Moritz Borcherding
Moritz Borcherding
Fixed it. @HsuJv could you try the current state of the merged branch one more time just to confirm this?
Hi @snakehand I've been thinking about this a bit, sadly the paper you linked isn't available anymore (at least not under the link and a quick google only turns up...
Started doing preliminary work here, no simd yet just understanding the algorithm: https://github.com/KillingSpark/crc-rs/tree/clmul Interestingly enough this is ~2x faster than the current table-less implementation even without any real thought on...
Just as an idea from another project I contributed to: They offload the changelogs to the pull-request makers. They require an changelog entry with all merge requests that add features...
Any progress? I'd really love to see this release to get the performance improvements out there
Fwiw I tried compiling the webrtc crate with the 3.1.0-beta.1 which worked great. Did you find any blockers?
@akhilles is there anything I can do to help this move forward?
I see the concern and I agree the SIMD implementations don't clearly fall into an order like the current implementations. Even though it means that crates that do not update...
#115 looks like it would allow for relatively seamless integration of SIMD implementations. That's pretty cool!
This should include a change that checks the decoded FSE tables so that the symbols do match the allowed ranges for literal length, match length and offset which would avoid...