Fix conv1d stream implementation hls directives #635
Merged
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Outdated suggestions cannot be applied.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.
Description
Fixed compiler directives for conv1d io_stream implementation. A loop was given the unroll directive, when it should have been pipelined. This loop is used to shift the kernel data when a new input is read by the stream. The new directives match those used in the conv2d implementation.
Type of change
Tests
The existing tests for conv1d still pass
Test Configuration:
To see the improvements of the fix, I just ran synthesis on a large conv1d layer. The layer's parameters in this test are 8 channels, 16 filters, length 4096, with a kernel size of 7 and a stride of 2. The fix reduces the latency from 635,968 cycles to 53,342 cycles when using a reuse factor of 1 and resource strategy.
Checklist