Apple A8x

[Roland00Address]

They failed in the $99 x86 tablet claim. The only $99 x86 tablets we have are bargain bin throwaways from companies who tried and failed and are forced to fire sale their product. We'll never have $99 x86 tablets as long as intel is charging a 60% margin, and intel will never compete in mobile with 60% margins either. There are half a dozen companies out there who are willing and happy to accept margins as low as 15%. Intel cant compete with that.

First they never claimed $99 for fact they said this.

We also heard from the CEO that we may see tablets as low as $99. Depending on how far the OEMs want to push it but the BOM now allows such price points.

Actual devices on the market.

$99 Toshiba http://www.newegg.com/Product/Produc...82E16834216801
$124 ASUS http://www.newegg.com/Product/Produc...82E16834231792
we are about to see the HP Stream 7 for $99 and acer is going to release its B1-750 and that is expected to go for 100 Euros who knows the US prices. All the skus I just mentioned are baytrail.

If you allow clovertrail and not baytrail to the picture you suddenly have shipping devices from ACER, Dell, and HP added to that sub $129 price point.

And if you want to add the crap brands from this you can get an intel atom processor with 2gb of ram for $68 currently on newegg.

 

[witeken]

But a $700 with 16/32gb?

A premium priced phone with only 16GB in late 2016?! If Apple doesn't want to pay a few bucks for at least 16GB more, than surely they won't put a premium processor in the phone either.

On the other hand, how many phones with a price higher than $700 exist? Intel can't price itself completely out of the market. There are plenty of Core CPUs in cheap laptops (<$500), so Core in a $500 phone doesn't seem like an impossibility to me.

 

[asendra]

Damn! ?? If true, competes in single core with Denver @2,5Ghz, and blows everything out of the water in multi core.

http://browser.primatelabs.com/geekbench3/search?utf8=&#10003;&q=ipad5

Air 2 Vs Air
http://browser.primatelabs.com/geekbench3/compare/1061742?baseline=1061162

3 cores @ 1,5Ghz, 2Gb RAM.

It also shows way better memory performance, as the previous passmark test did (although not as brutal..)

Confirmed then?

 

[liahos1]

Damn! ??

http://browser.primatelabs.com/geekbench3/search?utf8=&#10003;&q=ipad5

3 cores @ 1,5Ghz, 2Gb RAM.

Confirmed then?

pretty cool. i wonder what the die size on that is. now i can refresh anandtech forum pages that much faster :hmm:

 

[Roland00Address]

Damn! ?? If true, competes in single core with Denver @2,5Ghz, and blows everything out of the water in multi core.

Confirmed then?

No not confirmed.

It is so easy to spoof geekbench. http://browser.primatelabs.com/geekbench3/search?dir=desc&q=samsung+galaxy+note&sort=score

Notice all those samsung galaxy notes with intel core i7.

And if you want to go further with spoofing geekbench you start editing files such as the SMBios file.

 

[Mopetar]

pretty cool. i wonder what the die size on that is. now i can refresh anandtech forum pages that much faster :hmm:

I think they said ~3 billion transistors during the presentation, so if the A8 is ~2 billion, the A8x is probably about 50% larger though the size increase probably depends on exactly what was added and compact it can or cannot be made.

 

[asendra]

No not confirmed.

It is so easy to spoof geekbench. http://browser.primatelabs.com/geekbench3/search?dir=desc&q=samsung+galaxy+note&sort=score

Notice all those samsung galaxy notes with intel core i7.

And if you want to go further with spoofing geekbench you start editing files such as the SMBios file.

Yeah, I know it can be spoofed, but where's the fun in that?

We have a Geekbench dev around, maybe he can verify the entry (he isn't leaking anything right? should be no harm in that xD)

 

[pw257008]

No not confirmed.

It is so easy to spoof geekbench. http://browser.primatelabs.com/geekbench3/search?dir=desc&q=samsung+galaxy+note&sort=score

Notice all those samsung galaxy notes with intel core i7.

And if you want to go further with spoofing geekbench you start editing files such as the SMBios file.

user looks legit, though

 

[asendra]

user looks legit, though

Yeah, and by the looks of it, it seems it's not the first time he tests an Apple device before it being entered in the geekbench name database.

 

[liahos1]

I think they said ~3 billion transistors during the presentation, so if the A8 is ~2 billion, the A8x is probably about 50% larger though the size increase probably depends on exactly what was added and compact it can or cannot be made.

133 mm^2 is massive. approaching intel ivybridge levels. can't imagine its a cheap chip on 20nm.

 

[jfpoole]

No not confirmed.

It is so easy to spoof geekbench.

It's easy to spoof Geekbench on Android, but it's tricky to spoof on iOS. I'm convinced the 3-core iPad Air result is legitimate.

 

[asendra]

133 mm^2 is massive. approaching intel ivybridge levels.

Ipad 3 had a die area of 160 or so, If I remember correctly, so It would not be the first time they go "big".

It's easy to spoof Geekbench on Android, but it's tricky to spoof on iOS. I'm convinced the 3-core iPad Air result is legitimate.

And he speaks!! Good to hear, we'll know soon enough.

 

[Arachnotronic]

Great job, Apple's silicon team(s)!

 

[Roland00Address]

It's easy to spoof Geekbench on Android, but it's tricky to spoof on iOS. I'm convinced the 3-core iPad Air result is legitimate.

I can believe it is legit, but belief is not confirmation Whatever devices will soon be out.

 

[asendra]

Those single core results would be comparable to core 2 duo macs @ 3ghz from 2009, and multi core results are comparable to dual core macbook pros from 2010, and macbook airs from 2011, fyi.

Actually, this thing is pretty close to my 2012 Macbook Air, which is my main machine... Those Broadwell rMPB can't come soon enough.

 

[Eug]

Single-core IPC is increased over A8 (slightly).

L2 cache is doubled from 1 MB (A8) to 2 MB (A8X).

So, 50% more CPU cores and 50% more GPU cores, so it makes sense there are 50% more transistors at over 3 billion.

This thing is also significantly faster than my main laptop, which is a 2.26 GHz Core 2 Duo P8400 13" MacBook Pro from 2009, running Yosemite, and running it quite well I might add.

 

[Arachnotronic]

A8X likely has GX6650, obviously 3x Enhanced Cyclone confirmed. Wouldn't be surprised to see a 6-8MB L3$.

 

[teejee]

Let's do some math.

It costs about $300 million to develop an SoC in sunk, fixed R&D costs. If you want to be really, really generous, let's say it's $200 million.

No way a budget SOC for example from Mediatek costs $200 million to develop. Not even close.

 

[Mopetar]

133 mm^2 is massive. approaching intel ivybridge levels. can't imagine its a cheap chip on 20nm.

It might not be a 50% increase in die area if all the additional transistors are more densely packed. It's still going to be a big chip, but the A5X was over 160 mm^2, so it would hardly be the biggest chip Apple has made.

As far as the cost goes, I think Apple can afford the margins.

 

[krumme]

It all depends on how much of a margin hit Apple (or...) wants to take. A $1000 iPhone with 256GB, Cannonlake and Gen10 should be easily possible with sane margins.

Does it look like thats the strategy for apple?
In my world they want the profit themselves. Each an every penny. And every quarter of a penny.

Its the exact oposite happening. There is absolutely no sign other than they will continue what they set out 4 years ago. And why wait 4 years when they have the needed power right in their hands now? Makes no sense. They have succeded.

 

[Eug]

133 mm^2 is massive. approaching intel ivybridge levels. can't imagine its a cheap chip on 20nm.

It could be that the initial batch will be done on 20 nm and then the next batch will be done on 14 nm, with the same iPad innards and same battery, for a longer battery life. This wouldn't be the first time they've done that.

I do know that they have a 14 nm contract with global foundries and Samsung apparently.

This also makes me wonder if the original intent for the iPad mini 3 was to have it get a 20 nm A8 (not A8X), but due to deficiencies in 20 nm volume production and the stellar sales of the iPhone 6 and 6+, they chose to use the old A7 instead. If I remember correctly, there was an article a while back suggesting that the iPads or Pro may be delayed because of iPhone 6 demand.

 

[ams23]

Damn! ?? If true, competes in single core with Denver @2,5Ghz, and blows everything out of the water in multi core.

Hard to compare accurately with Denver because we don't have AArch64 results yet for that CPU core. Also note that Denver should have a real-world peak CPU clock operating frequency of 2.3GHz, irrespective of what Geekbench 3 shows in their info table, according to a forum member at B3D. By the time you factor in AArch64 floating point improvements to single-core and multi-core scores, Denver should be clearly ahead of enhanced Cyclone in single-core performance while not being very far behind in multi-core performance (and most mobile apps probably do not make good use of more than two CPU cores in the first place).

It looks like the Geekbench 3 single-core CPU performance of A8X in iPad Air 2 is ~ 12% higher overall than than that of A8 in iPhone 6, with most of the difference due to a higher CPU clock operating frequency in iPad Air 2.

With Cyclone and enhanced Cyclone CPU, the Geekbench 3 dual-core score appears to be ~ 1.777x higher than the single-core score. In other words, for every doubling of cores, the Geekbench 3 multi-core score should increase by ~ 1.777x. Since there are three cores in iPad Air 2, the multi-core score is ~ 2.47x higher than the single-core score ( 2.47 = 1.777*(1+1.777)*0.5 ).

 

[Eug]

64-bit speeds in Geekbench should be several percent higher than 32-bit from what I gather.

---

The specs have been unofficially confirmed now:

https://twitter.com/gruber/status/524727681334935552

iPad Air 2, the specs:

3-core CPU (Geekbench: 1813 single-core, 4539 multi-core), 2 GB RAM

Those scores are even higher (slightly) than the original Geekbench leak.

Also, there are now multiple iPad5,4 / iPad Air 2 entries in the Geekbench browser:

http://browser.primatelabs.com/geekbench3/search?q=ipad5

And also, benches from Engadget's review:

 

[Zink]

A8X single threaded performance is getting stupidly fast. With a ~4W chip at 1.5 GHz they are hitting 70% of the performance of a 2.6 GHz 15W i5? Put that in a Mac already!

3 Billion transistors is insane for a low end consumer device. Even the base model 15" MBPR's $400 chip has 1.7B transistors for CPU + GPU and then some more for L4 eDRAM. All of Apples 13" and smaller notebooks use 1.3B transistors.

It seems like there would be huge advantages for Apple if they could do their own custom 5B+ transistor chips for ultraportable laptops compared to using Intel&#8217;s slightly faster x86 chips. They would get the opportunity to add lots of fixed function hardware and custom features tuned to the OS while also keeping costs lower than using Intel.

The only big downside to that would be x86 software compatibility. Apple&#8217;s Mac sales are still growing and AMD isn't going to be making big profits from their x86 CPUs any time soon. Would AMD&#8217;s x86 license agreements extend to allowing Apple to somehow get a custom x86 SOC made? They could acquire part of AMD, pay AMD for a custom chips like current gen consoles etc. Apple could spend the money to redo their current architecture but with x86 instructions.

On the other hand OSX is getting a bit old. Maybe Apple will wait and make a big jump to ARM when the time is right.

 

[ams23]

Is it the CPU Physics score that is still holding back A8X in 3dmark Ice Storm Unlimited overall score? The third CPU core should have significantly improved the physics score. The Graphics score should be close to mobile Kepler, but based on a combined score of 21,659, it appears that the Graphics score is less than 30,000...