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
Type
/
to 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 }}
m96-chan
/
PyGPUkit
Public
Notifications
You must be signed in to change notification settings
Fork
0
Star
3
Code
Issues
51
Pull requests
0
Actions
Projects
Security and quality
0
Insights
Additional navigation options
Code
Issues
Pull requests
Actions
Projects
Security and quality
Insights
v0.2.20: Fused NN Kernels + Flash Attention 3 SM120 + FP8 Block-Scale MMA
- #193
#193
Merged
m96-chan
merged 23 commits into
main
m96-chan/PyGPUkit:main
from
release/v0.2.20
m96-chan/PyGPUkit:release/v0.2.20
Copy head branch name to clipboard
Jan 26, 2026
Conversation
Commits
23
(23)
Checks
Files changed
Merged
v0.2.20: Fused NN Kernels + Flash Attention 3 SM120 + FP8 Block-Scale MMA
#193
m96-chan
merged 23 commits into
main
m96-chan/PyGPUkit:main
from
release/v0.2.20
m96-chan/PyGPUkit:release/v0.2.20
Copy head branch name to clipboard
Commits
Commits on Jan 6, 2026
feat(ops): add native Conv1d CUDA kernel (#180)
Show description for 2411032
m96-chan
and
claude
committed
2411032
View commit details
Copy full SHA for 2411032
Browse repository at this point
Commits on Jan 12, 2026
feat(examples): add Llama Guard 3 content safety classifier
Show description for feb3304
m96-chan
and
claude
committed
feb3304
View commit details
Copy full SHA for feb3304
Browse repository at this point
Commits on Jan 16, 2026
feat(llm): add LLaMA 4 native CUDA kernels
Show description for 5fcf3c3
m96-chan
and
claude
committed
5fcf3c3
View commit details
Copy full SHA for 5fcf3c3
Browse repository at this point
chore: add security benchmark scripts to gitignore
m96-chan
committed
86c821b
View commit details
Copy full SHA for 86c821b
Browse repository at this point
chore: update cutlass submodule (alignment fix)
m96-chan
committed
9eb3a4b
View commit details
Copy full SHA for 9eb3a4b
Browse repository at this point
feat(attention): add Flash Attention 3 for SM120 (Blackwell)
Show description for ee6b1fe
m96-chan
and
claude
committed
ee6b1fe
View commit details
Copy full SHA for ee6b1fe
Browse repository at this point
feat(ops): add TMA utilities for SM90+ kernels
Show description for 1241b74
m96-chan
and
claude
committed
1241b74
View commit details
Copy full SHA for 1241b74
Browse repository at this point
wip(fa3): add TMA-enabled Flash Attention 3 kernel
Show description for adee44b
m96-chan
and
claude
committed
adee44b
View commit details
Copy full SHA for adee44b
Browse repository at this point
feat(attention): integrate TMA FA3 into SDPA dispatch
Show description for dcd1f8e
m96-chan
and
claude
committed
dcd1f8e
View commit details
Copy full SHA for dcd1f8e
Browse repository at this point
fix(fa3): resolve __syncthreads divergence causing kernel hang at scale
Show description for 5f763de
m96-chan
and
claude
committed
5f763de
View commit details
Copy full SHA for 5f763de
Browse repository at this point
refactor(fa3): parallelize softmax and fix consumer warp indexing
Show description for 028af30
m96-chan
and
claude
committed
028af30
View commit details
Copy full SHA for 028af30
Browse repository at this point
fix(fa3): resolve non-determinism in TMA FA3 attention kernel
Show description for a7c814c
m96-chan
and
claude
committed
a7c814c
View commit details
Copy full SHA for a7c814c
Browse repository at this point
wip(fa4): add Flash Attention 4 SM120 Phase 1 BF16 baseline
Show description for e3214b3
m96-chan
and
claude
committed
e3214b3
View commit details
Copy full SHA for e3214b3
Browse repository at this point
wip(fa4): Phase 2 NVFP4 Q@K^T external validation
Show description for b676995
m96-chan
and
claude
committed
b676995
View commit details
Copy full SHA for b676995
Browse repository at this point
bench(fa4): add Phase 3 full NVFP4 pipeline validation
Show description for b7f66c3
m96-chan
and
claude
committed
b7f66c3
View commit details
Copy full SHA for b7f66c3
Browse repository at this point
docs(fa4): add SM120 implementation report
Show description for d22576e
m96-chan
and
claude
committed
d22576e
View commit details
Copy full SHA for d22576e
Browse repository at this point
feat(fa3): add SM120 tuning configs with version selection
Show description for 7b99c62
m96-chan
and
claude
committed
7b99c62
View commit details
Copy full SHA for 7b99c62
Browse repository at this point
Commits on Jan 17, 2026
docs(fa3): document sync requirements in SM120 kernel
Show description for 1ac32c8
m96-chan
and
claude
committed
1ac32c8
View commit details
Copy full SHA for 1ac32c8
Browse repository at this point
feat(fp8): add native PTX inline assembly for FP8 block-scale MMA
Show description for dc3658f
m96-chan
and
claude
committed
dc3658f
View commit details
Copy full SHA for dc3658f
Browse repository at this point
feat(fa3): add FP8 block-scale MMA Flash Attention 3 for SM120
Show description for a9b75e2
m96-chan
and
claude
committed
a9b75e2
View commit details
Copy full SHA for a9b75e2
Browse repository at this point
Commits on Jan 26, 2026
feat(nn): add fused kernels for RMSNorm+Residual, SwiGLU, GeGLU
Show description for ce6de2b
m96-chan
and
claude
committed
ce6de2b
View commit details
Copy full SHA for ce6de2b
Browse repository at this point
fix(lint): resolve ruff lint errors in fused kernel files
Show description for bea759c
m96-chan
and
claude
committed
bea759c
View commit details
Copy full SHA for bea759c
Browse repository at this point
fix(mypy): add type annotation for scores_max in llama4.py
Show description for 0820572
m96-chan
and
claude
committed
0820572
View commit details
Copy full SHA for 0820572
Browse repository at this point
You can’t perform that action at this time.
