Josh Holmer

In addition to rebasing, should there be a comment about squashing commits? Do we enforce squashing commits?

I'd like some tests/benchmarks/etc. I tested a similar change for jemalloc a while ago, and found that it provides no significant performance change (at least on Linux), so opted to...

It seems like something isn't quite happy here: https://travis-ci.org/github/xiph/rav1e/builds/748500930

Re "We have a branch for calling sad vs satd." that branch is likely being elided by the compiler since we always inline this function. Other than that these seem...

Top 10 time consuming functions, from a 720p 4:2:0 10-bit test clip (10 frames): Speed 6: | | | | | | | |------------------------------------------------------------|----------|--------|------------------------------------------------------------------|-------------|---------------| | Function / Call Stack |...

I've discovered uftrace seems to generally be more accurate than Intel VTune. Here's a speed 6 trace of top 10 functions by self time from current master: ``` Total time...

Ah, sure, here is what I used: ``` $ uftrace record --no-libcall -- target/release/rav1e ~/Downloads/KristenAndSara_42010p.y4m -o /dev/null $ uftrace report --sort self ```

It's been a year since we checked in on this, here's the current top priority functions for potential HBD ASM as tested at various speeds on a i7-9700. Percents are...

It seems that, for me at least, this is working on a per-file basis. i.e. if I open a file, then close vspreview and reopen the same file, the dimensions...

Ah, I see what the other part of the issue is now... for HD videos, if the video is wider than the size of the window you have saved, it...