Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Apple's follow-up to M1 chip goes into mass production for Mac

asia.nikkei.com

Post reply on HN
Print view View original

471–480 of 661 posts

Re: Apple's follow-up to M1 chip goes into mass production for Mac

#471

Post by olliej »

Someone wrote: post #337

FTA: "the latest semiconductor production technology, known as 5-nanometer plus, or N5P. Producing such advanced chipsets takes at least three months” I know as good as nothing of this process, but I can’t imagine the latency from silicon wafer to finished product is 3 months. I also can’t imagine some inherent start-up delay for producing the first chip (but as I said: I know as good as nothing of this process), so…

Latency for a chip is 4 weeks at the low end of complexity, and 12 weeks (3 months) at the high end of complexity.

My mind was blown when I first found that out

Re: Apple's follow-up to M1 chip goes into mass production for Mac

#472

Post by paulpan »

mortenjorck wrote: post #39

This would be quite an accelerated timeline if Apple ships its second-generation M-series chip only eight months after the first. Typically, they’ve followed a sort of six-month tick-tock pattern for the A-series, launching a new major revision in the fall with the new iPhone, and launching an “X” revision in the spring with new iPads. I think most observers have been expecting an “M1X” for Apple's first pro-oriented…

I think getting out the M2 before the fabled "M1X" actually makes sense. This could explain the decision to put the M1 into the new iPad Pros, to re-use the M1 chips elsewhere once the new M2 becomes available.

Main reason being the M1 was a more proof of concept and rushed out (despite being as good as it turned out to be). The M2 will be a more refined M1 but with notable improvements such as LPDDR5 support - akin to AMD's Zen1 and Zen+ releases.

On the other hand, there could be a M1X being readied for release in the upcoming June WWDC. It may be architecturally older than the M2 but still superior performance on a big cores differential, e.g. M1 only has 4 big cores and 4 small cores, the M1X just needs more big cores to be notably more performant.

All highly speculative of course, will have to find out in about a month.

Re: Apple's follow-up to M1 chip goes into mass production for Mac

#473

Post by rubatuga »

nonameiguess wrote: post #320
forgetfulness wrote: post #175

Earlier quoted context omitted.

Maybe the commenter has a very interesting use for it, but why would you buy a 256 GB RAM machine (which isn't a lot of memory for this arena either) to develop the model, instead of using something smaller to work on it and leasing a big cluster for the minutes to hours of the day that you'll need to train it on the actual dataset?

I don't do this kind of thing any more, but back when I did, the one thing that consistently bit me was exploratory analysis requiring one-hot encoding of categorical data where you might have thousands upon thousands of categories. Take something like the Walmart shopper segmentation challenge on Kaggle that a hobbyist might want to take a shot at. That's just exploratory analysis, not model training. Having to do t…

I think there are studies showing that one hot encodings are not as efficient as an embedding, so maybe you would want to reduce the dimensions before attempting the analysis.

Re: Apple's follow-up to M1 chip goes into mass production for Mac

#474

Post by Elora »

nexuist wrote: post #383
amelius wrote: post #377

Earlier quoted context omitted.

> Just imagine Apple releasing the Mx chips for server infrastructure. Intel ought to be sweating bullets now. As a developer, I'm sweating bullets. Imagine a future where Apple is the only serious platform left on mobile, server and desktop, then developers can forget about starting that next disruptive startup and instead they will become just second-class Apple "employees" with contracts similar to Uber drivers.

Apple isn't harnessing magic here. The opportunity for competition exists, but everyone else is just sitting on their asses while Apple's chip team puts in the work. If you want to beat Mx, you have to think outside the box, just like how Apple thought outside of x86. That being said, I do enjoy the shift to ARM as the dominant arch. x86 has far too many decades of cruft behind it. I don't think Apple will take over…

> The opportunity for competition exists

This is so deep and powerful if you really think about it. This is what we were promised. While not perfect, I applaud Apple for at least trying.

Re: Apple's follow-up to M1 chip goes into mass production for Mac

#475

Post by astrange »

r00fus wrote: post #402
google234123 wrote: post #388

Earlier quoted context omitted.

ARM also has some decades of cruft attached to it.

ARMv8, ie, AArch64 is a redesign/new, and entirely 64 bit with legacy (32bit) support mode tacked on. The design was done to be able to remove legacy without another redesign. To my understanding, this is quite unlike AMD64 (ie, Intel's current 64 bit ISA licensed from AMD) - which extended x86.

The 32-bit mode is optional; M1 doesn't even support it.

Re: Apple's follow-up to M1 chip goes into mass production for Mac

#476

Post by edrxty »

acegopher wrote: post #405
edrxty wrote: post #397

Earlier quoted context omitted.

I'm really hoping one of the other chipmakers jumps on the RISC-V bandwagon. There's substantial potential to do to ARM what ARM did to x86 here. If Intel/AMD/Qualcomm/Broadcom/whoever started talking about a significant RISC-V offering I'd be buying as much of their stock as I could.

RISC-V doesn't offer the same advantage over ARM that ARM has over x86, so that's unlikely to happen.

I work with RISC-V, it has a lot of features that are not yet well explored. In particular the instruction set extension mechanism is extremely robust and capable, allowing much more FPGA fabric integration that you currently see on ARM devices. As we move towards more programmable logic it'll be a massive advantage going forward.

Re: Apple's follow-up to M1 chip goes into mass production for Mac

#477

Post by crazygringo »

nexuist wrote: post #385
crazygringo wrote: post #374

Earlier quoted context omitted.

But since so much software these days is cross-platform, apps/sites will still have to work performantly on Intel chips. E.g. Google can't slow down Chrome to take advantage of Mac speed increases, because it would become unusable on Windows. So I actually think that Mac software will hold the performance edge for a long, long time.

It logically makes sense for Windows to move to ARM as well at this point. They already ported it with Windows RT, and now all they have to do is provide a translation layer like Rosetta and release the full Windows experience (instead of the locked down version RT was).

But there's no M1 chip equivalent for PC's.

It's fine to have an ARM version of Windows but there's no equivalently high-performance ARM chip to run it on. Unless you're talking about Bootcamp on Macs.

Re: Apple's follow-up to M1 chip goes into mass production for Mac

#478

Post by taysondev »

Klonoar wrote: post #275
flohofwoe wrote: post #216

Earlier quoted context omitted.

Give Mac software a few years to "catch up" and everything will feel just as slow as usual (and I'm not even joking, so far, software was always excellent at undoing any advancements in hardware very quickly).

If you don’t run Electron apps, even an older Mac will fly. My 2015 is still astoundingly fast, it just doesn’t have Chrome bogging it down.

I actually used a MacBook Air 2011 up until summer 2020 as my main home computer. The only reason I even bought a new MacBook was because I couldn't get the newest version of macOS which I needed for the latest version of Xcode.

Finally sold the 2011 MacBook Air a few months ago for $300. I got so much value out of that laptop.

Re: Apple's follow-up to M1 chip goes into mass production for Mac

#479

Post by varispeed »

jjoonathan wrote: post #127
varispeed wrote: post #63

Earlier quoted context omitted.

Exactly, there are bodies that are supposed to protect consumers from that behaviour, unfortunately they failed everyone massively. That in itself begs for an inquiry how those institutions actually work and whether it is worth spending tax payer money on them if they consistently fail to deliver.

...and if that exact kind of navel-gazing is why they don't work? What then?

Charge people for failing to deliver, dispense huge fines and jail time. Then rebuild it in a way to avoid mistakes why the previous solution didn't work.

Re: Apple's follow-up to M1 chip goes into mass production for Mac

#480

Post by edrxty »

foobiekr wrote: post #441
edrxty wrote: post #397

Earlier quoted context omitted.

I'm really hoping one of the other chipmakers jumps on the RISC-V bandwagon. There's substantial potential to do to ARM what ARM did to x86 here. If Intel/AMD/Qualcomm/Broadcom/whoever started talking about a significant RISC-V offering I'd be buying as much of their stock as I could.

ARM didn’t do anything to x86. One specific team at one company did something. Maybe two if you’re super generous and include graviton.

ARM is destroying x86 across the board, not just at Apple. They've long won the mobile space and are now making significant inroads into the server market, long thought to be the last bastion of Intel/x86.
Post reply on HN