Apple A7 is now 64-bit

[Eug]

The fact that the didn't mention amount of RAM means nothing, for or against 1 GB, vs 2 GB. As thunng8 says, they didn't mention last time around either. In fact, I don't think they've ever mentioned RAM increases for the iPhone:

iPhone: 128 MB
iPhone 3G: 128 MB
iPhone 3GS: 256 MB
iPhone 4: 512 MB
iPhone 4S: 512 MB
iPhone 5: 1 GB
iPhone 5C: 1 GB
iPhone 5S: 1 GB ?

This does make me lean a little more towards the 16 GB model of the 5S though. I figure with 1 GB RAM, I might be inclined to keep it for a shorter time than I would with a 2 GB device so I may as well go with a cheaper model, esp. since my storage requirements have only been slowly creeping up on phones.

The only thing is that if 64-bit is a little more inefficient overall in terms of RAM and storage usage, that means the older 5 and 5C actually have a slight RAM advantage over the 5S.

I suspect the next iPhone 6 may use a descendant of this A7, but will gain 2 GB RAM. 4 GB RAM is a long way off.

 

[Nothingness]

Apple didn't say anything about RAM the last time they doubled it, so why should they say it now?

And even then if it has only 1 GB, why would that mean they shouldn't have switched to 64-bit now? They have the CPU, it seems to perform well both performance wise and battery wise, so switch and let devs get familiar with ARMv8. Add to that marketing buzz and it can't be denied it is a wise move.

People thinking they should have waited to hit the memory wall are dumb or just plain Apple haters.

 

[Eug]

Convert Your App to a 64-Bit Binary After Updating It for iOS 7

Xcode can build your app with both 32-bit and 64-bit binaries included. This combined binary requires a minimum deployment target of iOS 6 or later. The 64-bit binary runs only on iOS 7 or later. If you have an existing app, you should first update your app for iOS 7 and then port it to run on 64-bit processors. By updating it first for iOS 7, you can remove deprecated code paths and use modern practices. If you are creating a new app, target iOS 7 and compile 32-bit and 64-bit versions of your app.

Apparently, fat binaries with 64-bit are not yet supported in iOS 6, but will be in October.

http://9to5mac.com/2013/09/16/apple...t-submitting-iphone-5s-optimized-64-bit-apps/

You can submit 64-bit apps for iOS 7 today that take advantage of the power of iPhone 5s. Xcode can build your app with both 32-bit and 64-bit binaries included so it works across all devices running iOS 7. If you wish to continue to support iOS 6 then you will need to build for 32-bit only. Next month we will be making changes that will allow you create a single app binary that supports 32-bit on iOS 6, as well as 32-bit and 64-bit on iOS 7.

I guess that means iOS 6.1.5/6.2.

EDIT:

Weird. Apple pulled the message. It's gone now.

 

[bullzz]

http://anandtech.com/show/7335/the-iphone-5s-review/5

this is huge. at this speed, they might hit haswell performance in 3-4 generations

 

[Haserath]

So ArmV8 is a completely separate instruction set. And also comes with some performance improvements.

The DGEMM operations aren't vectorized under ARMv7, but they are under ARMv8 thanks to DP SIMD support so you get huge speedups there from the recompile. The SFFT workload benefits handsomely from the increased register space, significantly reducing the number of loads and stores (there's something like a 30% reduction in instructions for the A64 codepath compared to the A32 codepath here). The conclusion? There are definitely reasons outside of needing more memory to go 64-bit.

http://www.anandtech.com/show/7335/the-iphone-5s-review/4

And Apple's A7 competes with Baytrail at a much lower frequency.

 

[Zodiark1593]

So ArmV8 is a completely separate instruction set. And also comes with some performance improvements.


http://www.anandtech.com/show/7335/the-iphone-5s-review/4

And Apple's A7 competes with Baytrail at a much lower frequency.

Jeez, that iPhone is very close in benchmarks compared to my laptop's 1st gen i5 @ 2.53 GHz. My laptop, for example, hits about 250 in Sunspider. I truthfully don't believe an iphone is half the speed of my laptop.

 

[tempestglen]

x86 product- Mac only accounts for less than 15% revenue, I think.

In 2007, apple jumped to x86 and arm simultaneously, Apple's resurgence came mostly from arm devices(iPod,iphone,iPad etc) not from x86 chips.

Furthermore, the Britain ARM Holding co. was founded by apple and Acorn in 1990， arm chipset was for the famous first PDA "Newton" from apple. Apple is the "father" of ARM, but apple sold out ARM shares in1996 for saving apple itself.

DEC's P.A. center in 1990s designed the best arm micro architecture-- Strong ARM(later as intel's xscale) and apple cooperated with P.A.(DEC) tightly. In 2006, the P.A.'s pwrefficency chipset was chosen as Mac's heart, but 2007 apple turned to intel leaving P.A. in angry, and 2008 apple merged P.A..Steven Jobs gave P.A. a very radical design target, the fruit is Swift A6 in iphone5 and today's 64bit A7.

I believe that ARM and P.A. are completely involved in apple's history, not surprised if apple drops x86 for its customized ARM micro architecture.

Therefore, the story of ARM vs X86 will between former DEC(P.A.) and Intel, the "StrongARM "vs "Xscale"?

StrongARM MKII is coming! Watch out for it, intel!

I've told you STRONGARM MKII IS COMING.

 

[FwFred]

Jeez, that iPhone is very close in benchmarks compared to my laptop's 1st gen i5 @ 2.53 GHz. My laptop, for example, hits about 250 in Sunspider. I truthfully don't believe an iphone is half the speed of my laptop.

Have you tried IE 10 (or 11)? IE seems to do much better than Chrome here.

 

[Zodiark1593]

Have you tried IE 10 (or 11)? IE seems to do much better than Chrome here.

Firefox, though now that I think (right after I made the post), I haven't updated it in about a year, perhaps longer. Chronic procrastinator much, though I'm rather fond of the older task bar over the Chrome-style tabs.

edit: Running Waterfox 13.0

2nd edit: Scored 231 running the latest Firefox, though Mozilla Kraken went from 4000 to 2500.

 

[mavere]

I have a 2.0GHz Penryn (P7350) on latest OSX 10.9 preview with presumably up-to-date Safari/Webkit codebase: ~350 in Sunspider.

Go away smartphones. Stop making my laptop more obsolete.

Edit: I just ran Kraken (4610) and Octane (7000). Assuming Safari is equally optimized in both OSX and iOS, the A7 is very reliably ~25% slower than a 2GHz mobile Penryn. The iPad 5 will likely almost match this laptop in js-perf.

 

[Zodiark1593]

I have a 2.0GHz Penryn (P7350) on latest OSX 10.9 preview with presumably up-to-date Safari/Webkit codebase: ~350 in Sunspider.

Go away smartphones. Stop making my laptop more obsolete.

Overclocked (1.55 Ghz), the Tegra 2 in my smartphone (LG G2X) outpaces the Athlon XP 3000+ in my parents' desktop. Got about 1100 in Sunspider.

 

[Haserath]

Jeez, that iPhone is very close in benchmarks compared to my laptop's 1st gen i5 @ 2.53 GHz. My laptop, for example, hits about 250 in Sunspider. I truthfully don't believe an iphone is half the speed of my laptop.

Sunspider isn't that great of a CPU test by itself.

There hasn't been any good tests done to compare. I think theoretically it could be half speed in some tests. Both CPUs execute 128-bit internally.

It probably ends up around 1/4 in a real test.

I'm amazed they can scale speed up this quickly without thermal limitations.

 

[Zodiark1593]

Sunspider isn't that great of a CPU test by itself.

There hasn't been any good tests done to compare. I think theoretically it could be half speed in some tests. Both CPUs execute 128-bit internally.

It probably ends up around 1/4 in a real test.

I'm amazed they can scale speed up this quickly without thermal limitations.

No kidding. Suddenly, the rumor of Apple eventually going ARM for their Macbooks doesn't seem so far fetched, especially if their CPU designs can scale well for high power/high performance applications.

I would like to see some sort of raytracing or cg rendering test though. See the cpu in action in a real benchmark.

 

[StrangerGuy]

So ArmV8 is a completely separate instruction set. And also comes with some performance improvements.


http://www.anandtech.com/show/7335/the-iphone-5s-review/4

And Apple's A7 competes with Baytrail at a much lower frequency.

I bet Apple now is laughing off their asses outdoing a company whose entire existence revolves around fabs and CPUs, on an "ancient" 28nm node as a side project.

 

[Aikouka]

1GB RAM is beyond pathetic at this time.

Why? I have an iPhone 5 with 1GB of RAM, and I haven't felt like it could use more. Typically, the biggest things that iDevice users notice when their phone is running low on memory is that their previous websites (and other tabs) get chucked from memory. That's really not all that common for me on my phone -- even when I play a few games. Would it hurt to have more memory? Mmm as long as it doesn't eat up more power.

 

[krumme]

I bet Apple now is laughing off their asses outdoing a company whose entire existence revolves around fabs and CPUs, on an "ancient" 28nm node as a side project.

Well it looks my x86 hobby for 25 years will change to armv8 in a few years time. Lol. Those changes dont come often.

 

[StrangerGuy]

Well it looks my x86 hobby for 25 years will change to armv8 in a few years time. Lol. Those changes dont come often.

You can bet every ARM licensee and their dog from high to bottom tier is rushing out armv8 next year. Probably leaving BT to a still massive but shrinking <$400 craptop market or the oh-so-popular Win 8 tablets both with performance good enough to give consumers a good reason not to even bother with Haswell.

 

[Eug]

Why? I have an iPhone 5 with 1GB of RAM, and I haven't felt like it could use more. Typically, the biggest things that iDevice users notice when their phone is running low on memory is that their previous websites (and other tabs) get chucked from memory. That's really not all that common for me on my phone -- even when I play a few games. Would it hurt to have more memory? Mmm as long as it doesn't eat up more power.

A 64-bit 5S would be using more ram than the 5.

 

[Zodiark1593]

You can bet every ARM licensee and their dog from high to bottom tier is rushing out armv8 next year. Probably leaving BT to a still massive but shrinking <$400 craptop market or the oh-so-popular Win 8 tablets both with performance good enough to consumers a good reason not to even bother with Haswell.

Here's hoping Intel will dramatically increase performance of their i-series in the near future. I wonder if Arm can still see further leaps down the road, or if this was their Conroe moment.

 

[Eug]

So, are we sure now that the fab is Samsung? Anand seems to think it is.

 

[Mondozei]

You can bet every ARM licensee and their dog from high to bottom tier is rushing out armv8 next year. Probably leaving BT to a still massive but shrinking <$400 craptop market or the oh-so-popular Win 8 tablets both with performance good enough to give consumers a good reason not to even bother with Haswell.

I didn't buy into the 'Apple is doomed' crowd's arguments but Bay Trail tablets will not be horrible. Bay Trail will rival the A7 with software optimizations and do so at a much lower price point than the new iPad mini, even if that one will have the A7X. (Typically the GPU gains are the strongest in the 'X' chip variants and we know that BT will perform around the S600 level. Decent enough for most mainstream users).

Secondly, the mainstream tablet market is exploding right now, so your assertion that it will be 'shrinking' is slightly bizarre.

Most people will buy the cheapest tablet which will give them decent-to-good mainstream performance. Bay Trail will get them top-of-class CPU performance and middle-to-high GPU performance at very attractive price points.

 

[lefty2]

Have you tried IE 10 (or 11)? IE seems to do much better than Chrome here.

Those javascript benchmarks just measure browser performance between the Safari version in iOS7 and Chrome/IE rather than CPU speed. . It's a pity they didn't do any proper tests. It would have been interesting to see how A7 measures up against Bay Trail.

 

[BallaTheFeared]

Those javascript benchmarks just measure browser performance between the Safari version in iOS7 and Chrome/IE rather than CPU speed. . It's a pity they didn't do any proper tests. It would have been interesting to see how A7 measures up against Bay Trail.

Indeed... (i5-4670k OC)

Chromium 31 - 86.1ms

Fox Nightly 27.0a1 - 94.7ms

IE10 - 56.1ms

 

[Eug]

For those of you who care about this sort of stuff, here is Linpack (multithread) on the 5S:

http://reviews.cnet.com/iphone-5s/

iPhone 5S - 1634.4
HTC One - 648.7
Galaxy S4 - 636.5
iPhone 5C - 583.9
Moto X - 293.8

IOW, the GS4 scores < 40% of what the iPhone 5S does.

 

[TerryMathews]

For those of you who care about this sort of stuff, here is Linpack (multithread) on the 5S:

http://reviews.cnet.com/iphone-5s/

iPhone 5S - 1634.4
HTC One - 648.7
Galaxy S4 - 636.5
iPhone 5C - 583.9
Moto X - 293.8

IOW, the GS4 scores < 40% of what the iPhone 5S does.

Do you know, is that GS4 Snapdragon or Exynos?