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
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
.
ROCm
/
aiter
Public
Notifications
You must be signed in to change notification settings
Fork
572
Star
562
Code
Issues
325
Pull requests
643
Discussions
Actions
Projects
Security and quality
0
Insights
Additional navigation options
Code
Issues
Pull requests
Discussions
Actions
Projects
Security and quality
Insights
[HIP] [OPUS] Unify OPUS GEMM/BMM interfaces and use Torch workspaces
- #4961
#4961
Open
Fyzyukk
wants to merge 69 commits into
ROCm:main
ROCm/aiter:main
from
Fyzyukk:opus_kernel_workspace_management_and_interface_refactor
Fyzyukk/aiter:opus_kernel_workspace_management_and_interface_refactor
Copy head branch name to clipboard
Conversation
Commits
69
(69)
Checks
Files changed
Open
[HIP] [OPUS] Unify OPUS GEMM/BMM interfaces and use Torch workspaces
#4961
Fyzyukk
wants to merge 69 commits into
ROCm:main
ROCm/aiter:main
from
Fyzyukk:opus_kernel_workspace_management_and_interface_refactor
Fyzyukk/aiter:opus_kernel_workspace_management_and_interface_refactor
Copy head branch name to clipboard
Commits
Commits on Aug 21, 2026
[OPUS] Reapply Task1 and Task2 on upstream/main
Show description for ebf718f
Fyzyukk
committed
ebf718f
View commit details
Copy full SHA for ebf718f
Browse repository at this point
[OPUS] Consolidate task documentation
Fyzyukk
committed
e2002dc
View commit details
Copy full SHA for e2002dc
Browse repository at this point
[OPUS] Document Task1 and Task2 validation workflow
Fyzyukk
committed
977dd7a
View commit details
Copy full SHA for 977dd7a
Browse repository at this point
[OPUS] Preserve upstream MXScale tuned config
Fyzyukk
committed
6aca635
View commit details
Copy full SHA for 6aca635
Browse repository at this point
[OPUS] Guard C ABI priming on tensor device
Fyzyukk
committed
d18b804
View commit details
Copy full SHA for d18b804
Browse repository at this point
[OPUS] Drop Task1 and Task2 validation docs
Fyzyukk
committed
79456dd
View commit details
Copy full SHA for 79456dd
Browse repository at this point
[OPUS] Optimize Task2 Python dispatch hot paths
Fyzyukk
committed
995f343
View commit details
Copy full SHA for 995f343
Browse repository at this point
[OPUS] Adapt upstream gfx1250 split-K fixes (#4246)
Show description for 76beb3c
Fyzyukk
committed
76beb3c
View commit details
Copy full SHA for 76beb3c
Browse repository at this point
[OPUS] Document gfx942 and gfx1250 validation plan
Fyzyukk
committed
8f12a00
View commit details
Copy full SHA for 8f12a00
Browse repository at this point
[OPUS] Optimize A16 C ABI launch path
Show description for 17d895d
Fyzyukk
and
cursoragent
committed
17d895d
View commit details
Copy full SHA for 17d895d
Browse repository at this point
[OPUS] Isolate A16W16 caller policy
Show description for be54e9b
Fyzyukk
committed
be54e9b
View commit details
Copy full SHA for be54e9b
Browse repository at this point
[OPUS] Split public GEMM and BMM adapters
Fyzyukk
committed
bce5c47
View commit details
Copy full SHA for bce5c47
Browse repository at this point
[OPUS] Consolidate launch policy and repair tune routes
Fyzyukk
committed
5f7adab
View commit details
Copy full SHA for 5f7adab
Browse repository at this point
Minimize OPUS integration changes
Fyzyukk
committed
059b2b4
View commit details
Copy full SHA for 059b2b4
Browse repository at this point
Remove generated HTML reports
Fyzyukk
committed
366c8a9
View commit details
Copy full SHA for 366c8a9
Browse repository at this point
Reconcile OPUS branch with latest upstream
Fyzyukk
committed
aa295b1
View commit details
Copy full SHA for aa295b1
Browse repository at this point
Trim redundant OPUS tests and benchmarks
Fyzyukk
committed
c7034f9
View commit details
Copy full SHA for c7034f9
Browse repository at this point
Focus OPUS tests on representative regressions
Fyzyukk
committed
b621a6c
View commit details
Copy full SHA for b621a6c
Browse repository at this point
Drop redundant OPUS interface tests
Fyzyukk
committed
a081f33
View commit details
Copy full SHA for a081f33
Browse repository at this point
Optimize A16 eager dispatch and remove legacy workspace registry
Fyzyukk
committed
4bf3bfd
View commit details
Copy full SHA for 4bf3bfd
Browse repository at this point
Add OPUS validation coverage and update documentation
Fyzyukk
committed
a3b7632
View commit details
Copy full SHA for a3b7632
Browse repository at this point
Add MI308X gfx942 OPUS validation guide
Fyzyukk
committed
9e7ae8e
View commit details
Copy full SHA for 9e7ae8e
Browse repository at this point
Commits on Aug 24, 2026
Merge remote-tracking branch 'upstream/main' into opus_kernel_workspace_management_and_interface_refactor
Fyzyukk
committed
2af893d
View commit details
Copy full SHA for 2af893d
Browse repository at this point
Merge remote-tracking branch 'upstream/main' into opus_kernel_workspace_management_and_interface_refactor
Fyzyukk
committed
da3d008
View commit details
Copy full SHA for da3d008
Browse repository at this point
Remove MI308X OPUS validation guide
Fyzyukk
committed
5c7adcd
View commit details
Copy full SHA for 5c7adcd
Browse repository at this point
Remove branch-only OPUS test files
Fyzyukk
committed
4df3275
View commit details
Copy full SHA for 4df3275
Browse repository at this point
Fix OPUS lint failures
Fyzyukk
committed
95d7f01
View commit details
Copy full SHA for 95d7f01
Browse repository at this point
Fix gfx950 MXFP8 split-K Cvoid compilation
Fyzyukk
committed
8cf1198
View commit details
Copy full SHA for 8cf1198
Browse repository at this point
Format MXFP8 split-K regression test
Fyzyukk
committed
f6268f4
View commit details
Copy full SHA for f6268f4
Browse repository at this point
Fix no-argument OPUS A16W16 test entry
Fyzyukk
committed
7a43edb
View commit details
Copy full SHA for 7a43edb
Browse repository at this point
Commits on Aug 26, 2026
Fix OPUS compatibility and tuning paths
Fyzyukk
committed
d9e0f7a
View commit details
Copy full SHA for d9e0f7a
Browse repository at this point
Commits on Aug 27, 2026
[OPUS] Integrate PR #4903 gfx1250 CO kernels with exact-kid dispatch
Show description for 2f847d1
Fyzyukk
committed
2f847d1
View commit details
Copy full SHA for 2f847d1
Browse repository at this point
[OPUS] Remove obsolete dsv4pro BF16 tuning config
Show description for 92b9626
Fyzyukk
committed
92b9626
View commit details
Copy full SHA for 92b9626
Browse repository at this point
[OPUS] Align BF16 tuning tables with PR #4903
Show description for e3d9dd8
Fyzyukk
committed
e3d9dd8
View commit details
Copy full SHA for e3d9dd8
Browse repository at this point
Merge upstream/main into opus workspace refactor
Fyzyukk
committed
6df2316
View commit details
Copy full SHA for 6df2316
Browse repository at this point
Commits on Aug 31, 2026
[OPUS] Address PR 4961 must-fix review
Show description for 534a0fc
Fyzyukk
committed
534a0fc
View commit details
Copy full SHA for 534a0fc
Browse repository at this point
[OPUS] Keep family modules in the flat package
Show description for d59aa48
Fyzyukk
committed
d59aa48
View commit details
Copy full SHA for d59aa48
Browse repository at this point
[OPUS] Address PR 4961 should-fix review
Show description for 41bdd49
Fyzyukk
committed
41bdd49
View commit details
Copy full SHA for 41bdd49
Browse repository at this point
[OPUS] Fix Python style checks
Fyzyukk
committed
341813a
View commit details
Copy full SHA for 341813a
Browse repository at this point
Commits on Sep 2, 2026
Merge branch 'main' into opus_kernel_workspace_management_and_interface_refactor
Fyzyukk
authored
c422310
View commit details
Copy full SHA for c422310
Browse repository at this point
Fix MXFP8 BMM policy review issues
Fyzyukk
committed
92b36fb
View commit details
Copy full SHA for 92b36fb
Browse repository at this point
Commits on Sep 3, 2026
Address OPUS A16W16 review feedback
Fyzyukk
committed
2cbd16c
View commit details
Copy full SHA for 2cbd16c
Browse repository at this point
Refresh OPUS CO source provenance
Show description for 0aa9c71
Fyzyukk
committed
0aa9c71
View commit details
Copy full SHA for 0aa9c71
Browse repository at this point
Simplify OPUS tuned GEMM integration
Fyzyukk
committed
cc7af9d
View commit details
Copy full SHA for cc7af9d
Browse repository at this point
Remove OPUS A16W16 C ABI launch path
Fyzyukk
committed
1e41b66
View commit details
Copy full SHA for 1e41b66
Browse repository at this point
[FlyDSL] Remove stale TopK availability checks (#5210)
Show description for e986c56
2 people
authored and
Fyzyukk
committed
e986c56
View commit details
Copy full SHA for e986c56
Browse repository at this point
Merge upstream/main into opus workspace refactor
Fyzyukk
committed
515f062
View commit details
Copy full SHA for 515f062
Browse repository at this point
Commits on Sep 4, 2026
Drop unrelated setup.py comment change
Fyzyukk
committed
61466d8
View commit details
Copy full SHA for 61466d8
Browse repository at this point
Revert "Drop unrelated setup.py comment change"
Fyzyukk
committed
6665c85
View commit details
Copy full SHA for 6665c85
Browse repository at this point
Drop comment-only packaging diffs
Fyzyukk
committed
4aa1ead
View commit details
Copy full SHA for 4aa1ead
Browse repository at this point
Commits on Sep 7, 2026
Remove SHA markers from OPUS policy parity test
Fyzyukk
committed
cf0d9bc
View commit details
Copy full SHA for cf0d9bc
Browse repository at this point
Merge upstream/main into opus workspace refactor
Show description for ec73ed5
Fyzyukk
committed
ec73ed5
View commit details
Copy full SHA for ec73ed5
Browse repository at this point
Fix gfx942 split-K launch validation for short K
Fyzyukk
committed
5e5869d
View commit details
Copy full SHA for 5e5869d
Browse repository at this point
Size gfx942 workspace from converged split-K
Fyzyukk
committed
3c61785
View commit details
Copy full SHA for 3c61785
Browse repository at this point
Commits on Sep 8, 2026
Add OPUS A8W8 tuning and exact BMM replay
Fyzyukk
committed
ac6ea5c
View commit details
Copy full SHA for ac6ea5c
Browse repository at this point
Gate OPUS A8W8 tuner to gfx950
Fyzyukk
committed
a65d5ff
View commit details
Copy full SHA for a65d5ff
Browse repository at this point
Commits on Sep 9, 2026
Merge upstream/main and adapt gfx1250 OPUS split-K fixes
Show description for 6ea8d5b
Fyzyukk
committed
6ea8d5b
View commit details
Copy full SHA for 6ea8d5b
Browse repository at this point
Keep OPUS CO build tooling and artifacts aligned with upstream
Show description for e30142f
Fyzyukk
committed
e30142f
View commit details
Copy full SHA for e30142f
Browse repository at this point
Restore upstream OPUS JIT configuration
Show description for a05fe49
Fyzyukk
committed
a05fe49
View commit details
Copy full SHA for a05fe49
Browse repository at this point
Commits on Sep 10, 2026
[OPUS] Merge main and address CO batch and split-K review
Show description for 1f24221
Fyzyukk
committed
1f24221
View commit details
Copy full SHA for 1f24221
Browse repository at this point
Commits on Sep 12, 2026
[OPUS] Fix launch validation and compatibility follow-ups
Show description for fa54e3f
Fyzyukk
committed
fa54e3f
View commit details
Copy full SHA for fa54e3f
Browse repository at this point
[OPUS] Consolidate redundant regression coverage
Show description for 416e92b
Fyzyukk
committed
416e92b
View commit details
Copy full SHA for 416e92b
Browse repository at this point
Commits on Sep 13, 2026
[OPUS] Merge main and preserve launch validation
Show description for 50b844a
Fyzyukk
committed
50b844a
View commit details
Copy full SHA for 50b844a
Browse repository at this point
[OPUS] Simplify A16W16 merge resolution
Show description for 9a985ef
Fyzyukk
committed
9a985ef
View commit details
Copy full SHA for 9a985ef
Browse repository at this point
[OPUS] Remove A16 tuned sweep from test entry
Fyzyukk
committed
2bcf559
View commit details
Copy full SHA for 2bcf559
Browse repository at this point
[CI] Retrigger failed workflows
Fyzyukk
committed
753854f
View commit details
Copy full SHA for 753854f
Browse repository at this point
[OPUS] Merge latest upstream main
Fyzyukk
committed
22a2d22
View commit details
Copy full SHA for 22a2d22
Browse repository at this point
Commits on Sep 14, 2026
[OPUS] Simplify and activate test coverage
Fyzyukk
committed
a15009c
View commit details
Copy full SHA for a15009c
Browse repository at this point
Commits on Sep 15, 2026
[OPUS] Merge latest upstream main
Show description for 7dc86ed
Fyzyukk
committed
7dc86ed
View commit details
Copy full SHA for 7dc86ed
Browse repository at this point
You can’t perform that action at this time.
