RFC for the MIOpen-hipDNN shim.#7388
Draft
NolanHannaAMD wants to merge 5 commits into
Draft
Conversation
github-actions
Bot
added
documentation
ci:docs-only
averinevg
reviewed
May 13, 2026
|
|
||
| ## 4. Proposed Design | ||
|
|
||
| ### 4.1 Build-time feature flag |
Contributor
There was a problem hiding this comment.
Would we want to consider a dynamic run-time flag for switching behaviours? This would probably allow for easier testing, and would allow the customer to revert back to the pre-shim behaviour just in case they hit an issue after we switched over to routing the MIOpen api to hipDNN.
Contributor
Author
There was a problem hiding this comment.
Yes, we have been discussing that and will likely modify the RFC to reflect this plan after investigations are complete.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Motivation
Frameworks like PyTorch, TensorFlow, ONNX Runtime, and JAX/XLA already have substantial integration with MIOpen — call sites, tuning DBs, build infrastructure, CI coverage. There is external pushback against going all-in on a hipDNN backend at this juncture: the migration work for those framework teams is non-trivial, the timing is bad for several of them, and hipDNN is still maturing. At the same time, we want MIOpen consumers to start benefiting from hipDNN's engine ecosystem (performance work, fusion graphs, new architectures) without waiting for that migration. A forwarding wrapper lets us route selected calls through hipDNN behind the existing MIOpen API, decoupling the rollout of hipDNN-backed execution from any framework-side rework.