
Products
Platform
GitHub
Repository hosting, code review, and collaboration
Copilot
AI coding agent for tasks, projects, and automated work
Build & Ship
Actions
CI/CD and automation
Code Quality
Quality analysis, fixes, and enforcement
Codespaces
Cloud development environments
Secure
Code Security
Vulnerability detection and remediation
Secret Protection
Secret scanning and leak prevention
All Features
Developers
Learn
Documentation
Training
Certifications
Build
Build with Copilot
Customize Copilot
Integrations
Explore
Trending repos
GitHub Sponsors
Developer articles
Changelog
Changelog
Your source for the latest features, improvements, and fixes across GitHub products, all in one place.
Businesses
By Stage
Enterprises
Team
Startups
Nonprofits
By Industry
Healthcare
Financial services
Manufacturing
Government
See all industries
Evaluate GitHub
Why GitHub
Customer Stories
Business insights
Trust Center
Add-ons
GitHub Advanced Security
Enterprise-grade code security
Copilot for business
Enterprise-grade AI features
Premium Support
Enterprise-grade 24/7 support
Resources
Learn & Connect
Documentation
Resource center
Blog
Events & webinars
Community Forum
Customer Stories
Integrations
Support
Customer support
Training
GitHub Status
Company
About
Partners
Changelog
GitHub Universe
GitHub Universe
The event for developers, by developers — talks, workshops, and what’s next for GitHub.
Pricing
Search
/
Sign in
Sign up
Appearance settings
You signed in with another tab or window.
Reload
to refresh your session.
You signed out in another tab or window.
Reload
to refresh your session.
You switched accounts on another tab or window.
Reload
to refresh your session.
Dismiss alert
{{ message }}
NVIDIA
cudnn-frontend
Repository navigation
Code
Issues
150
(150)
Pull requests
118
(118)
Discussions
Actions
Security and quality
Insights
More
items
All pull requests
Labels
38
(38)
Milestones
1
(1)
New pull request
Search pull requests
is
:
pr
state
:
open
is:pr state:open
Clear filter
Search
Pull requests
Open
118
(118)
Closed
2,265
(2,265)
Author
Label
Projects
Milestones
Reviews
Assignee
Sort by
Newest
descending
Sort…
Sort by Newest descending
More items
Comfortable display density
Compact display density
perf(linear_attention): GDN takes the value-width split for one in-place sequence at every length below the chain
area:frost
area:linear_attention
cat-enhancements
mod-frost
orig-nv-eng
#2852
·
RomanAnders90
opened
Nov 21, 2026
Contributor
·
·
·
Frontend 1.34.0
9
sdpa fwd (sm80): dense d256 single-row decode takes the paged d256 split rule
area:frost
area:global_attention
cat-enhancements
orig-nv-eng
#2851
·
YangXu1990uiuc
opened
Nov 21, 2026
Collaborator
·
5 tasks done
·
·
5
feat(gemm): generalize dense GEMM + SwiGLU to gemm_glu_forward with act_func; remove the SwiGLU-only API
area:gemm
cat-feature
orig-nv-eng
#2850
·
hwanseoc
opened
Nov 21, 2026
Member
·
5 tasks done
·
·
·
Frontend 1.34.0
5
QSA backward: cheaper deterministic=True on the dense prefix
area:sparse_attention
cat-enhancements
mod-cutedsl
orig-nv-eng
#2849
·
adnios
opened
Nov 21, 2026
Contributor
·
5 tasks done
·
·
·
Frontend 1.34.0
3
[CPP Launcher] Linear attention: GDN-2, GDP, Mamba-2 and the summary ops skip the custom-op dispatch in eager mode
#2847
·
hwanseoc
opened
Nov 21, 2026
Member
·
·
1
sdpa fwd (sm90): placement for paged windowed decode and non-dividing-group MTP
area:frost
area:global_attention
cat-enhancements
orig-nv-eng
#2846
·
YangXu1990uiuc
opened
Nov 21, 2026
Collaborator
·
5 tasks done
·
·
·
Frontend 1.34.0
7
samples(models): the attention preparation on its entry, cudnn.AttentionPrepareDecode, over a paged attention cache
area:frost
area:linear_attention
cat-feature
orig-nv-eng
#2845
·
RomanAnders90
opened
Nov 21, 2026
Contributor
·
·
·
Frontend 1.34.0
9
Graph API: every engine takes its device and stream from the graph
area:frost
area:gemm
area:linear_attention
cat-bugfix
orig-nv-eng
#2844
·
YangXu1990uiuc
opened
Nov 21, 2026
Collaborator
·
·
·
Frontend 1.34.0
23
Cover wide SM107 D256 output strides in retained CUDA Graph replay
area:frost
area:global_attention
cat-bugfix
orig-nv-eng
#2842
·
YangXu1990uiuc
opened
Nov 21, 2026
Collaborator
·
·
·
Frontend 1.34.0
18
test(linear_attention): fix cached compile counts and Mamba-2 batch checks
area:linear_attention
cat-bugfix
orig-nv-eng
#2841
·
Anerudhan
opened
Nov 21, 2026
Collaborator
·
·
·
Frontend 1.34.0
4
test(sdpa): pin the natural walk on the SM120 wide-head split record
area:frost
area:global_attention
cat-bugfix
mod-frost
orig-nv-eng
#2837
·
yanzhuo607
opened
Nov 20, 2026
Collaborator
·
5 tasks done
·
·
·
Frontend 1.34.0
5
test(sdpa): MXFP8 unaligned-K decline cell asserts a decline with the guard lifted
area:frost
area:global_attention
cat-ci
orig-nv-eng
#2836
·
yanzhuo607
opened
Nov 20, 2026
Collaborator
·
5 tasks done
·
·
·
Frontend 1.34.0
5
gemm(frost projection decode): acquire the split-K reduction barrier at cluster scope
#2835
·
bastianhagedorn
opened
Nov 20, 2026
Contributor
·
3 of 5 tasks
·
·
1
Declare packed Stats stride support per SDPA implementation and validate SM120 split plans
area:frost
area:global_attention
cat-enhancements
orig-nv-eng
#2832
·
YangXu1990uiuc
opened
Nov 20, 2026
Collaborator
·
·
·
Frontend 1.34.0
19
Prefer dual KV streams for medium SM100 FP8 paged decode
area:frost
area:global_attention
cat-perf-bug
orig-nv-eng
#2829
·
YangXu1990uiuc
opened
Nov 20, 2026
Collaborator
·
·
·
Frontend 1.34.0
3
Fold the fixed load-map mode into D256 THD split attention
area:frost
area:global_attention
cat-perf-bug
orig-nv-eng
#2828
·
YangXu1990uiuc
opened
Nov 20, 2026
Collaborator
·
·
·
Frontend 1.34.0
3
Synchronize the FROST FP32 gate tile exchange
area:frost
area:linear_attention
cat-bugfix
mod-cutedsl
orig-external
#2823
·
icavan
opened
Nov 20, 2026
Contributor
·
·
·
Frontend 1.34.0
7
Isolate fused GDN reader tickets across concurrent streams and graph replays
area:frost
area:linear_attention
cat-bugfix
mod-cutedsl
orig-external
#2821
·
icavan
opened
Nov 20, 2026
Contributor
·
·
·
Frontend 1.34.0
6
Support per-row start offsets in DSA IndexerTopK
area:sparse_attention
cat-feature
mod-cutedsl
op:DSA
orig-external
#2817
·
icavan
opened
Nov 20, 2026
Contributor
·
4 of 5 tasks
·
·
·
Frontend 1.34.0
1
11
fix(dsa): acknowledge H16 dKV reads before MMA reuses TMEM
area:sparse_attention
cat-bugfix
mod-cutedsl
op:DSA
orig-external
#2816
·
icavan
opened
Nov 20, 2026
Contributor
·
·
·
Frontend 1.34.0
6
fix(dsa): acquire the H16 dKV4 pipeline stage before publication
area:sparse_attention
cat-bugfix
mod-cutedsl
orig-external
#2815
·
icavan
opened
Nov 20, 2026
Contributor
·
·
·
Frontend 1.34.0
2
fix(dsa): publish the H16 TMEM pointer before warp roles diverge
area:sparse_attention
cat-bugfix
mod-cutedsl
orig-external
#2814
·
icavan
opened
Nov 20, 2026
Contributor
·
·
·
Frontend 1.34.0
2
fix(dsa): converge dense indexer TMEM allocation waits
area:sparse_attention
cat-bugfix
mod-cutedsl
orig-external
#2813
·
icavan
opened
Nov 20, 2026
Contributor
·
·
·
Frontend 1.34.0
7
test(misc): accept the Python 3.13 argparse help for --max-tokens / --max-new-tokens
area:packaging
cat-ci
mod-infra
orig-external
#2812
·
0z5a
opened
Nov 20, 2026
Contributor
·
4 of 5 tasks
·
·
·
Frontend 1.34.0
6
sdpa(bwd/sm80): launch the packed Delta, dQ cast and dK/dV reduce on a 2-D grid
area:frost
cat-bugfix
mod-cutedsl
orig-external
#2811
·
0z5a
opened
Nov 20, 2026
Contributor
·
4 of 5 tasks
·
·
·
Frontend 1.34.0
6
Previous
1
2
3
4
5
Next
You can’t perform that action at this time.