Thoughts on M1 ?
seems like they smoked intel

let's wait until the devices ship and we'll see the actual performance figures

Intel 😂

So says ... Apple?

Seems a little early to count all the chickens and details that come with different processor vs different processor ...

We have a few marketing slides, a graph with no labels, and lots of statements like "nX faster than the bestselling laptop in the same class," with no definition of class or metrics.

What we actually know:

- it has 8 cores, 4 fast, 4 slow
- it supports up to 16gb ram
- it's an soc
- it's called the M1

I saw someone claim the M1 has 500 times the number of transistors of a Ryzen 3600

@12bitfloat
Don't forget that it also does the kessel run 1.2 parsecs faster than the millennium falcon.

@12bitfloat it has 16 billion transistors as per apple

Even if they beat Intel - that's not the real x86 competition anymore. Intel is also beaten within the x86 ecosystem.

Apart from that, Apple has the advantage of being the only PC manufacturer for macOS. Using this vendor lock-in, they can force anything upon their (remaining) customers.

Microsoft can't do that, which is why ARM-Windows is a flop. And that's a good thing given that ARMs are SOCs with binary blob shit, which is why x86 Linux is a lot better than ARM Linux.

I did read that new Mac books with this chip can't have more than one external monitor...

@linuxxx oh that’s bad , but you can always daisy chain, right ?

@Fast-Nop
Ding ding ding. We have the winrar. Profit margins set to maximum.

@hardfault No, as in the cpu's capabilities... But idk, not a hardware person.

@12bitfloat I saw people at work claim Apple Silicon will completely outclass everything on the market. I still don’t know why people actually believe marketing hype. 🙄

But these same people believe absolutely everything the media feeds them, so it’s not like I’m surprised or anything. If CNN told them the grass was black and the wind was pink, they would believe it and defend the claim.

@theKarlisK For our Dutch people: wij van WC Eend....

@Root ipad chips already smokes i5 single core performance, so based on that my hopes are high, usually apple always down play their score in performance and batteries

Sure you can play iPad games on it now but you can't go install Windows 10 (Windows X looks likr shit for any device rn) so that's a shame for gamers.

Apple didn't even mention what the "top tier PC laptop" they compared it to. All their claims are just hot air until they actually get proven.

@melezorus34 yes no windows is a big downer , but I think windows might go on same route with Qualcomm chips in future

Yeah, I think if Apple had the best performing CPU on the market, they wouldn't compare it to "the most sold windows laptop" like they did in their reveal presentation. If they were actually faster than an i9-10900k or 5900x, I think they'd mention

What I know:
- it won't play AAA games
- cheaper to build, but sells with the same price

Apple. Always innovating... At marketing

For everyone sniffing at performance numbers, go look at A14's numbers, which are out. That should give you a good base for comparison, the M1 is only going to be more powerful than that (for one thing it has twice the number of high performance cores and a much larger thermal envelope). Also keep in mind the low TDP and great battery life because of the big.LITTLE design and 5nm process. There's so much more to a laptop CPU than how many SGEMM iterations it can do per second.

Let's see, I'm expecting something pretty good. Intel et. al. might beat them for wide functional unit ops but apart from that I expect it to be quite competitive. Raw performance is meaningless without the full context of what Apple's trying to do here.

@12bitfloat they probably did that because "highest performance" is a fairly meaningless term. Highest performance at what, for what application, under what operating context? Sure a 5900X would beat it at raw perf, but what about energy/op, or energy-delay product, or power scaling? The M1's built in NPU would shred the 5900x in power efficiency for ML applications for example, so for the same number of watts the M1 has "higher performance". Even tests like geekbench and SPEC are not quite there because they're just particular datapoints measured in particular ways. tl;dr it's complicated.

Initial reviews are out seems like they were not joking

@hardfault I heard that compared to desktop chips the M1 may only be half as fast

@12bitfloat yes but comparing a 10W chip with a 45-60W desktops chip is not fair
The performance per watt is unmatched

@hardfault that and the single core performance is competitive

@12bitfloat I heard the same about x86 low power mobile chips vs x86 desktop chips.

Comparing to Intel doesn't make much sense because Intel is technologically way behind in x86, but the M1 stacks up well even compared to AMD's Zen2 mobile stuff.

Not only that, but you have to compare the M1 to AMD ultrabooks that use the 4600U to 4800U with no extra graphics card, i.e. APU graphics only - and these are difficult to actually buy because most laptop manufacturers have trouble understanding the idea of an APU and pair that with a pointless and power hungry extra graphics card.

In order to beat the M1, AMD would need to release the Zen3 Ryzens for mobile, but that won't happen this year.

@Fast-Nop We still don't know how fast the M1 actually is

@12bitfloat We do know e.g. the Geekbench results, that being 1732/7545 (single/multi). That's faster than the 4800U at around 1150/6500 with quite some variance, depending on the laptop manufacturer, used RAM and configured TDP.

@Fast-Nop They're not "offical" offical though plus GeekBench has a bad rep as a benchmark (some publications explicitely don't use it for that reason)

So we'll see once Apple actually ships them

@12bitfloat Geekbench has a bad reputation? I think you're mixing that up with Userbenchmark which manipulates its numbers until Intel comes out first no matter what.

@Fast-Nop Both have a bad rep iirc. GeekBench isn't very accurate and Userbenchmark always conveniently skews towards intel

@12bitfloat Geekbench is pretty accurate actually - that's the reason why its entries vary a lot even for the same CPU. RAM speed, mobo, thermals, everything can influence the result.

Which is good because then you can gauge your system against comparable ones to check whether your build is OK. Plus that it works under Linux.

@12bitfloat @Fast-Nop reviews for the M1 MacBook are out on YouTube, so we do have real world numbers. Look at eg. Dave Lee's numbers. Apple wasn't joking, that thing is hilariously fast.

@RememberMe
I'm going to let other people eat this early adoption. From what I hear the hardware decoder they included skews many video-centric benchmarks, and docker put out a long explanation of "why are we not even close to ready." I'm going to wait to see the tale of the tape on the real cross-platform dev comparability.

@SortOfTested Dave Lee looks at Xcode compile times. That's really hard to skew. Plus Cinebench. Also pretty much every CPU has a hardware decoder anyway. I'm going to check one out soon, a friend bought one, I'll run some tests too. I'd argue docker is a special case because of specialized requirements, that's not really to do with how fast the chip is.

Besides raw performance is only one side of the equation, and one Apple need not necessarily care about all the much. The high performance of the fast cores is also for increasing power efficiency by bringing down the overall energy-delay product of the system, so even if it is specialized it's still a win. That's just something inherent to asymmetric processor design.

Edit: https://anandtech.com/show/16252/...

@RememberMe
I think my point might have gotten lost in a single statement:

My entire world is cross-platform. Raw bench numbers to me don't matter as much as "does it do what I need it to do?" I'm still going to wait and see what happens when the dust settles on java, docker, jetbrains, etc.

I'm happy to see competition, but I'm old, and not willing to endure a third round as an early adopter of apple switching instruction sets. I'll let the young humans deal with that for me. :)

@SortOfTested fair enough point. I think given a year or so it's not going to matter that the ISA is different, given how fast stuff is being ported. Once all the low level stuff gets ported or shimmed I expect it to be really easy to switch. But eh, let's see.

Me, I'm young enough to be fine with suffering through an ISA change :p most of my work is either C++/Python or SML/Coq/custom language stuff, both of which are excellent for cross platform anyway. The heavy stuff runs on servers as it is, so I'm really looking for a light powerful efficient client device which can do *some* work locally, so this pretty much fits right in.

@RememberMe i am looking forward for 16 inches MacBook pro, that’s where they will have the absolute banger ,
Also i see noone tried the virtualisation like virtual box, and I remember apple released its own virtualisation tool, i just wanna see how it handles linux VMs