Skip to content
Navigation Menu
Sign in
Appearance settings
Platform
AI CODE CREATION
GitHub Copilot
Write better code with AI
GitHub Copilot app
Direct agents from issue to merge
MCP Registry
Integrate external tools
DEVELOPER WORKFLOWS
Actions
Automate any workflow
Codespaces
Instant dev environments
Issues
Plan and track work
Code Review
Manage code changes
Code Quality
Enforce quality at merge
APPLICATION SECURITY
GitHub Advanced Security
Find and fix vulnerabilities
Code security
Secure your code as you build
Secret protection
Stop leaks before they start
EXPLORE
Why GitHub
Documentation
Blog
Changelog
Marketplace
View all features
Solutions
BY COMPANY SIZE
Enterprises
Small and medium teams
Startups
Nonprofits
BY USE CASE
App Modernization
DevSecOps
DevOps
CI/CD
View all use cases
BY INDUSTRY
Healthcare
Financial services
Manufacturing
Government
View all industries
View all solutions
Resources
EXPLORE BY TOPIC
AI
Software Development
DevOps
Security
View all topics
EXPLORE BY TYPE
Customer stories
Events & webinars
Ebooks & reports
Business insights
GitHub Skills
SUPPORT & SERVICES
Documentation
Customer support
Community forum
Trust center
Partners
View all resources
Open Source
COMMUNITY
GitHub Sponsors
Fund open source developers
PROGRAMS
Security Lab
Maintainer Community
Accelerator
GitHub Stars
Archive Program
REPOSITORIES
Topics
Trending
Collections
Enterprise
ENTERPRISE SOLUTIONS
Enterprise platform
AI-powered developer platform
AVAILABLE ADD-ONS
GitHub Advanced Security
Enterprise-grade security features
Copilot for Business
Enterprise-grade AI features
Premium Support
Enterprise-grade 24/7 support
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 }}
Uh oh!
There was an error while loading.
Please reload this page
.
NVIDIA
/
Megatron-LM
Public
Notifications
You must be signed in to change notification settings
Fork
4.4k
Star
17.6k
Code
Issues
393
Pull requests
820
Discussions
Actions
Projects
Security and quality
0
Insights
Additional navigation options
Code
Issues
Pull requests
Discussions
Actions
Projects
Security and quality
Insights
[2/2] Wiring cuDNN fused DSA kernels support with THD, CP and IndexShare (GLM5.2)
- #5099
#5099
Merged
HollowMan6
merged 16 commits into
NVIDIA:main
NVIDIA/Megatron-LM:main
from
HollowMan6:dsa_cp_thd_cudnn
HollowMan6/Megatron-LM:dsa_cp_thd_cudnn
Copy head branch name to clipboard
Jul 8, 2026
Conversation
Commits
16
(16)
Checks
Files changed
Merged
[2/2] Wiring cuDNN fused DSA kernels support with THD, CP and IndexShare (GLM5.2)
#5099
HollowMan6
merged 16 commits into
NVIDIA:main
NVIDIA/Megatron-LM:main
from
HollowMan6:dsa_cp_thd_cudnn
HollowMan6/Megatron-LM:dsa_cp_thd_cudnn
Copy head branch name to clipboard
Commits
Commits on Jun 30, 2026
Add cuDNN fused DSA kernels
Show description for f90d43c
HollowMan6
committed
f90d43c
View commit details
Copy full SHA for f90d43c
Browse repository at this point
Address review issues
Show description for ca9beca
HollowMan6
committed
ca9beca
View commit details
Copy full SHA for ca9beca
Browse repository at this point
Clean up
Show description for 118fb09
HollowMan6
committed
118fb09
View commit details
Copy full SHA for 118fb09
Browse repository at this point
More clean up
Show description for fb31f60
HollowMan6
committed
fb31f60
View commit details
Copy full SHA for fb31f60
Browse repository at this point
nits
Show description for b2ae36c
HollowMan6
committed
b2ae36c
View commit details
Copy full SHA for b2ae36c
Browse repository at this point
Commits on Jul 1, 2026
fix memory and perf regression
Show description for e881503
HollowMan6
committed
e881503
View commit details
Copy full SHA for e881503
Browse repository at this point
Resume TP=1 (no SP) special case perf
Show description for 98fcc90
HollowMan6
committed
98fcc90
View commit details
Copy full SHA for 98fcc90
Browse repository at this point
Commits on Jul 2, 2026
Top-k sharing used the split hook use packed-sequence metadata
Show description for 8c19e4c
HollowMan6
committed
8c19e4c
View commit details
Copy full SHA for 8c19e4c
Browse repository at this point
Commits on Jul 4, 2026
update test cases
Show description for db27798
HollowMan6
committed
db27798
View commit details
Copy full SHA for db27798
Browse repository at this point
Commits on Jul 5, 2026
padding small local head counts
Show description for 7c8acf8
HollowMan6
committed
7c8acf8
View commit details
Copy full SHA for 7c8acf8
Browse repository at this point
clean up
Show description for 8fbed38
HollowMan6
committed
8fbed38
View commit details
Copy full SHA for 8fbed38
Browse repository at this point
skip ranking when K covers every score column
Show description for 35efdad
HollowMan6
committed
35efdad
View commit details
Copy full SHA for 35efdad
Browse repository at this point
checks indexer K against valid local CP lengths
Show description for 318fc1c
HollowMan6
committed
318fc1c
View commit details
Copy full SHA for 318fc1c
Browse repository at this point
Preserve packed query offsets for SP without CP
Show description for 8138d87
HollowMan6
committed
8138d87
View commit details
Copy full SHA for 8138d87
Browse repository at this point
Commits on Jul 6, 2026
perf and mem optimization
Show description for 9832712
HollowMan6
committed
9832712
View commit details
Copy full SHA for 9832712
Browse repository at this point
Commits on Jul 8, 2026
Mark 'cutlass.cute.core' has no attribute 'ThrMma' error case as flaky
Show description for 962f6e9
HollowMan6
committed
962f6e9
View commit details
Copy full SHA for 962f6e9
Browse repository at this point
You can’t perform that action at this time.
