Apple A8x

[mavere]

Anything else to run?

Basemark OS II also includes other, non-gpu tests.

Speedometer measures browser's responsiveness/latency.

Oh and can you confirm the oddly high storage results?

 

[III-V]

Oh and can you confirm the oddly high storage results?

The Air 2 would be the iPad 5,4, and 5,3, right?

eMMC 5.0 would easily explain the boost, if it were implemented in the new tablets.

 

[Steelbom]

I got my 64 GB Gold WiFi, iPad5,3
Geekbench says 1.51 GHz, 1.94 GB RAM.

Ran all these benchmarks twice

Kraken 1.1
4423.2 ms
4538.6 ms

Sunspider 1.0.2
329.2 ms
312.3 ms

Octane 2.0
6672 (did not restart browser as recommended)
8349 (restarted browser, as recommended)

Geekbench 3.2.2
1797/4496
1811/4482

Basemark X 1.1.1 Medium (default)
41603
41454

GFXBench 3.0.20
Didn't seem to work.
Went through all the tests and the tests seem to run OK, but then the test results were empty.

Anything else to run?

Awesome! Thanks for the results.

 

[Eug]

The Photobucket app I downloaded on the iPad is absolutely useless. In the meantime, I'm using my ancient mid-2009 MBP to upload these pics to Photobucket. However, it turns out they have two apps. One is from 2011. Doh! The proper one requires iOS 7 or later. The one I downloaded is for iOS 4 and later. I guess they're keeping the iOS 4 version around for legacy support, but it's the first one that showed up for this iPad Air 2.

Back to the benchmarks, etc.:

3DMark 1.3.1 - Ice Storm Unlimited:

Basemark OS II also includes other, non-gpu tests.

Basemark OS II 1.0.11

Speedometer measures browser's responsiveness/latency.

Speedometer 1.0

Oh and can you confirm the oddly high storage results?

The Air 2 would be the iPad 5,4, and 5,3, right?

eMMC 5.0 would easily explain the boost, if it were implemented in the new tablets.

Confirmed. The storage performance is through the roof. Assume these results are truly correct, this is awesome. Totally unexpected for me. I wasn't aware of eMMC 5.0. However, that 947 MB/s read speed seems too good to be true. Is this correct?

Yes, iPad5,4 and iPad5,3 are the iPad Air 2. They correspond to the cellular+WiFi and WiFi-only versions respectively.

 

[Enigmoid]

Maybe its caching to RAM?

 

[mavere]

Comparing Eug's Basemark OS II scores against Anandtech's numbers:

  6+    Air2    Scaling

1474    1851    1.26x    Overall
2082    3020    1.45x    System
 847     667    0.79x    Memory
2435    4608    1.89x    Graphics
1098    1265    1.15x    Web

Interesting that the memory score regressed. I wonder what that portion actually tests. (Edit: Rightware's official results for iPhone 6 indicate a large gap between Average and Max memory scores, so maybe the A8 in general is finicky here)

 

[III-V]

Maybe its caching to RAM?

I think so. The part used is actually the same as was used in the iPhone 5s, it seems. So, not eMMC 5.0.

 

[Eug]

Yeah, that read speed is far too close to the memory bandwidth.

However I'm quite happy with the write speed. ~160 MB/s is nothing to sneeze at.

 

[Eug]

One "feature" that I haven't seen documented anywhere is the MINIMUM brightness.

iPad 2 (iOS 7.1.2) on the left and iPad Air 2 (iOS 8.1) on the right. I checked in the settings and both iPads are at the lowest setting.

I much prefer the lower minimum brightness on the iPad Air 2 (right), for surfing and viewing videos in a completely dark room.

 

[III-V]

Neat. I definitely enjoy devices that can dial the brightness way down.

 

[asendra]

I got my 64 GB Gold WiFi, iPad5,3
Geekbench says 1.51 GHz, 1.94 GB RAM.

Ran all these benchmarks twice

Kraken 1.1
4423.2 ms
4538.6 ms

Sunspider 1.0.2
329.2 ms
312.3 ms

Octane 2.0
6672 (did not restart browser as recommended)
8349 (restarted browser, as recommended)

Geekbench 3.2.2
1797/4496
1811/4482

Basemark X 1.1.1 Medium (default)
41603
41454

GFXBench 3.0.20
Didn't seem to work.
Went through all the tests and the tests seem to run OK, but then the test results were empty.

Anything else to run?

I haven't run much tests, but in my Air 2 I'm getting ~4030ms in Kraken and in Sunspider ~305-310 consistenly.

 

[Eug]

Comparing Eug's Basemark OS II scores against Anandtech's numbers:

  6+    Air2    Scaling

1474    1851    1.26x    Overall
2082    3020    1.45x    System
 847     667    0.79x    Memory
2435    4608    1.89x    Graphics
1098    1265    1.15x    Web

Interesting that the memory score regressed. I wonder what that portion actually tests. (Edit: Rightware's official results for iPhone 6 indicate a large gap between Average and Max memory scores, so maybe the A8 in general is finicky here)

In Memory Mark, the speed is increased with the iPad Air 2.

I get 6240, which puts it at the top of the iOS charts.

 

[Eug]

These aren't my benchmarks because I could not get GFXBench to save the results for some reason. But here ya go:

Offscreen, A8X beats K1.

http://wccftech.com/initial-apple-a8x-gpu-benchmarks-beats-tegra-k1/

 

[sontin]

And it's much slower in the 3DMark graphics test.

A8X archives 31k points, nVidia shield gets more than 36k.

 

[Eug]

Antutu 5.0.1

Linpack 1.4

 

[Eug]

I finally got GfXBench to work. It turns out it had to be updated, and that update came Friday, as 3.0.23.

BTW, with regards to memory, I tried loading up about 8-9 tabs in Safari, and that consumed over 650 MB RAM. Of that, the tab data not including the app itself, was close to half a GB.

2 GB RAM is a nice amount for this tablet, but hell, my next iPad will have to have 4 GB. 2017?

 

[Hans de Vries]

Confirmed. The storage performance is through the roof. Assume these results are truly correct, this is awesome. Totally unexpected for me. I wasn't aware of eMMC 5.0. However, that 947 MB/s read speed seems too good to be true. Is this correct?

I was hoping for that 1 GB/s NAND performance...

Do you remember the leaked schematic from July which says 1GB-DDR
on a NAND interface:

 

[2is]

I wish you guys would stop talking about this before I end up spending $750 on something I don't really need.

 

[Hans de Vries]

I was hoping for that 1 GB/s NAND performance...

Do you remember the leaked schematic from July which says 1GB-DDR
on a NAND interface:

I would like to see some more conformation but this would be a nice development.

The high speed NAND toggle mode standard has 400Mb/s signaling. For 16
data lines that's 800MB/s. For 1GB/s you would need a 500MB/s part.

For those who want to delve into this: All the schematic drawings of
the previous iPad Air. Sheet 15 has the Toshiba NAND Flash part
(THGBX2T0BBJLA03) , but with the high speed control pins unconnected.
Nice picture here:

http://www.techinsights.com/apple-ipad-air/

The high speed control signals (for 400Mb/s/pin) are given here, on page 9:
http://www.jedec.org/sites/default/files/doug_wong.pdf

It will get better (faster) still with the new UFS V2.0 standard:

http://www.toshiba.com/taec/Catalog/Line.do?familyid=7&subfamilyid=900116&lineid=2095185

 

[Eug]

I still can't get over the fact they're still selling the original iPad mini. At least they had the good sense to retire the iPad 2 months ago.

http://www.macworld.co.uk/review/ipad/ipad-air-2-review-3544035/

Geekbench:

iPad mini 1: 260 (single), 494 (multi)
iPad mini 2: 1374 (single core), 2484 (multi)
iPad mini 3: 1376 (single), 2483 (multi)
iPad Air 1: 1468 (single), 2658 (multi)
iPad Air 2: 1818 (single), 4520 (multi)

SunSpider:

iPad mini 1: 1,503ms
iPad mini 2: 442ms
iPad mini 3: 449ms
iPad Air 1: 439ms
iPad Air 2: 287ms

This is the first time I've seen an iDevice (including the iPad Air 2) score well below 300.

 

[Hans de Vries]

I still can't get over the fact they're still selling the original iPad mini. At least they had the good sense to retire the iPad 2 months ago.

http://www.macworld.co.uk/review/ipad/ipad-air-2-review-3544035/

Geekbench:

iPad mini 1: 260 (single), 494 (multi)
iPad mini 2: 1374 (single core), 2484 (multi)
iPad mini 3: 1376 (single), 2483 (multi)
iPad Air 1: 1468 (single), 2658 (multi)
iPad Air 2: 1818 (single), 4520 (multi)

SunSpider:

iPad mini 1: 1,503ms
iPad mini 2: 442ms
iPad mini 3: 449ms
iPad Air 1: 439ms
iPad Air 2: 287ms

This is the first time I've seen an iDevice (including the iPad Air 2) score well below 300.

That's SunSpider on Safari using iOS. I was wondering what that would do
on my notebook PC with the latest and greatest Mac OS X Yosemite 10.10 update 1:

Result: 176ms on a 3.2GHz Ivy Bridge. This is with VMware. I would like to
see the result of a real Mac.

 

[Eug]

Last year's mid range 13" Retina MacBook Pro is about twice as fast as the iPad Air 2 at ~142 ms, in SunSpider 1.0.2. I note you're running version 0.9.1. He was running Safari 7.0.

http://www.webkit.org/perf/sunspide...string-validate-input":[5,5,5,5,5,5,5,5,6,5]}

That is a dual-core Core i5, which should be running at 2.6 GHz.

http://forums.anandtech.com/showthread.php?t=2349482

He also has a Peacekeeper benchmark, at 4262, also in Safari 7.0.

My iPad Air 2 gets 2771 in Safari 8.0.

http://peacekeeper.futuremark.com/results?key=9mm4&resultId=5247312

The iPad 4 gets 951, but that was Safari 6.

Anyone have an iPad 4 or iPad Air result with Peacekeeper in iOS 8.1?

Last year the iPad Air got 1855.

 

[Enigmoid]

Interesting.

http://www.notebookcheck.net/Apple-iPad-Air-2-A1567-128-GB-LTE-Tablet-Review.129396.0.html


Throttling under load. Worse screen. No improvements in battery life.

 

[Arachnotronic]

Interesting.

http://www.notebookcheck.net/Apple-iPad-Air-2-A1567-128-GB-LTE-Tablet-Review.129396.0.html


Throttling under load. Worse screen. No improvements in battery life.

Of course it throttles under load...it's a passively cooled mobile chip.

 

[Nothingness]

Of course it throttles under load...it's a passively cooled mobile chip.

Indeed and even after throttling its Geekbench score is still well ahead of the top of the line non-throttling Intel Moorefield (and BTW has it been proven theses chips don't throttle too?)