As far as I know WebGPU has a performance overhead, also he is using an experimental implementation. How do you explain the difference between the 2.6 TFlops (FP32) [1] of the M1 and the 950 GFlops (FP32) of the 855 [2]?
Thanks for pointing this out. This does put things in perspective. While it's true that achieving that in the browser is something, this sounds much less interesting and a lot more like the usual hype.
A passively cooled M1 beats a Ryzen 3900X, an i7, and an i9 in most benchmarks[0].
But if you want to pick this single benchmark so that you can conclude that the M1 is not better than a mobile phone SoC from a year ago, then you do you.
Some of those tests are realistically testing RAM speed more than CPU power. Yet he has the 3900x configured with only 1600Mhz memory clock compared to 2133Mhz on the M1.
The single core performance of these chips is very impressive, especially at such a TDP, but in raw multi-core CPU power it simply does not beat a 3900x.
But if you want to pick one set of benchmark tests so that you can conclude that the 3900X from two years ago is not better than an M1, then you do you.
> but in raw multi-core CPU power it simply does not beat a 3900x
I agree.
> if you want to pick one set of benchmark tests so that you can conclude that the 3900X from two years ago is not better than an M1
The thread was clearly about M1 vs mobile phone SoCs. I just gave some benchmarks to show that the M1 is not a mobile phone SoC but can instead compete with very decent laptop/desktop CPUs. The 3900X wasn't the main point. I even own a 5950X myself.
> you do you.
In all fairness: I shouldn't have used that phrasing.
