Apple's Next-Generation A8 Chip Said to Top 2 GHz, Remain Dual-Core

[yellowtruck]

I'd say 50% is about right. And Broadwell is getting 30%, so the lead is somewhat shrinking. I just don't think we'll see a point anytime soon, if at all, where we'll look at an i5 or i7 and then to an iPad, and we'll say they're just as powerful.

I do. But what do I know. I'm just thinking in terms of product growth and profit stability. It logically makes sense to gain total control of the very core of ALL of their products to eliminate dependency.
The fact is, they are gaining every year in speed and though I'm being even more, extra imaginative here, they are on a product schedule and its profitable to release iPhones each year with a measurable rate of technical advance over the predecessor chip, never seemingly going beyond a 100% advance over last-years as doing so may not be cost effective in such a high turnover device.
Could they be already separately working on designs that are already far ahead and completely separate than the 100% up in the yearly spec bumps, in-between the upgrade advances in tech spec chip-nip slips?

 

[AdonisSMU]

Please, tell me why my iPhone needs a quad core processor?

I never said an iPhone needed a quad core... Where did you get that from?

 

[spjonez]

Exactly. Which is why a desktop should be faster than a mobile, no matter which CPU architecture is used. (e.g. a new ARM desktop chip should be faster than an Intel mobile device chip, and vice-versa, when using the same semiconductor technology generation).

And power is quite important on a desktop. Ever seen the power and cooling subsystem that a Cray or IBM Z-series requires? Nobody wants that on their desk, even if it ran their apps tons faster.

Lots of PC users just hate the fan noise (one good reason to buy Mac Pros).

I'd take a Cray on my desk any day

For those who are interested, when you see a chip advertised in multiple speeds, say 2.2GHz, 2.4GHz etc. they aren't actually designed that way. Chip fabrication includes a margin of error, the top clock speed for a chip is what every chip was supposed to do, but due to errors in the build process certain areas of the chip are flagged as "bad", which reduces it's clock rate for the chip to remain stable. This is an accepted fact of manufacturing and why we have multiple tiers for every chip.

Heat is very much an issue, but it's not the primary issue holding back processor speeds. Electron leakage is bigger problem than heat and becomes more and more of a issue as die sizes decrease. That's something Intel has been fighting against for a while and ARM will as their die sizes shrink.

 

[yellowtruck]

It hasn't been double every year. :\



Then you'd kidding yourself.

----------



Multiple people saying this doesn't make it any less false.

Wow, you are REALLY on it when it comes to these ARM posts. Hard-stand and wont budge!

----------

"A7's performance is top-notch, there was still room for improvement as almost no iOS app took full advantage of A7's available processing power"
How is it A7's fault… "needs improvement" when apps don’t know how to utilize substantially resources increased?

 

[firewood]

My MacBook Pro scores over 13,000 on Geekbench 3.

I wonder what your MacBook Pro processor would score if you down-clocked the Intel processor enough so that you could run geekbench half a day without the bigger heat spreader or fan but still stay cool enough for your lap.

Unfair handicap? Well that's the handicap that the current A7 has (and rumored A8 will have as well).

Now if Apple had any reason to tape-out an A8UltraXtreme with more cores on a fast process designed to be packaged with honk'n big heat sinks and fans attached... ...could be interesting for Intel.

 

[MacSince1990]

then after this generation the iPad air may score ~5,000.

but my point was not that ARM chips have caught up to MacBook Pro performance, but I was claiming that MacBook Pro's do not need as much performance as they have.

----------



yea right. probably wont see SkyLake until end of 2016, and some won't even be out to 2017.

They kinda do if you do any kind of professional work or are a power user in general.


----------

 

[spjonez]

Unfair handicap? Well that's the handicap that the current A7 has (and rumored A8 will have as well).

Do you have a source for this? I've been using computers since 8086's and I've never heard of a chip that was designed to produce a given clock rate and then intentionally under clocked due to heat. Are you forgetting just how big heat sinks used to be to maintain their advertised speeds?

 

[MacSince1990]

I wonder what your MacBook Pro processor would score if you down-clocked the Intel processor enough so that you could run geekbench half a day without the bigger heat spreader or fan but still stay cool enough for your lap.

Unfair handicap? Well that's the handicap that the current A7 has (and rumored A8 will have as well).

Now if Apple had any reason to tape-out an A8UltraXtreme with more cores on a fast process designed to be packaged with honk'n big heat sinks and fans attached... ...could be interesting for Intel.

In what world does an iPhone last a half a day even with non-demanding tasks..? It wouldn't last 30 minutes running Geekbench nonstop. Also, the ARM wouldn't have integrated graphics, which is a huge part of the cost, transistor count, and power draw (HEAT) of Intel chips.

I.. you know what, the rest of your post is just ridiculous. You have absolutely no idea what you're going on about.

 

[firewood]

I've never heard of a chip that was designed to produce a given clock rate and then intentionally under clocked due to heat...

For the past couple decades, the thermal resistance/dissipation of the chip's expected packaging is one of the first constraints a chip design team takes into account. The range of clock rates expected from the process is set accordingly.

For wafer/die/package testing outside of the final system, the power duty cycle and clock rates are lowered to keep the die temperature within tolerances.

To test Cray-3 parts outside of liquid immersion cooling, they used to power them up only 1 millisecond at a time every few seconds to keep them from melting, for an average clock rate over 1000X lower than production speed.

 

[melendezest]

I see this as yet another step towards convergence (what many apologists call "Apple Focus").

The formula (depending on how soon this happens):

Apple's full control over HW and SW (ie. proprietary-ness) + Macs with this inside + Good enough for most users (supposedly) = Maximum profitability. Apple (and its cheerleaders): win!! + some of us: lose!!

 

[brendu]

I never said an iPhone needed a quad core... Where did you get that from?

Edit: I responded to the wrong post. My bad.

 

[firewood]

Also, the ARM wouldn't have integrated graphics...

Have you looked at the die photo of an A7 and the proportion of the area used for the GPU and video blocks compared to the CPU area? The 2 64-bit ARM CPUs take less than one quarter of the total die area.

And you are correct. Half a day is too long. Most iPhone 5s battery torture tests only run 4 or 5 hours. So set up a MacBook Pro sans heat spreader or fan to run continuously for 4 hours without smoking, and see what benchmark results you get.

 

[spjonez]

For the past couple decades, the thermal resistance/dissipation of the chip's expected packaging is one of the first constraints a chip design team takes into account. The range of clock rates expected from the process is set accordingly.

For wafer/die/package testing outside of the final system, the power duty cycle and clock rates are lowered to keep the die temperature within tolerances.

To test Cray-3 parts outside of liquid immersion cooling, they used to power them up only 1 millisecond at a time every few seconds to keep them from melting, for an average clock rate over 1000X lower than production speed.

Did you copy/paste that from a website? Starting to think you're trolling, that has nothing to do with what I just asked. You didn't answer any of the other questions I posted so I'll depart this thread now.

 

[firewood]

Did you copy/paste that from a website?

Nope. I used to work on ASIC designs, both big/hot, and smaller, low power chips. As for testing the Cray, they have several on display at the Mountain View Computer History Museum, one of the tour guides seemed to know a lot about them, so I asked tons of questions.

 

[mtngoatjoe]

I wasn’t going to say anything, but I’m glad somebody did.

You can take your DIMM and stick it up you a$$. I want a DIMM module to go with my PIN number!

 

[mdriftmeyer]

A7 is on par with Core 2 Duos from 2009. My guess, if the A8 can continue the doubling, it'll be on par with Intel Core i3s from late 2011 (Roughly 4500 geekbench 3 score). Potentially, could catch the MBA's from 2013 (roughly 5000 geekbench 3 score) running the Core i5-Us

With processor increases on the desktop and laptop side slowing down, it could only be another 2-3 years before ARM chips catch and pass "desktop class" chips of the same year.

Nope.

 

[2IS]

A7 is on par with Core 2 Duos from 2009. My guess, if the A8 can continue the doubling, it'll be on par with Intel Core i3s from late 2011 (Roughly 4500 geekbench 3 score). Potentially, could catch the MBA's from 2013 (roughly 5000 geekbench 3 score) running the Core i5-Us

With processor increases on the desktop and laptop side slowing down, it could only be another 2-3 years before ARM chips catch and pass "desktop class" chips of the same year.

A lot of guess work here, and guess work that assumes ARM has some super secret technology that Intel can't figure out that will allow them to continue to double their power every year (we're talking exponential growth here) and magically not run into node and power limitations that the largest and most advanced semiconductor manufacturer in the world is running into.

In a word... No

 

[KillerTree]

I wonder what your MacBook Pro processor would score if you down-clocked the Intel processor enough so that you could run geekbench half a day without the bigger heat spreader or fan but still stay cool enough for your lap.

Unfair handicap? Well that's the handicap that the current A7 has (and rumored A8 will have as well).

Now if Apple had any reason to tape-out an A8UltraXtreme with more cores on a fast process designed to be packaged with honk'n big heat sinks and fans attached... ...could be interesting for Intel.

Broadway already can be used without a fan. Intel has shown Broadwell tablets thinner than iPad Airs

 

[firewood]

A lot of guess work here, and guess work that assumes ARM has some super secret technology that Intel can't figure out that will allow them to continue to double their power every year (we're talking exponential growth here) and magically not run into node and power limitations that the largest and most advanced semiconductor manufacturer in the world is running into.

Intel can figure it out. They've tried at least 3 or 4 RISC architectures in their history. But for good business reasons they've pushed something that grew out of an 8-bit calculator architecture quite far instead.

Apple's iOS processors have not been doubling in performance every year. The rate of improvement since 2008 seems to only average somewhere between 55% and 85% per year, depending on how one measures. But eventually their fab partners will hit the same atomic limits to Moore's law as Intel. Evening things out at that level. Thus eventually potentially allowing the newer arm64+ ISAs to gain a tiny bit more optimization advantages within the same power envelope over the older stuff-bolted-on-to-a-386 ISA.

----------

Broadway already can be used without a fan. Intel has shown Broadwell tablets thinner than iPad Airs

With any benchmark numbers in that package?

Seen any articles like this: http://semiaccurate.com/2014/07/11/intel-castrates-broadwell-gutting-performance/

 

[2IS]

Intel can figure it out.

Not while remaining cost effective and profitable they can't. At least not with the technology available to them currently

 

[KillerTree]

Intel can figure it out. They've tried at least 3 or 4 RISC architectures in their history. But for good business reasons they've pushed something that grew out of an 8-bit calculator architecture quite far instead.

Apple's iOS processors have not been doubling in performance every year. The rate of improvement since 2008 seems to only average somewhere between 55% and 85% per year, depending on how one measures. But eventually their fab partners will hit the same atomic limits to Moore's law as Intel. Evening things out at that level. Thus eventually potentially allowing the newer arm64+ ISAs to gain a tiny bit more optimization advantages within the same power envelope over the older stuff-bolted-on-to-a-386 ISA.

----------



With any benchmark numbers in that package?

Seen any articles like this: http://semiaccurate.com/2014/07/11/intel-castrates-broadwell-gutting-performance/

Show me an article with actual details that aren't behind a paywall.

 

[Michael Goff]

Intel can figure it out. They've tried at least 3 or 4 RISC architectures in their history. But for good business reasons they've pushed something that grew out of an 8-bit calculator architecture quite far instead.

Apple's iOS processors have not been doubling in performance every year. The rate of improvement since 2008 seems to only average somewhere between 55% and 85% per year, depending on how one measures. But eventually their fab partners will hit the same atomic limits to Moore's law as Intel. Evening things out at that level. Thus eventually potentially allowing the newer arm64+ ISAs to gain a tiny bit more optimization advantages within the same power envelope over the older stuff-bolted-on-to-a-386 ISA.

----------



With any benchmark numbers in that package?

Seen any articles like this: http://semiaccurate.com/2014/07/11/intel-castrates-broadwell-gutting-performance/

Semi-accurate dot com?

 

[wizard]

I too find the A7 chip fast enough. What they need to do is add more ram as that would add the most noticeable performance input.

More and FASTER RAM would certainly help all iOS devices! However iPad certainly could leverage more cores! I'm fairly perplexed by people that say A7 is fast enough. It isn't even close im my estimation.

 

[wizard]

"almost no iOS app took full advantage of A7's available processing power" -Anandtech

That statement is a bunch of BS designed to rub the fur the wrong way on sensitive individuals. First off the processor is or was brand new at the time so yeah no app takes advantage of it. The very same can be said about Intel processors when released with new capabilities. Second; all of Apples apps shipping with A7 based systems are 64 bit so they are taking advantage of the processor. Third; if an app used 100% of A7's capability for any length of time it would become a huge drain on the battery.

That's coz developers are still focused on supporting all previous hardware to have the biggest market possible for more profits. Its only logical.

It isn't even that, at the time the article was written it wasn't possible for developers to have a large number of apps available to take advantage. In a nut shell the statement's make about as much sense as complaining about apps not taking advantage of the latest Intel processor.

But kudos to Apple for continuing with their processor roadmap. I hope this leads credence to more planned features like "desktop class" enhancements and eminent merging a few months from now. Mobile Era.

What merging! I see nothing in Apples development of iOS and Mac OS to indicate merging in the future. It would be absolutely asinine on Apples part to do so.

----------

No way!

It might be possible!

 

[wizard]

I think we're at the point where iOS hardware is outpacing the software. A8 will be wicked fast (approaching Macbook Air speeds in terms of raw processing power), but there are very few apps that can even use the A7 to its full potential.

This really isn't true! There has been a steady flow of new software optimized for A7. Developers have been getting onboard as fast as they can. That is what is happening today, with iOS 8 even more apps will leverage the platform.

In a nut shell if A7 makes something possible in an App that can't be done on older processors then the app is taking advantage of A7.