Apple's A11 Bionic matches single-core 13" MacBook Pro performance in alleged benchmark
Reported benchmarks of Apple's latest A11 Bionic system-on-chip design, which will see release in the iPhone X and iPhone 8 series, show it outpacing not only current generation A10 and A10X chips, but matching the performance of some 13-inch MacBook Pro configurations, at least in terms of single-core processes.
Officially announced to the public on Tuesday, iPhone X, iPhone 8 and iPhone 8 Plus units are apparently being tested in the wild, with users pinging Geekbench servers to evaluate the systems' performance metrics.
Identified as "10,2," "10,3" and "10,5," the iPhone models seen in Geekbench scores on Wednesday appear to run the same A11 Bionic SoC unveiled onstage at Apple's special event at Apple Park. The chip sports two Apple-designed high-performance cores, four high-efficiency cores and, for the first time, an Apple-designed GPU.
In a series of 12 Geekbench 4 tests, minus what appears to be an anomalous reading, the A11 achieved an average single-core score of 4169 and an average multi-core score of 9837.
By comparison, the A10 Fusion power the iPhone 7 Plus scores roughly 3514 in single-core mode and 5970 in multi-core processes. The latest 12.9-inch iPad with A10X Fusion processor, Apple's fastest currently available mobile chip, scores 3924 in single-core and 9371 in multi-core.
With a single-core high water mark of 4274, the A11 Bionic rates just above a mid-2017 13-inch MacBook Pro running a dual-core 3.1GHz Intel Core i5, and a shade below a late-2016 15-inch MacBook Pro powered by a quad-core 2.9GHz Intel Core i7 as noted in Geekbench's Mac Benchmarks. As expected, Intel's laptop-class silicon blows away A11 Bionic in multi-core performance, with Apple's mobile chip reaching a high score of 10438, just above a mid-2012 15-inch MacBook Pro.
In a statement to MacRumors, Geekbench's John Poole said he believes the A11 Bionic benchmarks are legitimate, saying the two high-performance cores are likely operating at 2.5GHz. The "25MHz" seen in Geekbench metrics related to A11 Bionic chips appears to be an anomaly, Poole said.
The latest Geekbench figures follow an early benchmark that popped up on the service yesterday. At the time, the Geekbench 4 database showed a single entry for what now appears to be an iPhone X.
Apple's latest A-series silicon is due to arrive in the hands of consumers next week when the first iPhone 8 series preorders ship out to customers on Sept. 22.
Officially announced to the public on Tuesday, iPhone X, iPhone 8 and iPhone 8 Plus units are apparently being tested in the wild, with users pinging Geekbench servers to evaluate the systems' performance metrics.
Identified as "10,2," "10,3" and "10,5," the iPhone models seen in Geekbench scores on Wednesday appear to run the same A11 Bionic SoC unveiled onstage at Apple's special event at Apple Park. The chip sports two Apple-designed high-performance cores, four high-efficiency cores and, for the first time, an Apple-designed GPU.
In a series of 12 Geekbench 4 tests, minus what appears to be an anomalous reading, the A11 achieved an average single-core score of 4169 and an average multi-core score of 9837.
By comparison, the A10 Fusion power the iPhone 7 Plus scores roughly 3514 in single-core mode and 5970 in multi-core processes. The latest 12.9-inch iPad with A10X Fusion processor, Apple's fastest currently available mobile chip, scores 3924 in single-core and 9371 in multi-core.
With a single-core high water mark of 4274, the A11 Bionic rates just above a mid-2017 13-inch MacBook Pro running a dual-core 3.1GHz Intel Core i5, and a shade below a late-2016 15-inch MacBook Pro powered by a quad-core 2.9GHz Intel Core i7 as noted in Geekbench's Mac Benchmarks. As expected, Intel's laptop-class silicon blows away A11 Bionic in multi-core performance, with Apple's mobile chip reaching a high score of 10438, just above a mid-2012 15-inch MacBook Pro.
In a statement to MacRumors, Geekbench's John Poole said he believes the A11 Bionic benchmarks are legitimate, saying the two high-performance cores are likely operating at 2.5GHz. The "25MHz" seen in Geekbench metrics related to A11 Bionic chips appears to be an anomaly, Poole said.
The latest Geekbench figures follow an early benchmark that popped up on the service yesterday. At the time, the Geekbench 4 database showed a single entry for what now appears to be an iPhone X.
Apple's latest A-series silicon is due to arrive in the hands of consumers next week when the first iPhone 8 series preorders ship out to customers on Sept. 22.
Comments
iPad Pro 12.9 (A9X) - Source
MacBook (Core m3-6Y30) - Source
- Single-Core: 2768
- Multi-Core: 5343
----------------------------------------------Anandtech iPad Pro review using Spec2006 - Source
So despite having a higher Geekbench score, the 6Y30 was determined to have greater performance. So I'm hesitant on Geekbench claims, but it still shows that Apple has been making considerable amounts of progress in trying to catch up to Intel's Y/U processors. Curious to see what the A11X looks like, mind you, Cannon Lake (10 nm) will be out by then, so we'll have to see if Intel can keep pushing forward.
I am wondering how is it possible that benchmarks are already there, considering the phone is officially not out in the open yet.
Who is posting these benchmarks, Apple employees?
It really doesn't matter how anybody spins this. It is remarkable and it is a game changer. The wall between laptops and mobile devices continues to crumble.
By adding a keyboard, pointer and file system to the iPad Pro Apple has already challenged low and medium powered laptops. All they really need now is:
1) A mouse or touchpad replacing the pencil/pointer and let the pencil serve for drawing and markup.
2) Replace the lightening connector with Thunderbolt/USB-C.
At that point, the MacBook and MacBook Airs would be obsolete and irrelevant.
Android: But teh multicore
Jon Srouji: lol
Android: ...Benchmarks don't matter now.
Granted for the laptop comparisons, form factor and software is still a huge driver. If I could dock my iPhone to an inexpensive clamshell dock, like the Atrix did way back, that would be something. My iPhone is already several times more powerful than my travel Acer CB14, I'd happily get rid of that and just take a laptop and cheap clamshell dock.
And god, I want to see A11X with more cores, higher wattage, and active cooling in a clamshell. The iPad Pro with A10X is very powerful, but the tablet form factor is still not as great for productivity as a clamshell.
Regarding the pointer, all they need is a touchpad that mimics the shape of the screen, then wherever you place a finger, it appears as a "touch" by blurring a finger sized shape on the screen in the same location. No arrow or other pointer needed, the OS is already designed around a "touch" which can be mimicked with a touchpad. The pointer doesn't float, it only appears when you touch the touchpad, just like iOS and the way it receives input. No need to train people on the new OS, it's not new, it's simply the same with additional pointing help - we're already some way there with tvOS and it's input device and paradigms. Imagine a laptop like this, it'd sell way more than laptops with macOS on them.
so the question is what Apple would do here. Could they convince small developers to go with it? Probably. But large developers would really have to,have a very good reason to go through this again. And as it’s nkt likely that Apple could use this for anything other than a Macbook, AMD just maybe, but not likely, a Mini, there might not be enough a market for all of this software.
so we would have to go back to the premise I first brought up a few years ago. That is that experts have found that about 80% of the slowdown when running emulation is due to a small number of chip instructions, somewhere around a dozen, or a bit more, depending on which chip families are being emulated by which other families. As Intel doesn’t have patents on individual instructions, Apple could take those few and incorporate them into their ARM chip. When xu6 software requires them, the chip would switch over to them for these calculations.
if Apple could get an A12X perhaps 40% faster than this A11, then with those extra instructions, that could work out about as well as a mid range i5 low power mobile chip for much less money that Apple pays Intel.
i could see that working.
Since Thunderbolt is part of the PCI bus spec, it’s not that easy to put it into an ARM based device. An iPad would have to be completely redesigned for that. Then Apple would need to buy the controller from Intel, which might not be a problem, but next year, Intel will be building the port into their chipsets, and once that happens, how much longer will the individual chips be available? In addition, Thunderbolt controllers use a lot of power, too much for a tablet that has a 4-5 watt SoC.
But, the iPad Pro is creeping up on that clamshell thingee step by step... First the keyboard, then the file system, then the more powerful processor & GPU, now the mouse.