> The time budget for high quality engineers working on this will also be limited as there's no market that I think of, only itch scratching. That's why I star

> The time budget for high quality engineers working on this will also be limited as there's no market that I think of, only itch scratching. That's why I started this project backed by a Patreon. I knew this couldn't be accomplished via itch scratching; it has to be a job (not quite full time yet, but a good chunk thereof). FWIW: the userspace GPU driver is now passing 90% or so of the GLES tests running on macOS; only the kernel side is not started yet. The ML cores are a niche use case and broadly not useful for anyone not doing machine learning on Linux (but some people are interested in them and have started taking a look). And power management already works to a significant extent; in the next week or so I'll be submitting v3 of the PMGR driver and v2 of the cpufreq driver for upstreaming, and starting work on SMC (needed for battery/charging info). And the security chip (SEP) has already been poked at; the mailbox interface to it is shared with other chips and has already been upstreamed, and the protocol on top isn't terribly complicated. It's mostly a matter of building a little driver to give userspace access to it, and writing some tooling to use it.