Readback results from HVM into Rust through interop #389
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Outdated suggestions cannot be applied.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.
This allows the resulting nets to be read back directly from Rust, without requiring each implementation to print it's networks in error-prone ways.
TODO: The CUDA version is still quite slow due to copying the entire
GNet
into the host from the device. The way to fix it is to compact the global network's memory and only copy the relevant portion before copying it into RAM.