• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Apple A9X the new mobile SoC king

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Oh, I was talking about Core M CPU vs console CPU.
NVidia compared the Tegra X1 with a XBox360. A 10 years old console. So console class really doesn't mean anything
 
ShintaiDK said:
Since A9/A9X got 2xbandwidth and the speedup isnt specified. Then people should be really careful in where the performance comes from. It could also simply be vector instructions.
Click to expand...

Not sure how accurate this year's performance claims are (we'll have to wait and see) .. but last years performance increase claims were underestimated.

Apple claimed up to 40% increase in CPU and up to 2X graphics increase when comparing a8X and a7 in the iPad Air.

Well CPU increase by more than 75% in geekBench multicore and >50% in several other multicore benchmarks mainly because of increase in core counts.

GPU increase by 2.5X in gfxBench Manattan offscreen using same drivers. from 13fps -> 32.5fps.

The a8X actually got another increase to 37.5fps with a driver upgrade (this driver upgrade did not increase a7 performance). And using Metal got 41fps.

So - Apple does not have a history (at least going by last year's claims) of over inflating the performance claims.
 
IntelUser2000 said:
GPU will be way faster too. HD 5300 in the Broadwell Core M gets 22.9 fps and 60.5 fps in Manhattan and T-Rex offscreen respectively. iPad Air 2's A8X gets 37.6 fps and 71.4 fps respectively.

Intel claims that in 3DMark Sky Diver, Skylake will be 40% faster. If that carries onto GFXBench, and assuming A9X is 2x A8X, A9X graphics will be 134% faster. That's 2.34x. I believe if that carries true everywhere that means Iris 540(15W, eDRAM 48EU) performance.

Quad core 45W, mobile Sandy Bridge performance. A Quad core Haswell 4900MQ gets 13k while 4670K Desktop gets 13k too. 8K does quality as desktop performance.
Click to expand...

I bet Intel execs must be losing sleep tonight over about not only Apple's phone business is already bigger than the whole of Intel, now they also have the better mobile chip supported by a tailor made software ecosystem.
 
thunng8 said:
Not sure how accurate this year's performance claims are (we'll have to wait and see) .. but last years performance increase claims were underestimated.

Apple claimed up to 40% increase in CPU and up to 2X graphics increase when comparing a8X and a7 in the iPad Air.

Well CPU increase by more than 75% in geekBench multicore and >50% in several other multicore benchmarks mainly because of increase in core counts.

GPU increase by 2.5X in gfxBench Manattan offscreen using same drivers. from 13fps -> 32.5fps.

The a8X actually got another increase to 37.5fps with a driver upgrade (this driver upgrade did not increase a7 performance). And using Metal got 41fps.

So - Apple does not have a history (at least going by last year's claims) of over inflating the performance claims.
Click to expand...

A8X also doubled the memory speed over A7. From the looks of it, A9X will have ~50GB/sec bandwidth.
 
ShintaiDK said:
GPU vs CPU part.

Apple%20September%202015-165_575px.jpg
Click to expand...

PS3 is still selling. So yeah Apple is not wrong there in claiming console class GPU. They never said PS4 class GPU. :biggrin:
 
raghu78 said:
PS3 is still selling. So yeah Apple is not wrong there in claiming console class GPU. They never said PS4 class GPU. :biggrin:
Click to expand...

Thats why PR slides like this cant be trusted. And even if it compares to something, it may only be in a very narrow spectrum.
 
ShintaiDK said:
Thats why PR slides like this cant be trusted. And even if it compares to something, it may only be in a very narrow spectrum.
Click to expand...

Apple A9X is going to be the most powerful tablet SoC for CPU performance by a good margin. For GPU performance A9x will be behind Nvidia Tegra X1 which uses Maxwell. A9x will still be definitely at the top of the mobile SoC charts for 2015/2016. I am keen to see A9X vs S820 vs Tegra X1 in GPU comparisons. Tegra X1 will be disadvantaged in CPU comparisons as it is 20nm planar process based while the other 2 are 16/14nm FINFET based. But Maxwell is vastly superior and so Tegra X1 will dominate graphics performance charts even while being on a inferior process node.
 
What I find to be interesting is that while most companies prefer to avoid overly-large dies, Apple has no qualms over nuking the competition with gpu-size mobile chips. No doubt, they are paying dearly for the ability, though they certainly can pay for it without a second thought.
 
Let me understand this. The A8X is 50% larger than Core M and is fabricated on either the world's most expensive foundry technology or the world's most expensive non Intel fabrication technology. It outperforms Core M CPU by 38% (8000/5800). Golly gee and all but at face value that sounds like one hand clapping. Once again, a bigger, more expensive SoC performs better than a smaller, cheaper chip. Testing may reveal it actually is better than it sounds, but the converse is also possible. At least efficiency should improve.
 
Zodiark1593 said:
What I find to be interesting is that while most companies prefer to avoid overly-large dies, Apple has no qualms over nuking the competition with gpu-size mobile chips. No doubt, they are paying dearly for the ability, though they certainly can pay for it without a second thought.
Click to expand...

Apple is building chips that it wraps in extremely expensive, high margin devices. This gives Apple a degree of freedom that a merchant chip vendor like Intel or Qualcomm might not have.
 
Not proven of course, but it does not take a stretch of imagination to imagine that a company (as Apple are) with a basically unlimited budget and a bunch of already proven very effective design teams have done very well.

Especially when they're also much more targeted in terms of what they're trying to do than either Intel, or the vanilla arm designs.
 
dealcorn said:
Let me understand this. The A8X is 50% larger than Core M and is fabricated on either the world's most expensive foundry technology or the world's most expensive non Intel fabrication technology. It outperforms Core M CPU by 38% (8000/5800). Golly gee and all but at face value that sounds like one hand clapping. Once again, a bigger, more expensive SoC performs better than a smaller, cheaper chip. Testing may reveal it actually is better than it sounds, but the converse is also possible. At least efficiency should improve.
Click to expand...


The price Intel sells the Core M for is quite obscene. And people get on Apple over their margins.
 
scannall said:
The price Intel sells the Core M for is quite obscene. And people get on Apple over their margins.
Click to expand...

The price Intel sells Skylake Core M is not public AFAIK.
From ARK - Recommended Customer Price: N/A

Do we know that price? Like, I know it is listed on Ark, but I also know that Ark isn't what things actually sell for.
Click to expand...

The funny thing is it's not even on ARK.
 
dahorns said:
Do we know that price? Like, I know it is listed on Ark, but I also know that Ark isn't what things actually sell for.
Click to expand...

Well the ARK-price is the only one we have to go by. OEMs can for sure negotiate individual deals, but that is true for all CPUs.

According to the ARK price, 5Y71 is $281. And that is obscene! No price is listed for 6Y75 yet.
 
i3 4010U cost 281$ on ARK.

A NUC with an i3 4010U cost 278$ on ARK.

So stop referring to it at all when it is pointless.
 
dealcorn said:
Let me understand this. The A8X is 50% larger than Core M and is fabricated on either the world's most expensive foundry technology or the world's most expensive non Intel fabrication technology. It outperforms Core M CPU by 38% (8000/5800). Golly gee and all but at face value that sounds like one hand clapping. Once again, a bigger, more expensive SoC performs better than a smaller, cheaper chip.
Click to expand...

But Intel isnt cheaper. There is no doubt that Apple is getting those huge SoCs for less than the cost of one tiny Intel skylake chip.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top