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 }}
AMD-AGI
Primus
Repository navigation
Code
Issues
8
(8)
Pull requests
62
(62)
Actions
Projects
Security and quality
Insights
More
items
All pull requests
Labels
14
(14)
Milestones
0
(0)
New pull request
Search pull requests
is
:
pr
state
:
open
is:pr state:open
Clear filter
Search
Pull requests
Open
62
(62)
Closed
1,400
(1,400)
Author
Label
Projects
Milestones
Reviews
Assignee
Sort by
Newest
descending
Sort…
Sort by Newest descending
More items
Comfortable display density
Compact display density
feat(megatron): add a tensorwise FP8 GPT output projection
#1497
·
alfuyao-amd
opened
Nov 19, 2026
Collaborator
·
·
1
perf(mi355x): tune Mixtral 8x7B FP8 pretrain config
#1496
·
vidushi8
opened
Nov 19, 2026
Collaborator
·
·
5
fix(megatron/moe): fused router ignores moe_router_pre_softmax=false
#1495
·
vidushi8
opened
Nov 19, 2026
Collaborator
·
·
6
[OOB Release] add JAX v26.10 dockerfile
#1494
·
hlira-amd
opened
Nov 19, 2026
Collaborator
·
·
1
feat(turbo): turbo_grouped_gemm_schedule=work_steal, DeepSeek-V3 proxy 5.0% faster with EP comm overlap on MI355X
#1492
·
boyang-amd
opened
Nov 19, 2026
Collaborator
·
3 of 5 tasks
·
·
1
[WIP] feat(megatron): MXFP8 through Primus-Turbo on MI455X (gfx1250) for DeepSeek-V3
ci:gpu
#1491
·
lhzhang333
opened
Nov 19, 2026
Collaborator
·
·
2
[OOB Release] add primus v26.9.1 dockerfile
#1488
·
hlira-amd
opened
Nov 19, 2026
Collaborator
·
·
1
perf(zaya1): speed up the ZAYA1 model and tune the MI325X BF16 and FP8 recipes based on MI355X recipes
ci:gpu
#1479
·
clairesonglee
opened
Nov 18, 2026
Collaborator
·
2/2
·
·
6
ci: add a gfx950 torch unit-test job on the MI355X runner
#1478
·
WangLingxun
opened
Nov 18, 2026
Collaborator
·
·
1
fix: make the unit tests pass on gfx950
ci:gpu
#1473
·
WangLingxun
opened
Nov 18, 2026
Collaborator
·
·
7
perf(automodel): return fresh tensors from FLUX single-stream blocks
ci:gpu
#1468
·
luiza-amd
opened
Nov 17, 2026
Contributor
·
·
9
perf(automodel): scale clipped gradients with one fused multiply
ci:gpu
#1467
·
luiza-amd
opened
Nov 17, 2026
Contributor
·
·
9
fix(megatron): apply the Turbo spec provider to gated-delta-net layer specs
ci:gpu
#1465
·
WangLingxun
opened
Nov 17, 2026
Collaborator
·
·
4
perf(megatron): MiniMax-M3 on MI455X: bit-identical step-time kernels
#1464
·
HuangWei-95
opened
Nov 17, 2026
Collaborator
·
·
29
fix(mlperf): restore multinode FLUX throughput on v26.9
#1456
·
zirui
opened
Nov 16, 2026
Collaborator
·
5 of 6 tasks
·
·
4
perf(megatron): tune Qwen3-30B-A3B and Qwen3.5-35B-A3B MI300X configs
ci:gpu
#1455
·
WangLingxun
opened
Nov 16, 2026
Collaborator
·
·
3
perf(kimi_k3): fused FlyDSL situ on gfx1250 with the router probability folded in, Kimi K3.1 Pro MI455X recipe 1.15x
#1451
·
boyang-amd
opened
Nov 15, 2026
Collaborator
·
4 of 5 tasks
·
·
1
feat(megatron): deferred_grad_nan_check — one host sync per step for the DDP grad NaN/Inf check
ci:gpu
#1448
·
Z-Y00
opened
Nov 14, 2026
Contributor
·
·
9
Projection: fix Megatron and TorchTitan benchmark-mode accuracy, and add hybrid-model support
ci:gpu
#1446
·
araina-amd
opened
Nov 13, 2026
Collaborator
·
·
16
chore(deps): bump tornado from 6.5.8 to 6.6.1 in /docs/sphinx
dependencies
python
#1444
·
dependabot[bot]
opened
Nov 13, 2026
Bot
·
·
chore(deps): bump soupsieve from 2.9.2 to 4.0 in /docs/sphinx
dependencies
python
#1441
·
dependabot[bot]
opened
Nov 13, 2026
Bot
·
·
feat(mistral_large4_text): add mistralai/Mistral-Large-4-1T-A52B
#1440
·
peymanr
opened
Nov 13, 2026
Collaborator
·
·
8
DO NOT MERGE -- docs(specforge): provide single-node training reference and fixture
ci:gpu
#1438
·
yudigege86
opened
Nov 12, 2026
Collaborator
·
·
1
perf(zaya1): speed up the ZAYA1 model and tune the MI355X BF16 and FP8 recipes
ci:gpu
#1430
·
clairesonglee
opened
Nov 11, 2026
Collaborator
·
1/2
·
15 of 16 tasks
·
·
14
Tuning agent: reach recompute and parallel layouts in the seed plan; hold selective recompute to the projected memory
#1428
·
araina-amd
opened
Nov 10, 2026
Collaborator
·
·
14
Previous
1
2
3
Next
You can’t perform that action at this time.