Hacker News Reader

Comment by coldtea | Hacker News Reader

>I would like the people downvoting you to explain why they think it's truly revolutionary.

For one, it's the first PC chip with these kind of thermals and power combo.

Second, it brings many new design decisions, from unified memory, to the SoC design, various coprocessors, etc, that have not been available in the form of laptop/desktop PCs. The last of this kind in the mass market would be something like the Amiga, which is still a very different design.

Third, it's hella fast and many different things, both given its power draw, but also irrespective of it.

Fourth, it makes ARM for PC computing mainstream suddenly, after decades where it was relegated to the mobile and hobbyist/specialist (rapsberry, etc) space.

If that's not a revolutionary change, I don't know what you expected. A CPU made from flour with alien technology that's 1000 times faster and uses quantum bits?

Yeah, they didn't make that.

View on HN

Like the iPhone and iPod, it's a revolutionary entrant from the consumer point of view in terms of user experience. And like the iPhone and iPod, from a technical point of view, it's merely a very competent iteration down a natural evolutionary pathway, in a preexisting product type.

And like the iPhone and iPod it has very real innovations (in engineering) that people always underplay because "No wireless, less space than a Nomad, lame!", or "we had smartphones before, I had some Ericsson/Nokia/MS gizmo, so what's evolutionary about the iPhone?".

You're right. At the end of the day, they're technically evolutionary, not revolutionary, which was the central conceit of this subthread. And there's nothing wrong with that. Technical evolution can lead to consumer revolutions.

Ted Nelson: "It's possible to insist that every change is merely a small difference in degree if you do not want to see a change in kind"

I disagree. The iPhone was revolutionary in that it made smartphones and phone apps accessible to the mass market, unleashing a wave of change in how people live, work, and play.

M1 Macs will be used basically like the previous Macs, but somewhat faster and somewhat longer without plugging in. So I don't think this qualifies as a revolutionary change.

I think the M1 Macs are a precursor for a truly revolutionary (for consumers) Apple ARM computer. Whether that will be realized with the M1X/M2 next year or further on, it remains to be seen.

Ok. I'm glad we now agree that this is not revolutionary.

What's the revolution you are hoping for with future generations of the processor?

These all strike me as evolutionary changes. People will be doing about the same things as before. They will do them a bit faster, and not have to plug in as often.

I expect revolutionary changes to overturn an existing order. E.g., the Mac was revolutionary. Smartphones were revolutionary. This is a solid and impressive evolutionary change. Kudos to them, of course. But when actual revolutions happen, I would like to have a word handy to acknowledge that.

It's not necessarily revolutionary, but it's not just a small evolutionary step. But either way it's just pedantic.

I'm glad we both agree that revolution is not the right word. I think precision in language is something that is generally valued at HN, especially when the lack of precision is marketing hype.

Smartphones can be just as easily trivialized into being something which lets us "do the same things as before, but faster and with less plugs".

I think it is wrong to say there is a clear difference between these kind of changes. We won't really know how "revolutionary" M1 was until we see the landscape of the market in the future, but I think the simple fact of dropping x86 alone is enough to enable some major new ways of looking at desktop computing.

They cannot in fact be that easily trivialized. You'd welcome to try, though if you think you can demonstrate the point.

If you really believe that non-x86 laptops are what makes for a revolution, then the credit here doesn't belong to Apple. For example: https://www.theverge.com/2020/1/6/21050758/lenovo-yoga-5g-sn...

or: https://www.cnn.com/2020/03/09/cnn-underscored/galaxy-book-s...

But personally, I'd say that those are an interesting step, but not particularly revolutionary.

The credit doesn't really belong to Apple for big screen primarily-touch based communication devices either, except that they popularized them.

I think that's incorrect. As a user of a pre-iPhone smartphone, I don't think Apple should get credit for inventing the smartphone, but I do think the iPhone was revolutionary. They didn't just popularize it; they created a true consumer-grade device via relentless user-focused polish. It's the same deal with the original Mac, which was also revolutionary.

But here, there's nothing radically different about M1 Macs that will open up vast new markets or notably change the daily lives of a purchaser.

Exactly how is it much better? It's not the first chip with a 15-25W TDP. When compared to other chips with such a TDP, it's incrementally faster in single core and it trades blows in multicore. It does this at a process node ahead.

For very light loads, it does have a slightly better power performance than existing low power PC processors, but not by that much. A Ryzen LPP core is somewhere between the big and the small cores of the M1 in power performance in such cases.

It's not the first PC chip to be an SOC, or to have unified memory, or to have many coprocessors. Literally all of those were done simultaneously, in chips of comparable performance, at the same power.

It's also been a decade that we know that instruction sets don't matter. All modern PC processors use a different instruction set and translate from x86. To switch them to ARM would be a matter of adding a few extensions to expose all the functionality, and change the instruction translation stage a bit.

>Exactly how is it much better? It's not the first chip with a 15-25W TDP. When compared to other chips with such a TDP, it's incrementally faster in single core and it trades blows in multicore. It does this at a process node ahead.

Being a process node ahead is already "much better", and IS a feat. Where are the process node ahead chips from the competitors?

On top of that, it's Apple's own SoC design, it has its own GPU design and co-processors, tons of work into making speed optimizations all around (from the buses to memory handling), and came with a new OS release, a port to the new architecture, a bridge layer for iOS apps, and a perfectly capable x86 translation layer.

Rosetta2 alone would be difficult to pull all for most of the industry (Microsoft failed at the same, and their ARM laptops are completely subpar to the M1 machines as well in many levels, despite costing the same or more).

>It's not the first PC chip to be an SOC, or to have unified memory, or to have many coprocessors.

Which mainstream PC chip in actual laptops/desktops people use has unified memory and doesn't use sockets/DIMMS?

>It's also been a decade that we know that instruction sets don't matter. All modern PC processors use a different instruction set and translate from x86.

That's a moot point, since they still carry all the x86 baggage in microcode, since tons of bad decisions leading to things like Spectra. ARM doesn't.

In terms of GPU cores it seems evident that Nvidia is clearly number one—disregarding any thermal limits—and AMD is clearly number two. It seems Apple is a solid contender for third place now. I don't think Intel or the smaller embedded players (Mali, PowerVR etc) have anything quite so good.

Am I missing something?

Dunno, do you want a GTX 3070 for $499 or a AMD 6800 for $580? The benchmarks come out at about the same per $ (AMD wins by 10-15%). Sure you could get the 3080 ... or the 6800XT. Pick your price point and AMD and Nvidia will have something close. Sure AMD has nothing to compete at the $1500 price point, but the reviews I've seen claim that the $1500 nvidia is just a big waste not close to justifying the price.

PS5, Xbox X/S, and Apple have been preferring AMD lately. Even Tesla seems excited by the RNDA2 cards.

Yes, Nvidia and AMD are #1 and #2. I was contemplating who might be #3 right now.

>Being a process node ahead is already "much better", and IS a feat. Where are the process node ahead chips from the competitors?

Not if you still don't beat them cleanly. The end result is what matters. Also, being able to pay more than other people isn't a technical feat.

>On top of that, it's Apple's own SoC design, it has its own GPU design and co-processors, tons of work into making speed optimizations all around (from the buses to memory handling), and came with a new OS release, a port to the new architecture, a bridge layer for iOS apps, and a perfectly capable x86 translation layer.

Half of that is software, which I didn't mention, and the rest is par for the course. The buses and interconnects in a Zen chip are far superior, which you notice as better I/O.

>Which mainstream PC chip in actual laptops/desktops people use has unified memory and doesn't use sockets/DIMMS?

You didn't mention Sockets/DIMMS, but literally any AMD APU has unified memory, and use memory DIMMs. Seems like the best of both worlds to me. Just as good memory performance, and it's upgradeable.

>That's a moot point, since they still carry all the x86 baggage in microcode, since tons of bad decisions leading to things like Spectra. ARM doesn't.

AMD is not affected by Spectre. ARM also uses speculative execution.

> they still carry all the x86 baggage in microcode, since tons of bad decisions leading to things like Spectra. ARM doesn't

> AMD is not affected by Spectre

This is not true. AMD, ARM, Intel, and IBM are all affected by Spectre (Bounds Check Bypass) and that particular variant has not been fixed in hardware, and short of disabling speculative execution or radical changes to ISA or microarchitecture, is likely to never be fixed purely in hardware.

https://arxiv.org/pdf/1902.05178.pdf

AMD's new CPUs, post FX, are not affected by the majority of Spectre vulnerability. The ones that do affect them are fixed in software with essentially no performance penalty. Less than even ARM CPUs.

* Exactly how is it much better? It's not the first chip with a 15-25W TDP. When compared to other chips with such a TDP, it's incrementally faster in single core and it trades blows in multicore. It does this at a process node ahead.*

Ok. Where can I buy a laptop with similar performance, battery life and cooling then?

Same arguments came out with the first iPhone. “It’s just a big screen! Other phones have a big screen!”

Similar performance and cooling? Buy a Ryzen 4800U laptop.

Similar battery life? A good part of it is down to the software, sadly.

For the trade-off of 8 instead of 12 hours of battery life, you'll have a repairable and open computer. Seems competitive. You also get upgradeable RAM, a GPU that is actually useful because it has drivers for a useful API, and much more I/O.

1) M1 is significantly faster than 4800U at single-core.

2) M1 is up to 20 hours of battery life.

3) Not the software since the existing Intel based Macs share the same OS.

M1 is somewhat slower in multicore, and has a fifth of the I/O performance. They are comparable.

Importantly, you're comparing the M1 in a Mac Mini, to the 4800U in a laptop. Comparing the M1 in the Air, for example, the 4800U is quite a bit faster in multicore and the single core performance advantage shrinks.

As for 2 and 3, you're comparing to AMD, not to Intel. AMD CPUs have significantly better power efficiency than Intel.

If you want to compare battery life without taking into account software, compare power consumption. The Zen chip uses slightly more power than the M1 on a laptop.

>As for 2 and 3, you're comparing to AMD, not to Intel. AMD CPUs have significantly better power efficiency than Intel.

Still nowhere near as the same battery life has been seen in any AMD laptop. Not even in ones with less perf.

Seems that you're grasping at straws at this moment :-)

No I'm not. I'm specifically considering two chips.

Those Macs have both much better software for power management, and much more battery. You can't compare chip power consumption by comparing total endurance of the system.

When you do compare the two, they're within 30% of one another.

Read the anandtech article again. 15-25w is the power consumption of the whole Mac mini measured at the wall plug. I.e it includes power supply, ssd, gpu comparable to RX560, external buses etc etc. If you want an x86 chip that beats its performance by a significant margin you have to pay more for just the chip than for the whole Mac mini.

The 4800U also includes a GPU, and 10W is very generous for SSD and RAM.

Since both are SoCs, external buses and so are done in the chip.

Reply on news.ycombinator.com