Implement ConvInteger operator
#566
Merged
Conversation
This file contains 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
robertknight
force-pushed
the
conv-integer
branch
10 times, most recently
from
4270f88 to
29a7121
Compare
Implement the ConvInteger operator for all combinations of int8 input signed-ness, with the caveat that combinations other than (input=int8, weights=uint8) will result in a shift-conversion of the input and/or weights to `input=int8`, `weights=uint8` which maps to the currently supported GEMM kernels. Depthwise convolution for int8 uses a naive conversion of the f32 depthwise kernel and both have significant room for optimization.
robertknight
force-pushed
the
conv-integer
branch
from
29a7121 to
ad266bd
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
Implement the ConvInteger operator.
This follows the implementation of
Convfor f32, with the addition of zero point inputs and the removal of the bias input.In the process fix handling of padding elements in int8 im2col packing when conversion from i8 -> u8 is enabled.
TODO:
u8 x i8 -> i32GEMM support, but this is not a super useful combination. ORT only supports input=u8, weights=u8 and there are issues open to support input=u8, weights=i8.ConvIntegerweights are not in the safe range (i7/u7). Validate weights at load time? (deferred to Improve handling of weights with a non-reduced range inConvInteger#574)