Throwing another log on the fire...
The M1 debate in the video may seem polite, but those are two Canadians, so it's essentially civil war.
I've only watched the first half of the video so far (and thanks for sharing it BTW!), but what I find frustrating is that Linus has two main concerns that are valid, but also have
already been significantly addressed less than 10 days after the public event.
Linus says he doesn't like Apple's vague, unlabeled graphs at the event, because:
- Nvidia and other companies provide clearer and more specific benchmarks and graphs in their keynotes; and
- Without saying exactly which i-series Intel chips they are comparing the M1 to - and without specifying the thermal implementation of the M1 - Apple is being misleading about performance.
But per the many videos and web links several of us have provided above, we've already seen multiple benchmarks and timed tests that provide a very specific and clear sense of how the M1 Macs measure up to Intel Macs. To the best of my knowledge, all the benchmarks can also be compared cross-platform to other manufacturer's computers as well (although of course in that case one is also comparing the optimization of the OS and the benchmark app itself as they interact with the processor, but that was always the case and has nothing to do with M1).
Similarly, there have already been tests that reveal how the combination of the one missing GPU core and the lack of active cooling impact MacBook Air performance compared to the MacBook Pro (and I presume the Mac mini, though I haven't studied the results on that model yet). The results are that the Air does throttle - minimally on shorter intensive tasks and more on longer ones, although it is worth noting that MaxTech's Cinebench testing showed that after the M1 thottles down the most, it actually throttles back up a bit in the later runs, suggesting that the M1's ability to be managed with passive cooling, while inferior to active cooling, is significantly superior to thermal management of the Intel chips it replaces - even though all those Intel chips are actively cooled. And keep in mind that based on tests so far, the M1 is thermal throttling at temps below 40 Celsius, while the Intel MacBook Air almost instantly pegs at 100 Celsius and stays there for the duration of the test, with the fan running full blast the entire time.
So Linus' complaint boils down to the fact that independent testing has to be done after Apple's keynotes in order to reveal the exact performance characteristics of their chips and hardware. To the extent he is suggesting that Apple's vagueness is hiding disappointing performance and therefore false or misleading performance claims by Apple, that is just FUD, since Apple's claims have already been shown to be substantially true.
And it's not like anyone - especially at a site like this - would look at
any manufacturer's benchmarks or performance claims and say, "Oh, they are nice and specific, so we don't need to independently test the chip." I'm not going to take any manufacturer's word for it, so what the manufacturer says at a keynote is, by itself, irrelevant to me unless they are actually lying or misleading, which all the evidence so far says Apple is not doing in this case.
Now yes, it is absolutely true that many of Apple's Intel Macs do throttle significantly, in a way that would make their performance claims misleading if they were relying on graphs comparing their performance to that of PCs with the same Intel chips inside them. But that's not Apple's approach or market. As several of the MacBook Air reviews have noted, the irony of the massive performance leap the M1 chip provides is that many, maybe even most, MacBook Air buyers (and probably a good number of mini and entry-level MB Pro buyers) aren't even going to know or care about it. They will just buy the machine because it's the cheapest Mac, and they will notice it's quite snappy, and that will be the end of it.
So IMHO much of the criticism of the M1 here is about style - folks don't like that Apple is catering to a customer base that is not, as a whole, focused on performance benchmarks as much as elements of the Windows and Linux customer base. And folks don't like that Apple's performance claims are vague enough to potentially conceal major "gotchas." Fair enough.
But there is a stubborn tendency so far for some folks to ignore the data we already have and insinuate that the M1 is actually an underwhelming performer, which is false. The Intel MacBook Air, some recent Intel MacBook Pros, and I believe some Intel iMacs,
are indeed underwhelming because of the thermals. But all evidence so far indicates that's simply not the case with the M1 - precisely because it is as low-wattage, cool-running, and power-efficient as Apple has designed it to be and claims it is. That's just what the evidence says so far, nothing more, nothing less.