Make im2col default option for quartus #1010
Open
+4
−8
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Outdated suggestions cannot be applied.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.
A# Description
With using the Quartus backend, with conv2d with kernel size 3x3 or conv1d with kernel size 3, the framework makes the default implementation
Winograd
instad ofim2col
. As the currentWinograd
implementation does not seem to play nicely with low bitwidth and hard to achieve bit-accuracy (e.g., with automatic precision setting), I would suggest not enable it by default.For the Winograd impl, type
uint8_t
from<cstdint>
is used. This header is not included by default in many environments, explicit#include <cstdint>
is added to corresponding files.Type of change
Tests
Checklist
pre-commit
on the files I edited or added.