Read more.Quote:
A12X Geekbench scores are soundly beaten only by the MacBook Pro Core-i9 version.
Printable View
Read more.Quote:
A12X Geekbench scores are soundly beaten only by the MacBook Pro Core-i9 version.
Geekbench like many of these benchmarks means nothing in the realworld and it's very easy to make sure you can run optimisations to make scores look higher.
Plus TBH for most basic stuff even an SB Core i3 would be fine. I wish instead of optimising for benchmarks these mobile devices worked on dropping power instead.
No wonder Apple is considering moving away from Intel's chips.
Yeah, it's Geekbench.
Still, the trend is clear. And soon we'll be able to compare Photoshop directly, and that will be a real world comparison.
Also I think we'd all agree that the GPU on this will smash any Intel iGPU into a thin watery paste.
Again will be an equivalent version with full functionality?
Will the performance be sustained over a hour of processing RAWs? Or just a best case scenario for the tablet? For instance some of the stuff I do involves processing over a 100 RAWs in a batch with noise reduction.
All these headline figures are probably short term benchmarks and with Apple and Samsung charging more and more they need to find a way to sell stuff.
Geekbench... yeah I never trust that one as it seems to be a little too heavily 'optimised' for arm cpu's and iOS versus other OS's and/or x86/x64.
It's also funny how they've benchmarked the 1TB model (it's the only one with 6GB ram supposedly)...
Basically I'll believe it when I see it when an iPad Pro can keep up with 'professional' requirements in graphic design.
But can it run crysis?
More expensive than 92% of portable PCs too.
Remember when Apple claimed the PowerPC macs were 6 times faster than a PC ... then switched to x86 and suddenly claimed the new ones were 6 times faster than the old?
Just sayin...
Ugh, Apple loves to trot this out every now and then to say how great they are.
Why don't they use these chips in their laptops/iMacs then?
Chips in phones/tablets will always be thermally limited compared to Laptops. Just like Laptops are thermally limited compared to desktops. The amount of performance that is coming out of battery powered devices in such a restrictive form-factor is impressive, don't get me wrong. But how about showing a bit more of a range of benches?
Oh, that doesn't fit your narrative?
Carry on then :P
Some clarification about what they mean by "92% of all portable pcs" would be nice, too. Are we talking about all laptops that are in use at the moment? Because I'm not surprised that a $1000 just-released tablet is faster than grandmas $120 laptop from 3 years ago in 1 particular benchmark.
I'm generally an Apple hater, but that doesn't blind me to how impressive these CPUs are. Slap a big heatsink on one of these, and yes I'm sure it could do all that. When you consider that Intel are trying to sell their chips in badly thermally constrained super thin laptops (something we don't quite see the bias for in this country, but they are) I can see where Apple are coming from here.
Really, all it needs is a big heatsink and this could make an interesting SFF machine. From Anandtech talking about SPEC results for the A12 (not this bigger A12X):
https://www.anandtech.com/show/13392...icon-secrets/4Quote:
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.
I still can't mentally square up that this is supposed to be a mobile chip. 7 instruction issue into 13 ports, for anyone else that would be a server part with threading rather than a phone.
My Ryzen 5 2600 just to do 50 RAWs with NR,can take 20~25 minutes alone,so all these short tests of mobile Apple/AMD/Intel ULV CPUs are great for more basic usage,since the benchmarks are short. They make mobile devices look better than they are.
They are also wide cores clocked low - plenty of designs existed in the past which did. In Russia the Elbrus designs were wide and lowly clocked and apparently had high IPC.
Its a single core short-term benchmark on different OSes,and a low clockspeed server CPU(weird choice) not one of the Intel ULV chips. The test is also under iOS versus another OS too,so you cannot fairly compare stuff. They need to be both under the same OS.
Then as usual a website takes one benchmark and runs with it feeding into the Apple/Samsung hype. Its the best case scenario for Apple and worst case scenario for everything else. Then when you consider Apple is using a cutting edge process node,and only has a design which seems to only scale to a few cores.
The Intel and AMD cores have to scale from low TDPs to high TDPs,low clockspeeds to high clockspeeds and are not running on the latest nodes,have loads of PCI-E lanes,etc. Things like IF draw a massive amount of power too. All these are design considerations Apple does not need to worry about - they are selling CPUs to just cover a phone and maybe a laptop. Intel and AMD need their uarchs to cover more than that and that means they need to make compromises.
On top of that they are commodity cores - they need to be cheap enough to make viable £30 desktop and laptop CPUs,and potentially work on older nodes. The AMD and Intel cores are far more general purpose in the markets they target.
A lot of these mobile SOCs look great until you look at the ton of transistors they use,and the need for cutting edge nodes. We all know nodes are getting harder and harder. The Apple CPUs are massive in terms of transistors too,and there is no guarantee they can scale well with more cores and higher clockspeeds. The A12X is already around 120MM2 on 7NM for a quad core(the small cores are more for low load power) and has 10 billion transistors. The A12 is 83MM2. The A12X has double the transistors of an AMD APU,more than double that of a Ryzen 8 core CPU,which are both SOCs,and apparently more than double the transistors of the Core i9 9900K.
What happens when Apple,Samsung,etc cannot jump to new nodes as quickly??
So when they can be stuck on less than optimal process nodes,can have a design that can scale well to desktop level and replace something like Ryzen 2700X,then I might be interested.
All I see is a hyper specialised niche core design,using the best node possible,under an optimised OS. The latter alone can make a big difference.
Remember how Intel was nicknamed "Inteltech",now since Anand went to work for Apple,they keep hyping up Apple parts,so now are more like "Appletech". Its all about page clicks now.
Apple is having declining sales:
https://www.anandtech.com/show/13540...e-up-sales-not
So they need to hype the "performance" but its not working in actual sales,so they need to up the price. But when they up the price,people keep the products longer and so on. Its a vicious circle.
They are not alone.