Naive question: if the team behind this is essentially the same folks who pioneered Apple's A/M series, does that mean that basically there's basically very few people in the world who are capable of designing these high-performance SoCs?
When I tried to skim what was so "special" about the apple M1 secret sauce, older HNs posts said that it was basically just a well-executed design (large RoB, beefy cache, wide pipelines, etc.) using the latest tsmc node. But even in the years since then it doesn't seem like any other companies have managed to design something comparable, and it basically took this team of ex-apple folks to produce something similar.
The short answer is: Yes. The most best and most talented chip designers (think the top 0.01%+) - especially if you put together the right team of them - are orders of magnitude better. Would you have 1 Magnus Carlsen or 100 average chess players? Jordan, Pippen, Rodman, et al. or a bunch of college players?
Chip design is one of those fields where there is no skill ceiling to how good you can be, so having the best people means you can crush the competition.
The run of the mill engineer is indeed quite underpaid, but thats mostly because there is a mountain of gruntwork required to develop a modern chipset even with all of the automation tools out there. The engineers architecting the next generation of chips (think Jim Keller and the like) have millions or tens of millions in stock options.
It all sounds very interesting, but five more days -- then we'll see reviews and find out how much is hype, and how well they executed. (And how well Microsoft is handling translation.)
The laptop is limited to 2.5 GHz (possibly for battery life or heat reasons) meaning the benchmark scores are far below what Qualcomm was showing off in prerelease press coverage.
> What microsoft has done to make the transition as smooth as possible?
Building the software translation (i.e. emulation) layer and testing / validating the top 20 or so most popular "apps": social media, spotify, office, etc. I think most of these are ARM native.
> What are game developers told to do to make their games compatible?
Qualcomm gave a GDC talk about this, though I can't seem to find it now. There are a handful of options, including just running it via x86 emulation. You can also recompile for ARM directly, or use some kind of hybrid solution where some system-related components use native ARM code and the rest remains emulated x86.
I'm no more a fan of Microsoft than the next average HN reader, but I think Microsoft is doing some real leg work here. That being said, I get the impression that it's very much Qualcomm-driven.
Disclaimer: Qualcomm employee but don't have anything to do with any of this
I feel like Microsoft's current stand on this is weird. They're either not as enthusiasting as Qualcomm and just riding along, or their terrible execution makes them move really slow.
There's not much Qualcomm can do about Microsoft lifting its bum to help them, but they can support a kickass Linux experience.
And why intel sits around licking Microsoft's boots. All these decades when they too could have been pushing good Linux support as a hedge against Microsoft doing exactly this I'll never know.
The second m1 dropped, amd and Intel should have been all in on pushing a well-supported Linux desktop to compete with that. They dropped the ball of course and let Microsoft just fumble away with Windows like they have been for the last decade and a half.
Implementing this hedge in these gigantic multi-billion dollar companies would have been what maybe five $10 million a year? Peanuts.
When I tried to skim what was so "special" about the apple M1 secret sauce, older HNs posts said that it was basically just a well-executed design (large RoB, beefy cache, wide pipelines, etc.) using the latest tsmc node. But even in the years since then it doesn't seem like any other companies have managed to design something comparable, and it basically took this team of ex-apple folks to produce something similar.