Christopher Berner
Christopher Berner
I removed most of the sparse additions, but the ones in phase 1 are still there. They were a relatively small amount of time before. They're next up on my...
Ah, actually the sparse additions in phase 1 weren't taking much time. I removed them and it improved throughput by ~2%. The profile shows 4.1% in `SparseBinaryMatrix::add_assign_row`, but seems that...
I tried the HDPC generation optimization you suggested, but wasn't able to make it work. Perhaps I implemented it wrong? https://github.com/cberner/raptorq/pull/101 For this part "set each column in every row...
Oh yes, of course, compute it recursively, very clever! Thanks, I'm adding this in
Nice. Another ~10% speedup from that HDPC optimization :) Looks like the next thing I need to do is revisit some of the matrix methods. Things like iterating through a...
I have a fork which I've been developing. I released 0.4.0 from my fork, since I needed several of the new features since 0.3.1. Feel free to use it, until...
@zargony any feedback on this PR? Happy to make changes if you'd like
Ok, that's fine. I'll follow #126 to see when it's merged, and then rebase on it. Sure, named feature flags sounds easier to use. It doesn't seem like it will...
@zargony just checking in. Any estimate when you will merge #126?
Ya, I'm happy to help out! The feature gates sound like an easy place for me to start, since I already looked at those a bit. I think it makes...