Apple M1 latencies, throughput and iops
dougallj.github.ioAlready discussed here: https://news.ycombinator.com/item?id=26375139
Those are two different pages with two different sets of information from the same person.
The other one was linking to the overview, the first page of the series on that specific core and that specific SoC. Discussions on HN are often interesting, and it is not ideal to fragment them just because the work is spread across 6 pages. It's like linking to the 3rd page of a 10-pages article.
In any case, I wasn't voicing an opinion on the re-post (absolutely nobody is interested in that), just pointing out that there was already a discussion on this, from a couple of days ago.
It's a sneeless dupe by HN's Theory of Dupes:
https://hn.algolia.com/?dateRange=all&page=0&prefix=false&so...
Lots of volatility in the snee market.
repo for the GPU: https://github.com/dougallj/applegpu
So now were programming against unofficial specs? Great, Apple! /s
Although this is interesting and I'm always impressed by what people can reverse engineer, I kinda wish the techies would just give up this battle.
I know M1 is the latest hotness, but Apple wants to own everything and won't stop until we're all plugged into our iAnal probe and paying to store the temperature readings on iCloud.
Just let them have their user-hostile yearly subscription iTunes appstore disposable product island, and work on something where the company actually helps you to use their stuff in the way you want.