Arseny Kapoulkine

I'm reluctant to merge this because I feel like a better solution is to keep both thread id & name and display both. That way you don't have to deduplicate...

In your renderer, how do you define `projWidth`? Typically znear shouldn't affect the projection results, but if you define `projWidth`/`projHeight` as the view-space bounds of the camera slice at znear...

I don't think this is targeting the correct branch; it should target `master`.

It's been a while since I've seen this code. Can you explain why the order of query result retrieval and submission of the next frame was changed? fwiw I think...

`-cc` option should be fine to use as a default instead of `c`, gltfpack may drop the distinction long term. `-c`/`-cc` is complementary to `gltfpack.quantize`, it's recommended to use both...

> Projects like gltf-import-export have taken this to the limit and only offer these two options for output, nothing else. [gltfpack](https://github.com/zeux/meshoptimizer/tree/master/gltf) does the same; earlier versions supported .glb with external...

> the signed version should have a single instruction lowering to CVTTPS2DQ This is not the case, see #173.

Note that 64-bit conversions were dropped because they didn’t have an efficient implementation on *any* architecture, short of AVX512. What does lowering of unsigned conversions look like for ARM/Power?

I feel like AVX512 is not going to be practical for years to come as the dominant target ISA. The first laptop CPU that supports it shipped late last year....

Worth noting is that repeated shuffle masks should compress reasonably well. I've compiled both examples into a binary (extracting just "machine" code) and used zstd to compress these (zstd over...