Apple A8 Benchmarked

[Eug]

^^^ Re: 6 plus gfx benchmark discussion about on-screen vs. off-screen:

And a bit of browser stuff:

The iPhone 5s definitely holds its own overall.

 

[liahos1]

Some benchmarks courtesy of AnandTech.

The battery life looks good.

http://www.phonearena.com/reviews/Apple-iPhone-6-Review_id3796/page/4

its all over the place depending on the test methods

 

[Nothingness]

http://www.phonearena.com/reviews/Apple-iPhone-6-Review_id3796/page/4

its all over the place depending on the test methods

Add iPhone 5 and 5s in the phonearena test and you'll see even then their methodology shows that battery life has improved

 

[liahos1]

Add iPhone 5 and 5s in the phonearena test and you'll see even then their methodology shows that battery life has improved

its like 20 minutes - 6.6% improvement - more on a 16% larger capacity battery no? doesn't seem groundbreaking for the 6 despite the move to 20nm. not that i'm complaining though...seems like a fine product with my use on the 6 so far.

 

[witeken]

The battery life of the Plus is pretty disappointing.

 

[Rakehellion]

The battery life of the Plus is pretty disappointing.

Why?

 

[jdubs03]

Anyone notice how the Shield Tablet is like 6500 points lower than it should be?

Obviously we can see what happened.

 

[witeken]

Why?

Because it has a 3000mAh battery, 50% more or so than iPhone 6.

 

[Nothingness]

Because it has a 3000mAh battery, 50% more or so than iPhone 6.

You posted a slide from Intel showing how much the screen consumes, I guess we are seeing this here with a larger and higher resolution screen.

 

[Rakehellion]

Because it has a 3000mAh battery, 50% more or so than iPhone 6.

That's battery capacity, not battery life.

 

[Rakehellion]

Anyone notice how the Shield Tablet is like 6500 points lower than it should be?

Obviously we can see what happened.

Who says it should be higher?

 

[ams23]

Who says it should be higher?

Clearly Anandtech made a mistake in their 3DMark Unlimited Physics graph. The data point they listed as "NVIDIA Shield tablet" in the iPhone 6/6+ preview is actually the data for "NVIDIA Shield [portable]", while the actual Shield tablet Physics score is much higher in comparison.

 

[ams23]

^^^ Re: 6 plus gfx benchmark discussion about on-screen vs. off-screen

The strange thing is that, if you look at the actual GFXBench website, the data they have compiled seems to show that iPhone 6 and iPhone 6 Plus are consistently rendering at the lower iPhone 5S resolution (!) for the GFXBench T-Rex and Manhattan On-screen tests. And most of the other reviews/previews reflect that. Anandtech's On-screen data is more sensible given the actual native screen resolution of these phones, but they are the exception rather than the rule.

 

[Eug]

I contacted one reviewer, who in turn contacted the GFXBench people. Long story short: GFXBench has not yet been updated to support the iPhone 6 or 6 Plus, so their results are not accurate.

 

[CHADBOGA]

Considering how the A7 wowed everyone and the A8 has the benefit of a shrink, this is overall a pretty disappointing effort in my view.

What happened to all that ARM magic pixie dust that was supposed to leave Intel & x86 in its wake?

On the basis of the A8, I don't see ARM challenging x86 on the desktop, anytime soon.

 

[Eug]

Considering how the A7 wowed everyone and the A8 has the benefit of a shrink, this is overall a pretty disappointing effort in my view.

What happened to all that ARM magic pixie dust that was supposed to leave Intel & x86 in its wake?

On the basis of the A8, I don't see ARM challenging x86 on the desktop, anytime soon.

Personally, I just think that was wishful thinking. But then again, that's not the point either. I'm perfectly happy to use x86 on the desktop and ARM in my phone. What's the most interesting to me now though is to see x86 moving toward fanless laptop designs. Core M is what I'm looking to for Apple's next MacBook revamp. This is the crossover point to higher end tablets / 2-in-1s where it could compete against ARM in some instances.

 

[jdubs03]

Clearly Anandtech made a mistake in their 3DMark Unlimited Physics graph. The data point they listed as "NVIDIA Shield tablet" in the iPhone 6/6+ preview is actually the data for "NVIDIA Shield [portable]", while the actual Shield tablet Physics score is much higher in comparison.

Yea, I thought the 2nd graph illustrated that succinctly. But ehh, yea idk. Edit: They fixed it.

 

[Nothingness]

A8 throttling: http://arstechnica.com/apple/2014/09/iphone-6-and-6-plus-in-deep-with-apples-thinnest-phones/3/

The reviewer also says that every day battery life on the 6 plus is better than what their review seems to show.

 

[ShintaiDK]

A8 throttling: http://arstechnica.com/apple/2014/09/iphone-6-and-6-plus-in-deep-with-apples-thinnest-phones/3/

The reviewer also says that every day battery life on the 6 plus is better than what their review seems to show.

Apple sure have another definition of substained performance.

Why am I not surprised.

 

[Nothingness]

Apple sure have another definition of substained performance.

Why am I not surprised.

Indeed Anyway it's still better than A7.

 

[ams23]

Clearly Apple was referring to sustained GPU (rather than CPU) performance, but they didn't exactly spell that out. Anyway, previous Apple iOS products also do not suffer from long-term GPU throttling. Do note that this is due in part to the use of lower power and lower precision FP16 ALU's (modern day desktop, notebook, and console GPU architectures and the games created for them are designed with higher FP32 precision in mind).

On a side note, as analyzed at B3D forum, some of us have come to the conclusion that A8 is using a 4 cluster Series 6XT GX6450 GPU running at a similar GPU clock operating frequency as the 4 cluster Series 6 G6430 GPU in A7. The GPU performance improvement is up to ~ 50% per cluster and per clock, with 33.3% improvement due to an increase in FP16 ALU's, and a bit more than 10% improvement due to improved throughput efficiency.

 

[Eug]

It's almost as if this is an A7.2X.

Minor CPU upgrade, but significant GPU upgrade.

 

[Qwertilot]

Depends if they got that claimed 50 per cent efficiency gain or not I'd think?

Chip of course also only about half cores of one form or another. No idea what the rest is all doing but there does seem to be rather a lot more of it. Presume they had a reason to put it there

 

[Eug]

GFXbench has been updated (but not yet released) as 3.0.2, to include the iPhone 6 and 6 plus. The results for the 6 plus for performance at the native resolution are not great:

Specifically, in "on-screen" native resolution tests using the OpenGL ES 3.0 API, iPhone 6 Plus delivered a score 22 percent lower than last year's iPhone 5s, rendering the "Manhattan" test scene at 19 fps (at its internal 2208x1242 resolution, scaled down to fit its 1920x1080 display), compared to the 24.4fps (albeit at a much lower 1136x640 resolution) achieved by iPhone 5s. iPhone 6 rendered the same scene at 26.6 fps (at its native 1334x750 resolution), delivering a score 9 percent higher than the 5s.

I don't think this will be as big of a problem as the iPad 3 was though. Some might argue the iPad 3 was somewhat gimped even at release.

 

[Eug]

The results seem to be similar for GFXbench 3.0 too, so I guess they did work for native on-screen resolution. This is from AnandTech's review:

Notice how much slower the 6 Plus performs. This is a factor of two difference.