Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Intrinsity (Formerly Exponential Technology) Now Speeding Up ARM CPUs

leukotriene said:
Ok, A4...A8...A9.

Can somebody boil this down in laymen terms? What does the "A" refer to? ARM? And the number after it means what? Is a higher number better?

I know there are some chip-savants on this thread. Help us. :D
Click to expand...
Since Jobs described the iPad's CPU as Apple's "own design", the simplest explanation is that the "A" refers to Apple, not ARM. If you look at the screen grab from the promotional video, the Apple logo is featured prominently next to the "A4" on the CPU.

As to the number, it is currently meaningless to the layperson or anybody else except Apple Engineering. Since there is no other publicly disclosed Apple CPU with the ARM architecture, there is no reference point. The number will only begin to have meaning after we've seen one or more other CPUs from Apple.
 
cvaldes said:
Since Jobs described the iPad's CPU as Apple's "own design", the simplest explanation is that the "A" refers to Apple, not ARM. If you look at the screen grab from the promotional video, the Apple logo is featured prominently next to the "A4" on the CPU.
Click to expand...

Wouldn't that actually suggest the Apple icon itself refers to Apple and not the 'A'?

Anyways, according to an article referenced by engadget, the Apple A4 is a system-on-a-chip that uses the ARM Cortex-A9 CPU.
 
Gray.2 said:
I think what we are seeing here is not necessarily Apple getting into the chip-making business, but rather taking more control over what they want the chips to actually do.

By all accounts, the chips are being manufactured in Asia, like all or most of Apple's other hardware.

Steve Jobs is an absolute control freak (and for good reason) and the addition of the A4 is simply an extension of that.

If Intel had made a chip that was as good as or cheaper than the A4, then I would be willing to bet, Apple would have simply used that. Because Intel (or whoever) did not have what Apple wanted, they did it independently.

Also, what happened to A1-A3?
Click to expand...

Not good to have all eggs(apples) in one basket. ;)

It a logical business practice, if you ask me. What happens if intel cannot deliver or is running into some problem down the road. :apple: learnt its lesson with the PPC "G" series, and plans to be as flexible as possible. Nothing is wrong with that, I bet there is a PPC version of Snow Leopard in they labs as we speak. :)

The ARM processors have been used in mobile computing and devices, and if :apple: is heading in that direction count on more iPhone OS X to utilize the cost efficient processor in the future. Maybe even a MacBook Air running some near full version of iPhone OS X/Mac OS X on an ARM. :D
 
macmacey said:
I agree it wasn't the design of the chip that was the problem. There just wasn't enough of a product line and roadmap to convince Apple to build a product line around it.
Click to expand...

Well, we weren't supposed to be generating a full line of chips for them. They just wanted a high-end chip. The x704 worked, and ran at around 450-475MHz (target was 533) on the first tapeout. We taped out an update that ran at least 533 but it was pretty much over by then. Our chips required a slight bios change to fire up the caches - even after Apple cancelled their orders, the clonemakers wanted to buy the chip, but they couldn't get permission to modify the bios supplied to them by Apple. Of course Apple didn't want to have to compete against clone makers using a better chip.

And Apple probably never intended to use our chip - I think they just wanted more leverage in negotiations with IBM and Motorola.

Links:
A somewhat accurate account by Fast Company: http://www.cs.utexas.edu/users/karu/papers/exponential.html

The JSSC article describing the chip (written by me!):
http://ieeexplore.ieee.org/Xplore/l...0641683.pdf?arnumber=641683&authDecision=-203
 
cvaldes said:
Since Jobs described the iPad's CPU as Apple's "own design", the simplest explanation is that the "A" refers to Apple, not ARM. If you look at the screen grab from the promotional video, the Apple logo is featured prominently next to the "A4" on the CPU.

As to the number, it is currently meaningless to the layperson or anybody else except Apple Engineering. Since there is no other publicly disclosed Apple CPU with the ARM architecture, there is no reference point. The number will only begin to have meaning after we've seen one or more other CPUs from Apple.
Click to expand...

Maybe its a reference to the G4 chip which made the last "just works" line of Macs. :p
 
My 2 cents says: Lets not get carried away with 'CHIP HYPE'. Cause most of yall don't even know what a chip really does other than sporting a certain amount of GHZ.
 
I realize it's a Sunday (lack of anything noteworthy to put on page one), but I for one appreciate the (potential) historical significance of the new A4 chip. Thanks, Arn. :)
 
arn said:
The article also has some fundamental errors, making it entirely suspect. While the A4 may be the Cortex A9, nothing about this article made me think it is more likely or not.

arn
Click to expand...

Since the presently-discussed Macrumors article also has some fundamental errors, take your pick of theories.
 
leukotriene said:
Ok, A4...A8...A9.

Can somebody boil this down in laymen terms? What does the "A" refer to? ARM? And the number after it means what? Is a higher number better?

I know there are some chip-savants on this thread. Help us. :D
Click to expand...

cvaldes said:
Since Jobs described the iPad's CPU as Apple's "own design", the simplest explanation is that the "A" refers to Apple, not ARM. If you look at the screen grab from the promotional video, the Apple logo is featured prominently next to the "A4" on the CPU.

As to the number, it is currently meaningless to the layperson or anybody else except Apple Engineering. Since there is no other publicly disclosed Apple CPU with the ARM architecture, there is no reference point. The number will only begin to have meaning after we've seen one or more other CPUs from Apple.
Click to expand...

Most chip manufacturers compose their part numbers starting with one or more letters that identify the manufacturer and chip class, then letters and/or numbers identifying the chip family, followed by one or more digits corresponding to the specific chip. After that you may find a letter (A, B, C...) if the chip design has gone through major revisions. Then a few letters and numbers specify packaging (the chip carrier) and finally the external packaging method (tape&reel, tray, tube..).

The markings on the chip are usually a small subset of the part number, along with a manufacturing date code and the company logo, when space permits.

The A on Apple's CPU is clearly standing for Apple, but does not prevent them from starting a new series for a completely different class of chip, like G for a GPU or P for support chips.

You can tell the manufacturer both by the logo and by the first characters of the part number.

It is highly unlikely the number 4 is a sequential revision. It is more likely identifying the family, maybe for the number of cores, so Apple could possibly come up later with the A2 or A1 in the next iPhone.

Future chips of the A4 family could be marked A41, A42, and the current one could be revised to A4A or A40A, A40B

There is no relation between A4 and the ARM families.
 
HLdan said:
I'm just happy that Apple is getting on the chip bandwagon so they don't have to rely on 3rd party companies like Intel. I love Intel's chips but the more Apple is not reliant on the them the better. The iPad's A4 chip is a great start.
Click to expand...

That would be stupid, the reason I jumped over to the Apple desktops is due to the intel cpu.
 
carmenodie said:
My 2 cents says: Lets not get carried away with 'CHIP HYPE'. Cause most of yall don't even know what a chip really does other than sporting a certain amount of GHZ.
Click to expand...

That's sounds like a good example of your knowledge of chip architecture, but not everyone here.
 
cmaier said:
Well, we weren't supposed to be generating a full line of chips for them. They just wanted a high-end chip. The x704 worked, and ran at around 450-475MHz (target was 533) on the first tapeout.
Click to expand...

I think the problem was the x704 was TOO fast. This was when MHz still mattered and it was a marketing problem to move too fast to 500MHz and then not be able to follow it up with something faster soon enough.

cmaier said:
And Apple probably never intended to use our chip - I think they just wanted more leverage in negotiations with IBM and Motorola.
Click to expand...

Since the chip met close to the target, your hypothesis seems correct. Exponential delivered pretty much what they promised, and so it does leave you wondering why Apple asked for it. Perhaps it seemed like a good idea at one time, and then didn't later.
 
ptysell said:
Ug. I can only pray to god that Apple doesn't dump Intel and move back to the RISC architecture.
Click to expand...

That's such an uninformed statement, I can only pray to God that you switch to MS Windowns to raise the average IQ of Mac users by 5 points.
 
macmacey said:
I think the problem was the x704 was TOO fast. This was when MHz still mattered and it was a marketing problem to move too fast to 500MHz and then not be able to follow it up with something faster soon enough.



Since the chip met close to the target, your hypothesis seems correct. Exponential delivered pretty much what they promised, and so it does leave you wondering why Apple asked for it. Perhaps it seemed like a good idea at one time, and then didn't later.
Click to expand...

A lot of it probably had to do with a change of strategy caused by the return of Steve Jobs.

Those were interesting times. I saw macs running NT (with our chip), heard from Apple engineers that none of them knew how their custom ASICs worked anymore (due to mass defections), etc. We used to get resumes from people with no chip design background, desperate to work for us just so they could be involved.

We did have two followups planned. The x704 was designed to be extendable/convertible into an x86 if we decided to go that way, which is really what the austin team was working on (though they were going to start from scratch), and the california team was working on a faster ppc (2x speed) with all the new (at the time) ppc extensions. We didn't figure out how we were going to do altivec, though. Last thing I worked on was adding support for the floating point square root instruction for the follow-on chip.

At the end we ran out of financing, and a handful of us stayed on for a couple months without pay just in case we managed to find some. It was sad.

Several of us ended up at Sun, and several at AMD. At AMD the Exponential folks had a lot to do with Opteron/Athlon 64. The rest pretty much scattered, though one of the architects has worked for me as an expert witness recently (I'm a lawyer now).
 
Actually yes I think they would.

elhungarian said:
:confused:

Think Steve will do business with a company that sued them?
Click to expand...

Certainly! There are many reasons to bring suit against a company. Remember it has been rumored that PA Semi was purchased by Apple in part due to Apple dropping a big PPC project with PA. That put PA in a tuff situation with respect to money. This resulted the possibility that later iPhone chip projects would have failed. Thus Apples rapid purchase of PA to make sure chip development continued.

This is all second hand so if anybody has more details please pipe in.

In any event sometimes in business you find a deal going south. It happens. Good leaders though should recognize how they factor into the break up and make amends accordingly. It would be interesting to see how the legal action was settled after the break up. Who knows maybe Apple invested in Intrinsity.



Dave
 
wizard said:
Certainly! There are many reasons to bring suit against a company. Remember it has been rumored that PA Semi was purchased by Apple in part due to Apple dropping a big PPC project with PA. That put PA in a tuff situation with respect to money. This resulted the possibility that later iPhone chip projects would have failed. Thus Apples rapid purchase of PA to make sure chip development continued.

This is all second hand so if anybody has more details please pipe in.

In any event sometimes in business you find a deal going south. It happens. Good leaders though should recognize how they factor into the break up and make amends accordingly. It would be interesting to see how the legal action was settled after the break up. Who knows maybe Apple invested in Intrinsity.



Dave
Click to expand...

1) Apple was not working on a PPC project. PA Semi was, but Apple was just an investor - they were also an investor in Exponential.
2) Neither Intrinisity, nor its actual former name, EVSX, sued Apple. Exponential did not become Intrinsity. The very basis for the question was wrong.
 
!¡ V ¡! said:
Not good to have all eggs(apples) in one basket. ;)

It a logical business practice, if you ask me. What happens if intel cannot deliver or is running into some problem down the road. :apple: learnt its lesson with the PPC "G" series, and plans to be as flexible as possible. Nothing is wrong with that, I bet there is a PPC version of Snow Leopard in they labs as we speak. :)

The ARM processors have been used in mobile computing and devices, and if :apple: is heading in that direction count on more iPhone OS X to utilize the cost efficient processor in the future. Maybe even a MacBook Air running some near full version of iPhone OS X/Mac OS X on an ARM. :D
Click to expand...

It's not an issue of Intel vs ARM for desktop computing, it's one of one of chasing the power-sipping specs for mobile computing. Right now mobile computing is broken into two groups: 1) Intel chips that are optimized for mobile applications. These chips run MS Windows and Apple OSX, and are aimed at fulfilling the applications needs of enterprise, government, and legacy users. These chips may never be as power efficient as the ARM chips.

2) ARM chips that due to their design, are the best path to devices that have much longer battery life between recharges. These are most suitable for smart phones and other highly portable devices. (Low battery consumption means lighter weight, cooler operating, and smaller devices as a consequence)

All ARM chips lacked was a good application base to become broadly successful to a large user base. The iPhone/iTunes mashup created an astounding 140,000 application base in an amazing short timebase. (its next competitor's application offering is 100 times smaller)

I don't envision the current Intel architecture going away as the mobile computer segment grows. There are many backroom applications that it should continue to well serve.

What I do see is the needs of mobile users being better served by ARM style hardware. And I see that arena broken into two segments:

1) Closed, mostly vertical systems, such as Apple's Mobile OSX.

2) Open systems, such as the Android OS

While history shows that the market favors open systems 10:1, history need not repeat itself this time around. Short term agility and long-range strategic planning have both become more important in the last 10 years.

Additionally, companies that can keep their core products from becoming commodities can command better margins and develop a more loyal customer base. These kinds of companies can also effectively lead the market, which also gives them the first-mover advantage at all times.

To recap: It's not ARM vs Intel competition for desktop computers. It's a matter of choosing what works best in the super-portable computer arena of intelligent phones and touch-based computers. ARM brings the right specs to this market segment. Apple brings the OS and applications.
 
cmaier said:
1) Apple was not working on a PPC project. PA Semi was, but Apple was just an investor - they were also an investor in Exponential.
2) Neither Intrinisity, nor its actual former name, EVSX, sued Apple. Exponential did not become Intrinsity. The very basis for the question was wrong.
Click to expand...

PA Semi also had a good head start on designing low power chips at the time Apple bought them. Apparently they were doing this work for the federal government under several contracts.

Apple was so insistent about getting the PA Semi engineers working on its goals that Apple,at first, even wanted to get out of the government contracts.
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back