Consolidating DSP-based MVU with Automated Flexible Laning #1398
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Outdated suggestions cannot be applied.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.
Adopt automated flexible laning in DSP-based MVU implementation with all design featuring soft vectorization. This only leaves the implementation mode leveraging the INT8 mode of DSP58 explicitly diverting to an alternative kernel implementation.
Leveraging the gained flexibility, the instantiation of the DSP-based MVU in FINN is simplified by eliminating previously needed bitwidth-dependent decisions on the kernel selection.