New Multi-Core iPhone with Firmware 3.0?

[Rot'nApple]

But wouldn't nano imply it being smaller, thus the pro being larger?

Why?

Why can't "nano" imply reduced processor power as in 3g and the new iPhone Pro can be the exact mirror of the "nano" 3G iPhone but come equipped with a bigger, more powerful processor?

I mean, the MacBook Pro 15" is smaller in size and less powerful in processor power than the 17" but we don't go around calling the 15" the MacBook Pro and the 17" the MacBook Pro'er!

 

[randyhudson]

Power consumption is not square of clock-speed

I don't think CPU power consumption increases as a square of the clock speed. There is a certain minimum voltage below which the CPU will not even function, so slowing down the clock speed further when it is operating at Vmin means roughly linear power (and performance) reduction.

When you start over-clocking CPUs, and you have to increase the voltage above Vmin in order to do so, then it is true that the power consumption grows faster than the performance gains, but I don't think it is an exact science. It often depends on the imperfections in the chip, packaging, and cooling.

But, as an example, a Mobile Core 2 Duo (T9300) has a MHz range of 800*-2,500, but the voltage only increases 1.062V-->1.150V.

(*I'm guessing 800 is around the minimum clock speed, matching its FSB).

 

[darkblu]

I don't think CPU power consumption increases as a square of the clock speed. There is a certain minimum voltage below which the CPU will not even function, so slowing down the clock speed further when it is operating at Vmin means roughly linear power (and performance) reduction.

When you start over-clocking CPUs, and you have to increase the voltage above Vmin in order to do so, then it is true that the power consumption grows faster than the performance gains, but I don't think it is an exact science. It often depends on the imperfections in the chip, packaging, and cooling.

But, as an example, a Mobile Core 2 Duo (T9300) has a MHz range of 800*-2,500, but the voltage only increases 1.062V-->1.150V.

(*I'm guessing 800 is around the minimum clock speed, matching its FSB).

to continue from where randyhudson left it at, for a multi-core (MP) design to win versus a single core (SP) desing in the power/watt department, all it takes is:

* at full opetational configuration (in our case - all four cores on), the performance of the MP desing should approach the SP design in watts.

* from there on, the MP desin allows for powering down cores one by one, effectively achieving a linear power saving, which the SP core does not allow. IOW, if a mobile c2d clocks down as 1.150V@2500 --> 1.062V@800 (as per randyhudson's c2d example), the quad core desin would go 1.150V@2500* (equivalence of) --> 0.2875V@600 (equivalence of).

that'd be a clear win for the MP desin in the power-saving deparment. in practical terms, it'd allow an MP tablet to run much longer from battery (reduced core count down to 1), and yet be very close to the 'big iron' SP performance-wise when plugged in the wall socket, and power draw is not a concern.

* assuming quad arm cortex approach a mobile c2d in performance, and power draw, which is a gross simplification for the sake of the argument.

 

[holmesf]

Pretty much.

As far as phones, though, I suspect the next iPhone will allow background process too. The current model COULD do it, and jailbreakers will, but Apple won't enable it in order to encourage upgrades.

The reason for this is actually because the iPhone has only 128MB of RAM and does not support virtual memory in the sense of paging RAM contents to flash memory like a desktop pages memory to the hard disk. Since the system consumes some of this memory, that means that < 128MB is available to any given application, and since there is no paging, any application that tries to consume more than the available memory will be forced to either free up memory by the system, or will be terminated.

This is all information Apple has stated in its developer videos.

Apple thus felt that to provide the best experience for the end user (applications not causing each-other to malfunction) they should disallow background processes.

 

[Ochire]

Wicked. Maybe it'll give it enough power to copy and paste.. hell, maybe even forward messages while we're at it. I'm not expecting mms or emailing attachments till we're at octocore though.

 

[dex22]

C'mon guys. Apple's not going to release this year a complete product based around a technology they won't even get their hands on for another 6 months.

Even then, it will take 6 months for apple to design the IC, and three months more to qualify the phone as a complete, integrated system. Then they have to port the OS and qualify it. The rational timeframe for this all-singing all-dancing all-cut-n-pasting phone is June 2010.

I'm not saying they won't release a dual-core ARM11-based iPhone in the meantime...

 

[winterspan]

This is highly unlikely unless the iPhone v3 is way off down the road.


Here's the deal. The current ARM processor core in the iPhone, ARM11, does indeed have an optional multi-core architecture if system-on-a-chip manufacturers want to go that route, and nVidia's "Tegra" platform actually uses multiple ARM11 cores.

That said, ARM11 is old; The next generation of ARM is the Cortex series. The first version, the Cortex-A8 is literally just now coming out in smartphones (like Palm's new Pre I believe), and is what powers T.I.'s OMAP line and Qualcomm's Snapdragon. The Cortex-A8 is *NOT* multi-core capable.

The second-generation core in the "Cortex" line is the Cortex-A9 (which is an out-of-order core), which does indeed add multi-core support. But system-on-a-chip manufacturers are not even close to using the Cortex-A9 core. Although it's architecture has been completed, it's not even shipping to 3rd party chip designers. Additionally, the embedded market moves much slower than say the x86 desktop market. Even after the ARM architecture has been completed, it has to be extensively tested and verified, and then 3rd party chip designers have to integrate the cores into new system-on-a-chip models, which also have to be extensively tested and verified. Even once that is done, then the device manufacturer has to integrate that complete chip into a new end-user device. This whole chain from ARM core conception and design to an end user actually having the product takes years.

For this reason, I think it's *highly unlikely* that the next-gen iPhone would use a multi-core-capable Cortex-A9 if it is released in 2009. Its much more likely Apple uses a custom single-core system-on-a-chip based on the Cortex-A8.

C'mon guys. Apple's not going to release this year a complete product based around a technology they won't even get their hands on for another 6 months.
Even then, it will take 6 months for apple to design the IC, and three months more to qualify the phone as a complete, integrated system. Then they have to port the OS and qualify it. The rational timeframe for this all-singing all-dancing all-cut-n-pasting phone is June 2010.

I'm not saying they won't release a dual-core ARM11-based iPhone in the meantime...

I agree. I also don't think they would use a dual-core ARM11, as a single-core Cortex-A8 is so much more powerful it would be pointless, particularly at the Gigahertz+ clock speeds they've gotten them running at.

 

[numbersyx]

I agree that this doesn't ring true. Far too big a leap for a cell phone PDA device. No doubt that Apple have invented something which has caught the imagination but someone's imagining a little too much.

 

[branjosef]

I'm holding out for a quad-core G5 iphone 3G with blu-ray

 

[mtizle]

ok, I hate to take a trip to negative town on this topic, but I feel like I need to be realistic. For any of you that own a mobile device with a core processor (being macbook, macbook pro or a pc laptop with a core processor) you know how hot they get. My Macbook's fan run's at about 6200 RPM when at full strength, which is somewhat often. Being a cell phone, there is no way that a cooling system can be built into a phone without increasing the size of the phone or killing the battery life by a ton. Sorry guys, but this just isn't happening.

 

[holmesf]

ok, I hate to take a trip to negative town on this topic, but I feel like I need to be realistic. For any of you that own a mobile device with a core processor (being macbook, macbook pro or a pc laptop with a core processor) you know how hot they get. My Macbook's fan run's at about 6200 RPM when at full strength, which is somewhat often. Being a cell phone, there is no way that a cooling system can be built into a phone without increasing the size of the phone or killing the battery life by a ton. Sorry guys, but this just isn't happening.

It probably won't happen ... however there is no direct relation between number of CPU cores and heat produced by the CPU, especially comparing a theoretical ARM CPU with an Intel one.

 

[iMacmatician]

What about the possibility to run Mac OS X instead of just OS X on the iPhone and iPod touch? For the ultimate wired and wireless presentation remote using NATIVE Apple Keynote and Microsoft PowerPoint files.

Mac OS X is too hardware-intensive for an iPhone, iPod touch, or even a mini-tablet. And I think we will see mobile iWork on the mini-tablet.

What? Why? So you can have a cursor-arrow and drop-down menus on a tiny screen? How annoying!

Wouldn't it make MUCH more sense to simply write software for the iPhone that can use these files you're talking about?

Of course not, because, you know, why not just fill up the 3.5" display with all the menus, toolbars, and Inspector windows that come with Keynote/PowerPoint? Don't forget that Keynote files require a cursor. Not a touchscreen, a cursor-based touchscreen.

Seriously though, if the iPhone OS was modified to support tabs, slide-out sheets (instead of Inspector window), and probably a few other things, mobile iWork is very possible, even if it was limited to a device slightly larger. I don't think some advanced effects and stuff will be in the mobile version, due to processing / app size limitations (maybe just a way to input it in the file but not to actually process/view it), but that happens on older, lower-end Macs too.

I don't think CPU power consumption increases as a square of the clock speed. There is a certain minimum voltage below which the CPU will not even function, so slowing down the clock speed further when it is operating at Vmin means roughly linear power (and performance) reduction.

I think Power = (something)(clock speed)(voltage)^2.

 

[Jig]

It doesn't work out all neat and pretty like that. Multithread scaling is not perfectly linear by any means. If that was true, why stop at four! why not have 16 ARM11s in that bitch all running at 1/16 of the clock speed...0mgZ0rs!

I would much rather have more battery life than a moronic quad-core crippled iPhone / iPod Touch. The battery tech isn't there yet (and the new 17" MBP's battery is nothing special. It's the same lithium-polymer technology just in a different form factor and larger). I'd much rather have push or regulated background process support than a dual or quad cpu in my Touch.

LOL. I think not. I need quad-core to do copy & paste? Simply moronic.

The original poster about the power consumption growing quadratically to the clock rate was entirely right, despite the ridicule he received. Many cores in embedded systems will save power. That's the point.

Firstly, the iPhone has lots of stuff running, as anyone can see from jailbreaking, and I think quad cores would certianly help performance.

But more importantly, it would help battery life. Twice the clockrate = 4x the power. Twice the cores, 2x the power. Although it is true programs need to be written in a way to take advantage of this parallelism, this really can be done for a lot of things, like for instance mp3 decoding, h264 decoding etc. In fact it already has been done.

For an embedded product, where performance and battery life are important, there is no reason at all why this couldn't work. In fact, it's the only way forward.

You ask 'why not 8 cores, or 16 cores?' Why not indeed! This will happen, and will happen, and will reduce power consumption further. There already exist processors that have this kind of parallelism, and it will come to the embedded world. Like everything, it takes time.

 

[iMacmatician]

The original poster about the power consumption growing quadratically to the clock rate was entirely right, despite the ridicule he received.

It's linear.

http://software.intel.com/en-us/art...the-performance-details-of-mobile-processors/
http://www.microcontroller.com/Embedded.asp?did=155

 

[pedmed]

Power mac G3 Beige

Hello areyouwishing,
I was wondering if you still have your Powermac G3 333Mhz for sale?
If you still have it for sale, please contact me at: pedmed@ispmonsters.com
Pedmed

Wow, surprised I didn't see anyone comment about the Pre by Palm.

Calendar in iPhone still can't use multiple calendar sources (exchange, gmail, facebook) and it certainly can't show the concurrently.

I cannot go from 3rd party app directly to 3rd party app without making a pitstop at the home screen.

I still don't have spotlight/inquisitor for the entire phone (gotta go to safari to search the web, gotta go to contacts to search for them, and forget searching email)

I still don't have turn by turn directions.

Still no flash on the camera

still no option of a physical keyboard

I still can't read a consolidated inbox that shows ALL my inbound email (from multiple accounts)

Still can't have 2 drafts open at the same time

still don't have native unified messaging (IM, SMS, MMS) in a single thread

BUT I am getting 4 cores...AWESOME DUDE!

 

[pedmed]

power mac G3 333 MHz

Hi areyouwishing,
I was wondering if you still have your Beige powermac G3 with 333Mhz for sale? if so, I am interested in purchasing it.
Please email me at: pedmed@ispmonsters.com
Thanks
pedmed