Apple's new A8X processor dominates the benchmarks

[Captain_Slow]

so the iPhone 6, a dual core, is faster than the Galaxy s5, a quad core. Seems like I was right all along, Apple's dual cores>Snapdragon quad cores. Haters come at me.

 

snapturtles, as they shall now be known

[Captain_Slow]

It ain't no monster, the A8X trades blows in single core performance with the 801. The 810 is coming to the market next year packing eight (4+4) 20nm cores. So not long after the new Apple devices go on sale Qualcomm will be shipping these monsters for commercial use. I rather wait 4-6 months for an Android device with what will be a much faster SoC (Android 5.0 also supports 64-bit).

 

I bet the 810 will have shitty single core performance. 8 cores is just plain stupid.

 

also, the a8x doesnt trade blows with a 801, it destroys it in single core, as did the a8, and the a7 before it.

[Captain_Slow]

So are you saying the ipad air 2 beat the nexus 9 by 0.28% in a benchmark that many say has no real world value yet this is considered domination by some?

 

it's considered domination because the a8x is 25% better in multicore and the Nexus 9 or the Nvidia Denver K1 is the only thing that can hold a candle to the a8x

[LegendKillerRG]

Consider this the iPad and iPhone are clocked at half the speed of the nexus 9 and yet has similar single threaded performance.

It's like AMD 8 cores against Intel 4 Cores basically...

[LukaP]

ive done some quick math and it seems the A8X has some IPC improvements as well, compared to the A8, since its more than 7% better

[Mew]

It's like AMD 8 cores against Intel 4 Cores basically...

Exactly, it's like an Intel dual core vs an AMD quad core. I'll take the Intel dual core any day.

[LAwLz]

Single core:

Nexus 9 - 1940

iPad Air 2 - 1812

Note 4 - 1273

 

Multi core:

iPad Air 2 - 4477

Note 4 - 4442

Nexus 9 - 3297

 

I am very impressed but there is 1 thing you have to remember. Geekbench on iOS is ARMv8 optimized but the Android version is still on ARMv7.

In terms of integer performance we saw an average of 88% performance increase in Geekbench 3 on the iPhone 5S (by simply changing from ARMv7 to ARMv8, and running the test on the exact same phone). The FP performance was an increase of ~45%.

 

If you remove the biggest gains (new crypto instructions) the results become a big more realistic but the int performance increase was still an ~8% increase and in FP it was an ~17% increase. These are the performance increases when you remove things like a tripling in SHA1 performance, and an x8 performance increase in AES.

So I don't think it's unreasonable to say that with a simple update to the Android version of Geekbench we might see the results become a bit more even.

(Edit: Forgot the source)

 

 

I give Apple two thumbs up for not only increasing the RAM to 2GB (fucking finally) but also making it a 3 core CPU. That explains the big difference in transistor count between the A8X and the A8.

Geekbench isn't the be-all and end-all benchmark, but these results aren't that far off from the Macbook Air 2011 (which I thought was a really shitty laptop but whatever, lots of people like it). That shows how good these ARM devices are getting.

 

Phone Arena are also quoting some 3D Mark Ice Storm Unlimited numbers at 21,660. For comparison, the K1 gets about 25,700.

[LAwLz]

I bet the 810 will have shitty single core performance. 8 cores is just plain stupid.

 

also, the a8x doesnt trade blows with a 801, it destroys it in single core, as did the a8, and the a7 before it.

8 cores is one of the best ideas ever for smartphones.

The Snapdragon 810 is very similar on the CPU side to the Exynos 7 Octa. Four low clocked Cortex A53 cores and four higher clocked Cortex A57 cores and you can schedule threads to any combination of cores.

The Note 4 benchmarks I posted has the A53s running at 1.3GHz and the A57s running at 1.9GHz (although they are probably thermal throttling in that tiny body during that long test).

 

We don't know the clock for the 810 but they will probably be similar to the Exynos. It should be a massive upgrade in performance and power usage compared to the Snapdragon 805 (which has basically the same CPU as the Snapdragon 800, which came out over a year ago).

[TroubleKlef]

How is it that Apple's SoCs are outperforming competitors with less cores and a much lower clockrate? They can't be THAT far ahead can they?

[Opcode]

I bet the 810 will have shitty single core performance. 8 cores is just plain stupid.

 

also, the a8x doesnt trade blows with a 801, it destroys it in single core, as did the a8, and the a7 before it.

The 810 uses the big.LITTLE configuration, so it's not exactly the same thing as a full blown octa-core.

[mr moose]

Consider this the iPad and iPhone are clocked at half the speed of the nexus 9 and yet has similar single threaded performance.

 

Why consider only that? why not consider the iteration of most android processors and why not consider that this is only one benchmark that again is rather arbitrary by many peoples reasoning?

 

As has been pointed out throughout this thread that yes it's moving forward, yes it's better, but does domination spring to mind? no one claims domination when nvidias new gpus are a few FPS in front of AMDs equivalent that are half the price.  But it doesn't dominate beyond reasonable expectation of any new device.

 

it's considered domination because the a8x is 25% better in multicore and the Nexus 9 or the Nvidia Denver K1 is the only thing that can hold a candle to the a8x

 

That's not very qualitative, your saying it's 25% percent better if you exclude the two nearest competitors.  That's not a very realistic look at how things are fairing.

 

 

How is it that Apple's SoCs are outperforming competitors with less cores and a much lower clockrate? They can't be THAT far ahead can they?

 

It's easy that they are faster, apple are a successful company that design successful products, every iteration they produce on whole will be faster, or use less power, have new features or just in general be better.  But no, I wouldn't expect anyone's new product to be THAT far in front. In the 30 odd years I have been a tech fanatic I have yet to see a product released to market that is exceedingly more advanced that the competitor. I'm sure a few products have come close but they have had a price tag that makes them unobtainable to most consumers.

[LukaP]

In the 30 odd years I have been a tech fanatic I have yet to see a product released to market that is exceedingly more advanced that the competitor. I'm sure a few products have come close but they have had a price tag that makes them unobtainable to most consumers.

Sandy bridge springs to mind though. that was such a big jump 

[Mew]

Sandy bridge springs to mind though. that was such a big jump

g92(8800gt, 8800 ultra) and 5870.

[mr moose]

Sandy bridge springs to mind though. that was such a big jump 

I didn't think it was such a big step.

 

This is just an example of what I an trying to say, not my resting evidence:

 

http://www.tomshardware.com/reviews/cpu-performance-comparison,3370-11.html

 

While the new architectures are at the top of the chart there is no sudden increase in performance that stands out.

[LukaP]

I didn't think it was such a big step.

 

This is just an example of what I an trying to say, not my resting evidence:

 

http://www.tomshardware.com/reviews/cpu-performance-comparison,3370-11.html

 

While the new architectures are at the top of the chart there is no sudden increase in performance that stands out.

fair enough. but it was still a bigger jump than before if i remember correctly

[Mew]

Why consider only that? why not consider the iteration of most android processors and why not consider that this is only one benchmark that again is rather arbitrary by many peoples reasoning?

Because the other "android processors", to be correct-snapdragon, have very low single threaded performance, as evident in the graph in the op. As for multithreaded performance, it's about even with the quad core snapdragons, also in the op as well as shown by other benchmarks. http://www.anandtech.com/show/8554/the-iphone-6-review/5

Although the anandtech review is on the iPhone 6, single threaded performance should be similar to the iPad air 2, and multithreaded should be about 50% better based on scaling.

[mr moose]

Why consider only that? why not consider the iteration of most android processors and why not consider that this is only one benchmark that again is rather arbitrary by many peoples reasoning?

Because the other "android processors", to be correct-snapdragon, have very low single threaded performance, as evident in the graph in the op. As for multithreaded performance, it's about even with the quad core snapdragons, also in the op as well as shown by other benchmarks. http://www.anandtech.com/show/8554/the-iphone-6-review/5

Although the anandtech review is on the iPhone 6, single threaded performance should be similar to the iPad air 2, and multithreaded should be about 50% better based on scaling.

 

I don't think we are talking about the same thing.  I refer to a benchmark as being rather arbitrary  and you link me to another one which shows pretty much the same thing, a newer processor doing better.  Isn't that what we should expect?  It's what I was expecting.  I am responding to the terminology people are using, I don't consider this domination, domination is total control out of the box, not average improvements for the next iteration of a  processor.   It seems every time  a new product is released that is faster, more efficient or whatever, someone reviews it or posts it and claims domination, or that it annihilates the competition, when the reality is that its just the next iteration and on the scale of advancement not outside of the curve.

[LukaP]

single threaded performance should be similar to the iPad air 2, and multithreaded should be about 50% better based on scaling.

Single threaded is improved. if i remember correctly, pure clock speed boost should be 7%, but its actually 13% in that graph. 

[mr moose]

fair enough. but it was still a bigger jump than before if i remember correctly

 

Which is pretty standard with technology evolution, consider the advances made over 5 years back in the 80s versus 5 years today.  In another 20 years we should be able to make improvements in one iteration that would have taken several years with today's technology.   If that makes any sense?

[LukaP]

Which is pretty standard with technology evolution, consider the advances made over 5 years back in the 80s versus 5 years today.  In another 20 years we should be able to make improvements in one iteration that would have taken several years with today's technology.   If that makes any sense?

it does. its called exponential growth in short hehe

[Valentyn]

Battery life, and even plain wattage consumption are considerations. No one who does more than casual editing would think to render on a tablet or smartphone... or would they? I'll be damned if people do. They're not computers; they're computer-like toys. They can do some interesting things and also show some moderate productivity capabilities for many consumer subgroups, but at the end of the day they're still blocked in in some way or another.

 

iPads already have very good batteries and wattage consumption.

 

Also sure they may be "computer-like toys" for now, but at the rate ARM tech is increasing yearly it won't belong before you can happily edit decently on it.

 

The fact that Apple already has the iLife, and iWork suit on there, along with MS Office, and now Pixelmator it all only benefits the end consumer.

 

There should never be a reason to want to halt progress because of a personal opinion that they're simply toys.

 

If someone wants to edit on an iPad they can, nothing to stop them. There are already people that use it as a digital sketchpad with good results.

I don't see why they should increase the uses and capabilities of tablets, and even smartphones.

[mr moose]

it does. its called exponential growth in short hehe

 

Yes, having issues with my worms this morning.  :unsure:  I seem to be using redundant vocabulary in an effort to articulate a specific point, yet it appears I am declining into a perniciousness diatribe at times.

[Mew]

I don't think we are talking about the same thing.  I refer to a benchmark as being rather arbitrary  and you link me to another one which shows pretty much the same thing, a newer processor doing better.  Isn't that what we should expect?  It's what I was expecting.  I am responding to the terminology people are using, I don't consider this domination, domination is total control out of the box, not average improvements for the next iteration of a  processor.   It seems every time  a new product is released that is faster, more efficient or whatever, someone reviews it or posts it and claims domination, or that it annihilates the competition, when the reality is that its just the next iteration and on the scale of advancement not outside of the curve.

 

People are under the very common misconception that the Snapdragon 80x is much faster than the A7, but in reality the A7 has much higher single core performance despite the lower clock speed and very similar multicore performance despite being down 2 cores. As it is right now, the A7 has no competition for single core performance and the A8 improves upon that. Only the Tegra k1 can match Apple in single threaded performance, but that's not ready for mobile phones. For me, the two most noticeable things when using a new cpu is the improved IPC and efficiency-Apple is able to do both. So in short, Apple is able to deliver high performance cores at low power consumption (just like Intel) while Snapdragon requires more cores to match the multithreaded performance of the A7, which requires highly optimized applications. The mobile space right now is very very similar to the current situation between Intel and AMD; there's essentially no competition between Snapdragon and the A7/A8.

 

Single threaded is improved. if i remember correctly, pure clock speed boost should be 7%, but its actually 13% in that graph. 

 

Interesting this might be like last year where the iPad mini with retina and the Air had the same clocks but the mini throttled, resulting in lower performance. Maybe the 6 is throttling under heavy loads, whereas the Air 2 has more surface area to dissipate the heat and able to maintain the higher clocks?

[Captain_Slow]

8 cores is one of the best ideas ever for smartphones.

The Snapdragon 810 is very similar on the CPU side to the Exynos 7 Octa. Four low clocked Cortex A53 cores and four higher clocked Cortex A57 cores and you can schedule threads to any combination of cores.

The Note 4 benchmarks I posted has the A53s running at 1.3GHz and the A57s running at 1.9GHz (although they are probably thermal throttling in that tiny body during that long test).

 

We don't know the clock for the 810 but they will probably be similar to the Exynos. It should be a massive upgrade in performance and power usage compared to the Snapdragon 805 (which has basically the same CPU as the Snapdragon 800, which came out over a year ago).

 

8 core sucks balls, big.LITTLE is for OEMs who cannot be bothered to scale power for their cores, make strong proper cores, and want to impress dumb people. 

 

Applications will use one quad-core or the other, not both. 

[Captain_Slow]

The 810 uses the big.LITTLE configuration, so it's not exactly the same thing as a full blown octa-core.

 

I know, samsung already has that, and it's amazing in single core performance.

 

 

oh wait, no it isn't...