David Michael Barr

To aid prioritizing the many steps required to advance HBD performance, it would be informative to have at lease one `perf` trace of a short HBD encoding session.

I note that the `_dav1d` variants of the listed test failures also failed, recorded in the detailed log. This makes it unlikely to be a decoder bug and more likely...

I noticed in some of the debian patches that features were removed from the crate, mostly for features that have not been stabilised. However, we also have some features for...

This needs some additional testing for efficient configuration. We will also need to choose where to persist the corpus and coverage reports.

The logic is sound. I started an AWCY run on [objective-1-fast at default speed](https://beta.arewecompressedyet.com/?job=master-c113c00353e605612a41c8dbca124aca3ce2a5dc&job=fix-mv-rate%402022-08-27T07%3A39%3A11.029Z): | PSNR Y | PSNR Cb | PSNR Cr | CIEDE2000 | SSIM | MS-SSIM |...

I think the confounding factor is that this function is called for motion estimation on subsampled frames.

> @barrbrain I also saw your other commit ([barrbrain@5b45ccf](https://github.com/barrbrain/rav1e/commit/5b45ccf7e7417a779db8d95066b8804f977e042f)) This was a mistake, I shouldn't have changed the scale. I tried another experiment which gave [modest gains](https://beta.arewecompressedyet.com/?job=fix-mv-rate%402022-08-27T07%3A39%3A11.029Z&job=fix-mv-rate%402022-08-28T06%3A27%3A57.707Z) on top of...

> In addition, this will require a modern version of nasm to be installed (for example the version in the Ubuntu 20.04 APT repo is too old). As long as...

This is largely superseded by recent PRs, #3271 and #3277. There are a couple of places that are more difficult to resolve: - `src/asm/x86/transform/forward.rs` - `src/me.rs`

At this time, `cdef_dist_wxh_8x8` or its components likely top the list.