Anandtech just released their iphone XS review and with it some SPEC2006 testing, the results show just how well the Apple chip performs and that the single-thread performance is up there with the fastest dekstop CPUs.
What is quite astonishing, is just how close Apple’s A11 and A12 are to current desktop CPUs. I haven’t had the opportunity to run things in a more comparable manner, but taking our server editor, Johan De Gelas’ recent figures from earlier this summer, we see that the A12 outperforms a Skylake CPU. Of course there’s compiler considerations and various frequency concerns to take into account, but still we’re now talking about very small margins until Apple’s mobile SoCs outperform the fastest desktop CPUs in terms of ST performance. It will be interesting to get more accurate figures on this topic later on in the coming months.
Apple is making some mighty impressive CPUs and I’m wondering if we’ll see those long rumored ARM macbooks soon.
Apple SoCs reached desktop-class performance 4 years ago or so, but this is the first time it’s reached similar IPCs to current desktops.
I’m sure ARM macbooks are coming eventually, but it’s important to note exactly how Apple reached these performance levels. In short, they have massive L1 caches on their chips. That’s expensive, takes a lot of silicon budget. So the question isn’t whether Apple can make ARM macbooks, but whether that would be price-competitive with Intel, and worth the trouble of transitioning all that 3rd party software.
Well the price does not matter to apple customers, they have proved that amplly now. So there is no concern on that part.
Worth the trouble? That’s the big one. But I believe it would be, they all ready tailor a super custom version of a BSD based OS to their needs with things like Metal APIs and their own specific projects like Logic DAW. So if they can in house enough of it, which they are trucking away on, I would say it is just a matter of time.
IMO it already exists. I’ve been using an iPad Pro for work and from what Geekbench shows this guy can reach haswell CPU levels which is pretty damn fine considering the weight and battery life.
And yeah I work in IT. The fact that everything nowadays is made through IaC software like salt and gitlab made this possible. It is pretty easy to reach out for some remote hosts for the things I cant yet do on iOS.
Apple had Rosetta for their PowerPC to x86 transition that ran nearly flawless for most users. I don’t doubt they could do the same for an x86 to ARM transition.
That’s because intel chips were way faster than the essentially abandoned PowerPC chips. There’s a definite and noticeable performance hit, see all the reviews of Windows 10 on ARM.
I don’t know about this kind of comparison. There are way too many pipeline and instructions differences between an Apple ARM chip and an x86 chip.
Maybe the ARM architecture, being an evolution of RISC, can improve much faster than x86 which uses CISC type of instructions.
If someone can explain me what kind of meaning saying “Performance: Reaching Desktop Levels” has let me know.
I am by no means a benchmarking expert, but the way I understand it the SPEC2006 testing is as apples to apples as it gets for synthetic comparisons between ISAs.
I’ve been reading through their website and what I gathered is that you can have that comparison but an ARM chip gets to the same speed as an x86 in different ways, working on a much less complex instruction set (fixed lenght of instructions) and a less complicated pipeline.
I guess it makes some kind of sense but still I see it like comparing a train and a car: both can do the same mph but they do different things in a different way.
This type of analogy has been made a lot in the x86 vs ARM discussions, but I’m not sure it has much meaning anymore.
Other than compatibility x86 doesn’t seem to have much “advantage” over ARM nowadays.
Either I’m blind, retarded or missing something.
But there’s one mention of Skylake in that article, that article points towards a Thunder X2 test.
Now no test here does any comparison of A12/A11 vs Thunder X2.
According to this a single Vortex A12 core at 2.5Ghz (The A12 has two Vortex cores) is a tonne faster at 403.gcc (44.56) than a Single Xeon thread at 3.8Ghz (31)
Either this is really amazing or i’m being stupid or something is off.