4 ms·
There are some cool takeaways from this massive collection of benchmark results. But before we get to them, things we already know: * You cannot buy Apple Sil
by neogodless 4y ago
There are some cool takeaways from this massive collection of benchmark results.
But before we get to them, things we already know:
* You cannot buy Apple Silicon without buying a Mac (and you can't get the unique characteristics of a Mac and macOS without Apple Silicon)
* Apple Silicon is optimized for macOS (and vice versa)
* Apple Silicon is remarkably efficient (particularly under macOS)
This comparison, of course, is focused on Linux users. It is not going to determine some sort of ambiguously defined but ultimate superiority of silicon across all scenarios. However, if you use Linux on one of the laptops used to compare the CPUs, it can give you insight on performance for those specific use cases.
So what are the takeaways?
* The fan-less M2 (4p/4e) in the Air with 8GB performs about as well as an air-cooled 15-28W Ryzen 7 Pro 6850U (8p) with 16GB on average (and vice versa)
* The performance of the M2 is likely more than satisfactory despite Asahi Linux not being the absolute optimal operating system for the silicon
* The performance of the Ryzen is likely more than satisfactory despite lower efficiency compared to Apple Silicon
* Laptop APU performance in 2022 is really remarkable (even at these low TDPs), and any user (regardless of preferred OS or chosen CPU designer) will likely have a good experience with one of these high performing chips
I think running these benchmarks appeals to our curiosity about the state of hardware. It is not a perfect, controlled experiment, but it compares two real-world CPUs (APU and SoC respectively) and gives us insight into their performance given the actual hardware configurations tested.
What this does not tell us (again): that Ryzens are better than Apple Silicon, or vice versa, in all cases or in every metric. That if one CPU beats another CPU in some benchmarks, it's the one you want, regardless of all the other considerations you have to put into purchase decisions. That either of these CPUs are "bad".
- jeswin 4y ago> Apple Silicon is optimized for macOS (and vice versa) > Apple Silicon is remarkably efficient (particularly under macOS) It is actually quite possible that Apple's advantage was more about being on TSMC's 5nm before everyone else, than something intrinsic about their processor. > The fan-less M2 (4p/4e) in the Air with 8GB performs about as well as an air-cooled 15-28W Ryzen 7 Pro 6850U (8p) with 16GB on average (and vice versa) The air cooled M2 MBP doesn't do significantly better than the fan-less MBA; so the results will probably still stand.
- hu3 4y agoFurthermore, it would be nice if both M2 and AMD could use the same TSMC transistor density. As I understand Apple bought out TSMC's state-of-the-art production and competitors are left with less efficient nodes.
- hajile 4y agoA13 on the same 7nm as AMD still beat it out in performance per watt by a large margin. The Air throttles after just 5-10 minutes, so anything requiring consistent high-performance will be adversely affected by the passive cooling.
- nijave 4y agoI'm curious what Linux vs Android power efficiency looks like. I would think Apple's experience with mobile could translate back to laptop efficiency. It seems on the Linux front, things are a bit more mixed. Androids optimizations include things like app background usage limits that are harder to integrate in a Linux distro