Geekbench scores for Apple's new iPad Pro have appeared, and they sure are impressive. At least one configuration of the iPad Pro tablet, which you can see on Geekbench here, with its A12X Bionic system on a chip (SoC) achieved a single core score of 5,030 and a multi-core score of 17,995.
Some sites, like 9to5Mac, which first found the scores, point out that the A12X Bionic's performance isn't far behind some powerful Intel chips. However, it's possible that variants in what tests are being run (AArch64 for ARM versus 64- or 32-bit for Intel) could have an effect. When we tested the 13-inch MacBook Pro with an Intel Core i7-8559U, it achieved a multi-core score of 17,348.
The model tested, named iPad8,8, lists 5,650MB of memory, which confirms speculation that at least some iPad models have 6GB of RAM. It has been speculated that 1TB iPad Pros get 6GB of RAM, while those with lesser storage only get 4GB. It's possible that the configurations with 4GB would produce slightly lesser scores.
On the GPU-based compute test, the same iPad model earned a score of 42,038.
It's rumored that iOS 13 could include a big redesign with new features, so it's possible that this power is simply future-proofing the iPad Pro. Either way, Apple's team in Cupertino is proving that it's chip division is nothing to mess with.
The iPad Pro is being tested in the wild now, but for most people it will be available on November 7 starting at $799 for the 11-inch model and $999 for a 12.9-inch model.
Now when Apple can meet Intel or AMD on a per core level it will be impressive. But I doubt they will do it in the small TDP window they are trying to be in since it has to be more energy efficient than performance biased.
I would love to see a comparison using x86-64 optimized bench mark software against the A12X to counteract the blatant bias here.
The ARM configuration is 4 powerful cores and 4 weak cores, for power saving. It's not even clear if a benchmark like Geekbench is using all core, but even if it does, that about equivalent to 4 real cores + 4 hyperthreaded cores. So the comparison is quite apt, in that it's quite possible that IPC is the same for a powerful core in the A12X and an Intel core.
Those 4 weak cores are vastly more powerful than Hyperthreading. HT adds at best 20% performance when it is programmed to take advantage. Those 4 cores are full cores just not as powerfull. The original idea behind the big.LITTLE design was the 4 big cores were the new "top of the line" ARM cores while the old ones were the older ones.
For example the Snapdragon 810 had a 2GHz Cortex A57 and a 1.5GHz Cortex A53. However the newer designs are just lower clocked versions of the same CPU, the Snapdragon 845 for example is 4 Kryo 385s at 2.8GHz and 4 at 1.8GHz.
So no it is not an apt comparison. The i7 is a 4 core CPU with SMT that basically allows multiple threads to be pushed through the same 4 cores. The A12X is a 8 core CPU with 4 high performance and 4 lower performance. And we don't know how Geekbench tests this but I would not be surprised if it does use all 8 cores in testing. Either way I doubt this CPU could replace any full fledged x86-64 CPU in raw performance.
With the elephant-sized exception of recompiling *everything* for a new arch. Sure, Apple can do it but what about all those third party devs?
If you compare this A12X to any chip Intel or AMD could reasonably place in a tablet(heat / power) which is what the A12X is going into, aka the new iPad pro. I do believe Apple are getting higher per core performance than AMD or Intel in that power envelope. Apple is rumored to be working on a desktop class CPU so we will likely have to wait and see to compare them to the high performance chips but nonetheless this is somewhat impressive we are seeing this kind of CPU power in a lower power tablet.
I would compare it to that. Apple isn't though. Apple is comparing it to the i7 U series. Apple is comparing it to the XBox One X. They are putting themselves in this position to be dissected.
I have no doubt that the CPU is the most powerful ARM CPU out right now. However when they state its more powerful than a i7 U series they have to be looked at in every aspect especially when they use a synthetic benchmark for it and are using it on an OS optimized specifically for that chip. Lets take iOS out of the equation and synthetics out. Lets take an OS that's not specifically designed for either and run real world tests on it and see.
However we can't do that as Apple likes to keep everything closed so all we can do is use their setups specifically designed for their hardware. I really don't care if they decide to replace everything again and make people go through the hassle of changing all their software again and leave anything non Apple CPU based in the dust. I don't feel bad for people who decide to buy an Apple product knowing that soon they may not get any new OS updates. Let them waste their money. I just have an issue with Apple making proclamations for a product that has no use outside of their ecosystem.
Stop spreading misinformation. You know the small cores are completely different than the large cores. You know Apple is comparing with the Xbox One S, not X. You know for the same amount of power draw, the A12X equals Intel easily. You're just digging in for no reason.
I literally stated that the big.LITTLE was originally two new cores with two older style cores designed with performance/power in mind.
Still doesn't change the fact that they are full cores and not anything similar to HT.
I misspoke on the Xbox version. Still they are comparing themselves to a dedicated gaming system not me.
For power draw, possibly. For actual real performance that is not based on a synthetic benchmark with too many variables that cannot be equalized? No. Until you can give me just the CPU in an OS that's not catered to that CPU or the Intel/AMD, that doesn't use a synthetic benchmark showing real world performance matching then no they only match it when its on their OS using their design for their ecosystem. It has not relevance in the laptop or desktop market.
Right now they are comparing a mobile OS to a desktop OS. The overhead difference alone can easily account for the performance.
Tell you what. Get with Apple and have them allow someone to develop a Linux OS for both x86-64 and the A12X and then have them compile some real world tests and lets see if it really matches that CPU or if it changes once their control over it is no longer there. I bet it wouldn't look nearly as good as it does right now.