View Single Post
chucker
 
Join Date: May 2004
Location: near Bremen, Germany
Send a message via ICQ to chucker Send a message via AIM to chucker Send a message via MSN to chucker Send a message via Yahoo to chucker Send a message via Skype™ to chucker 
2020-12-16, 17:26

Quote:
Originally Posted by kscherer View Post
Apple is crushing itself with the M1, and now they're crushing Windows on ARM, and by no small margin.
That headline is… well… technically accurate but kind of silly. What's being "smashed" is the utterly terrible Qualcomm SQ2 they put in the Surface Pro X. Windows is largely irrelevant in that comparison.

I don't know if Microsoft messed up or Qualcomm did or both of them, but hopefully, all these M1 headlines "inspire" someone at Microsoft and/or Qualcomm to put the foot on the gas again.

The M1 and SQ2 are actually amusingly simple to compare: they draw similar amounts of power, they both run at roughly 3.2 GHz, and they're both a 4+4 core setup. (edit) Oh, and both use TSMC as the manufacturer*.

(Side note: I hope some future Geekbench update will take these heterogenous setups into account. Even Intel is moving towards them, so listing cores as a singular integer number of "8 cores" just doesn't make sense any more.)

Core by core, the M1 is roughly 2.1 times as fast: it scores around in single-core on Geekbench, whereas the SQ2 scores around 800. On top of that, Apple's core scheduler seems to be more efficient (or their use of efficiency cores in addition to performance ones is more aggressive), as the factor is even higher for multiple cores, at about 2.4.

Qualcomm's phone CPUs are actually about 10% faster, which is not nothing, and they have to accomplish that with less thermal headroom, so I'm not sure why they do so poorly on the tablet CPU front. The explanation I come away with is they're not trying.

*) to be fair, the SQ2 is still 7nm, not 5nm. So that does give Apple some gains, but it doesn't come close to explaining the discrepancy.
  quote