3 ms·
Something is fishy here. The JPEG XL report [1] measured between 240-270 Megapixels/s on 6 cores using the C++ implementation (disclosure: I was responsible fo
by janwas 20d ago
Something is fishy here.
The JPEG XL report [1] measured between 240-270 Megapixels/s on 6 cores using the C++ implementation (disclosure: I was responsible for its SIMD/threading), about twice as fast as the then-current libaom.
Measuring on a single core is deeply misleading because our code was designed to scale well. I believe AVIF requires tiling in order to parallelize, which causes artifacts at tile boundaries.
[1]: https://arxiv.org/pdf/2506.05987 https://arxiv.org/pdf/2506.05987
- spider-mario 20d agoIt’s a measurement error: the `hyperfine` command uses `jxl_cli --speedtest` which, by default, does a warmup run that ends up in hyperfine’s measured time.
- xfs 20d agojxl-rs is single threaded. It may be not a good idea to use many threads per image as one page can have many images.
- jaffathecake 20d agojxl-rs is not single threaded.
- xfs 19d agojxl-rs as it is used in Chrome is single-threaded.