Check each conv model for Winograd gains
Maintainers usually reply within 3 days
Nobody has claimed this yet.
Assessment
- Difficulty
- 5/5
- Estimated time
- Over a week
- Newbie friendliness
- 35/100
- Issue type
- Feature
- Clarity
- Mostly clear
- Activity status
- Active
- Tech stack
- typescript
- Domain
- machine-learning, performance
Research direction
Start with the convolution implementations and model definitions for YOLO26, MobileNetV4, DepthArt, and Moonshine's convolution front end, then identify the current kernel and how model timing is measured. Count qualifying 3×3 stride-1 convolutions and estimate their share of runtime and the gain from a Winograd route. Done means documenting the per-model estimates and deciding where adding the route is worthwhile; the issue names no specific files or tests.
Written by the indexing model from the issue text.
Description
We should check, model by model, whether the conv-based tasks (YOLO26, MobileNetV4, DepthArt, Moonshine's conv front end) would benefit from a Winograd conv route.
Winograd pays off for 3x3 stride-1 convs with enough channels; it does nothing for 1x1, depthwise or strided convs. So the answer depends on each model's layer mix, not on the op in general. YOLO is the obvious first candidate.
For each model:
- Count the convs that qualify (3x3, stride 1) and how much of the model's time they take.
- Estimate the gain against the current conv kernel.
- Decide whether a Winograd route is worth adding for that model.
- Dominant language
- TypeScript
- Stars
- 121
- Forks
- 7
- Avg merge
- 3d 12h
- Merged PRs (30d)
- 48
Getting set up
- No Dockerfile or Docker Compose file
- Has a pull request template
- No contributing guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
More from software-mansion/runntime
-
Difficulty 2/5 1-3 hours Newbie friendliness 55/100
software-mansion/runntime#121 ·
Maintainers usually reply within 3 days
-
core enhancement performance
Difficulty 5/5 Over a week Newbie friendliness 15/100
software-mansion/runntime#118 ·
Maintainers usually reply within 3 days
-
bug zoo
Difficulty 3/5 1-2 days Newbie friendliness 40/100
software-mansion/runntime#117 ·
Maintainers usually reply within 3 days
-
enhancement high priority
Difficulty 3/5 1-2 days Newbie friendliness 73/100
software-mansion/runntime#113 ·
Maintainers usually reply within 3 days
-
Difficulty 5/5 Over a week Newbie friendliness 45/100
software-mansion/runntime#103 · 1 reaction ·
Maintainers usually reply within 3 days
All issues in software-mansion/runntime
Similar issues
-
Difficulty 2/5 1-3 hours Newbie friendliness 75/100
Maintainers usually reply within 2 days
-
nip98: validateEvent skips the payload check for a string or bytes bodyMay be free again A pull request for this issue was closed without being merged. Open
Difficulty 2/5 1-3 hours Newbie friendliness 72/100
nbd-wtf/nostr-tools#599 ·
Maintainers usually reply within 1 day
-
bug
Difficulty 2/5 1-3 hours Newbie friendliness 68/100
LemmyNet/lemmy-ui#4442 · 1 comment ·
Maintainers usually reply within 1 day
-
Difficulty 2/5 1-3 hours Newbie friendliness 78/100
Zettlr/Zettlr#6664 · 1 comment ·
Maintainers usually reply within 1 day
-
Microsoft Outlook add attachment reads the binary field from item 0Possibly taken @joshbarros claimed this today. Openstatus:in-linear
Difficulty 2/5 1-3 hours Newbie friendliness 85/100
n8n-io/n8n#44021 · 2 comments ·
Maintainers usually reply within 1 day