AMD Hints at Possible ARM Partnership

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Status
Not open for further replies.
No, AMD never had a horse in the phone/tablet race. The first generation of fusion was meant for netbooks, lappies and desktops. and it's not as if it's not been selling...

http://www.xbitlabs.com/news/cpu/display/20110424165526_AMD_Sells_Three_Million_Fusion_Chips_in_Q1.html

Intel's atom was their attempt to shrink x86 down to a platform that was feasible for phones and tablets and they never accomplished that (maybe the new gen will appear in tablets, they have OEM partners iirc), but fusion is just what it means... CPU + GPU fused. when AMD bought ati they didn't do it because their CPU business stunk, they did it because they saw fusion tech as something they can do very well, but it's all still x86, and thus inherently leeches way too much power and incredibly costly to produce in comparison to the ARM stuff.
 
Fusion turned out to be utter crap as many reviews showed, something i said it will be a crap before it was even released. If i was AMD or Intel i would slap Athlon XP into mobile device or let's say Intel Pentium IV. With the current technology they can make those cpus so small and power draw couple W.

x86 is future for mobile market. Why re-invent the wheel when wheel is already there and plenty to choose from!!!
Having x86 cpu in mobile device you get limitless options.

ARM -> CRAP, Nvidia is wasting their time with it.
 
fusion isn't crap, which reviews are you reading? you just admitted you had a bias before the thing even launched.

it made it into a tablet already ... Acer Iconia W500. I would imagine its battery life strugles, but its a full fledged win 7 machine.
 
or... or... maybe ARM can have a tie up with AMD for graphic system. why are we not thinking the other way around.
 
ARM + Radeon seems like a natural choice for Fusion, especially if you use the shader cores to strengthen the ARM chip's floating-point operations. NVidia's Tegra can already use CUDA, but developers haven't really taken advantage of that yet. AMD could stall them by promising OpenCL! And imagine combining Bulldozer's modular approach (sharing certain functional units) with Fusion's goals! AMD's biggest obstacle might be manufacturing, since 32nm gave them so much trouble. It would be really cool to have Windows 8 ARM using AMD Fusion/Radeon, as long as we also get software that can run on it.
 
I would hope that their partnership would be more to do with manufacturing smaller chips as ARM is already releasing samples far smaller than AMD. With Intel boasting 22nm or w/e AMD really have to do something radical to catch up
 
[citation][nom]lradunovic77[/nom]Fusion turned out to be utter crap as many reviews showed, something i said it will be a crap before it was even released. If i was AMD or Intel i would slap Athlon XP into mobile device or let's say Intel Pentium IV. With the current technology they can make those cpus so small and power draw couple W.x86 is future for mobile market. Why re-invent the wheel when wheel is already there and plenty to choose from!!!Having x86 cpu in mobile device you get limitless options.ARM -> CRAP, Nvidia is wasting their time with it.[/citation]

Nearly all the reviews for fusion have been nothing but positive. In fact the first gen fusion chips have been selling like hotcakes.

AMD is much more willing to look past x86 than Intel is. Intel has been trying to forcefully ram x86 down into a low-power segment of the market where, at least for the near future, it doesn't belong. The new atoms will still have a much higher power draw and they're far more expensive than ARM chips. The issue with x86 isn't that it's not feasible -- it is, but that's years down the line. When you take into account the manufacturing costs of bringing x86 down to ARM level then you realize that it may just not be worth it.

The one thing that may change that is an incredible revolution in batteries for these mobile devices, but that's something that's holding technology back as a whole and not just x86. Furthermore, intel/amd on x86 would still have to compete with a cheaper product that draws less power.
 
[citation][nom]gamerk316[/nom]The only reason for AMD to go ARM is if they lost faith in Fusion...wasn't tat suppost to be their big thing for tablets/smartphones?[/citation]

Er...no? Fusion is just AMD's name for integrated graphics built into a CPU chip. That CPU component can well be a ARM cpu. For AMD this may be the way to go with their rather limited R&D budget compare to intel.

As for intel, I actually would not be surprised if intel somehow brute forcing the x86 chip to be as power efficient as ARM, given them enough time. After all, they got the money.
 
[citation][nom]pita[/nom]Er...no? Fusion is just AMD's name for integrated graphics built into a CPU chip. That CPU component can well be a ARM cpu. For AMD this may be the way to go with their rather limited R&D budget compare to intel. As for intel, I actually would not be surprised if intel somehow brute forcing the x86 chip to be as power efficient as ARM, given them enough time. After all, they got the money.[/citation]

Given enough time, I can see it happening. The question though isn't whether intel has enough money or time to do it, but rather do we as consumers have the patience and the wallet size to pay for it? x86 at that level will inevitably be much more expensive, and bear in mind that ARM-based advancements are coming along quite nicely. We're seeing 1ghz dual core stuff already, and by next year there's talk of quad core arm architecture hitting the market and running low profile and power servers. In that time frame Intel simply can't manage to shrink x86 down to the same level. 2-4 years maybe, but if you take into account the timescale of atom advancements and just how minimal they are with respect to what they should be accomplishing, and don't forget the price, it just seems too damn slow and costly.
 
well... here's my 2 pence...

Chip companies are going to do what chip companies choose to do. It's up to the software developers to write software to run on the products that are on the market. That's why we have competition, to give those developers a choice.

Personally, I'd love to see AMD eat up a bit more market share. To me, they seem more reliable and a bit more honest in business practice than Intel. I really think they've got a good base now, strong CPU background, now a major player in the GPU market, and if they make ties with ARM, then the mobile market may get shaken up. We'll just have to see what happens! I think the CPU market gets more interesting each day.
 
[citation][nom]pelov[/nom]Given enough time, I can see it happening. The question though isn't whether intel has enough money or time to do it, but rather do we as consumers have the patience and the wallet size to pay for it? x86 at that level will inevitably be much more expensive, and bear in mind that ARM-based advancements are coming along quite nicely. We're seeing 1ghz dual core stuff already, and by next year there's talk of quad core arm architecture hitting the market and running low profile and power servers. In that time frame Intel simply can't manage to shrink x86 down to the same level. 2-4 years maybe, but if you take into account the timescale of atom advancements and just how minimal they are with respect to what they should be accomplishing, and don't forget the price, it just seems too damn slow and costly.[/citation]

What you're saying makes no sense. x86 is a cumbersome, bloated, inefficient instruction set, and will be two or four years from now. Intel can try to get it to work, but it's always going to be a disadvantage - then or now. They can use their superior manufacturing to offset it to some extent.

But, then, ARM is an old, crappy instruction set, just not as bad. AMD has been run by a monkey named Ruiz, and now that they are out of his dark shadow have a chance to innovate. Using an obsolete instruction set that was heavily influenced by the 6502, a 1970s processor, is not the solution. They should create a new instruction set that is highly efficient, and offer it at a price that it can become a new standard. ARM doesn't matter at all, who cares if they have an ARM processor or not? Do you think people buying an iPhone really care? They care what it runs, and ARM doesn't have a huge installed base. Better to start off fresh, and have Intel try to compete with their horrible instruction set against your very efficient one, instead of your not as bad instruction set.

The answer doesn't lie in the 1970s. We don't need x86 or ARM. We need something 2011.
 
[citation][nom]TA152H[/nom]What you're saying makes no sense. x86 is a cumbersome, bloated, inefficient instruction set, and will be two or four years from now. Intel can try to get it to work, but it's always going to be a disadvantage - then or now. They can use their superior manufacturing to offset it to some extent.But, then, ARM is an old, crappy instruction set, just not as bad. AMD has been run by a monkey named Ruiz, and now that they are out of his dark shadow have a chance to innovate. Using an obsolete instruction set that was heavily influenced by the 6502, a 1970s processor, is not the solution. They should create a new instruction set that is highly efficient, and offer it at a price that it can become a new standard. ARM doesn't matter at all, who cares if they have an ARM processor or not? Do you think people buying an iPhone really care? They care what it runs, and ARM doesn't have a huge installed base. Better to start off fresh, and have Intel try to compete with their horrible instruction set against your very efficient one, instead of your not as bad instruction set. The answer doesn't lie in the 1970s. We don't need x86 or ARM. We need something 2011.[/citation]

I'm not disagreeing with you, in fact we're in nearly full agreement. I too think x86 is too bloated to work on small-scale mobile devices where power consumption is pivotal. But i too think intel can close that gap considerably, but it'd take a LOT of time and a LOT of money, both of which mean less success when considering just how cheap ARM architecture chips are to make. The new atom processors are essentially halved old atom processors, where the clock speed is reduced and as is the power consumption, yet it's still way too high.

The issues are the same with either architecture but at opposite ends. I've read interviews with Nvidia heads where they've said ARM can handle server applications that require low wattage and don't need lots of horsepower (lots of cores, low power), but replacing the desktop CPU is not even on their minds. x86 faces the issue of not being able to shrink itself to a point where it can leech off of a relatively small-sized battery for extended periods of time.

Both Intel and AMD have stated that they can potentially shrink x86 down to a point where it competes with ARM in terms of power consumption, but whether that's wishful thinking and hoping to maximize profits as the only real x86 license holders, or whether it's actually feasible in the near future is a big big question for today's leaders in that field.

Can they introduce another player in the field? Yes, but getting people to back it, particularly in the field where so many different parts are working in unison that it requires (or really favors) a steady and stagnant approach is very risky.

I do think, though, that AMD does hold a bit of an advantage if it were to decide to try something new. CPU and GPU under the same roof along with a history of making big innovations that required risky behavior (first to 64bit, first APU) is something that should breed creativity.
 
[citation][nom]DjEaZy[/nom]... and the other thing... Windows 8 will support ARM too...[/citation]

+1 Exactly! When the software no longer needs x86 whoever makes products faster and cheaper will win market share. If AMD partners with ARM they are widening their potential market share.
 
Ok people need to stop with the nonsense. You can not compare "x86" and "ARM" because both are ambiguous words used for multiple things. There hasn't been a "x86" CPU made in a very long time. All currently implementations of the x86 instruction set, and its 64-bit extension are done through front side instruction interpreters. Example is the AMD CPU, every since the original Athlon its been a RISC CPU running with an instruction decode unit that receives CISC x86 instructions and created micro-ops and dispatch's them to the internal CPU engines. Intel follows a similar design, though theirs tends to be more "CISC" like.

ARM on the other hand isn't a CPU, its a licensed architecture. Companies license the design's from ARM and can either built the reference design or customize (Samsung) it for their own use. Each CPU produced this way isn't 100% compatible with other ARM CPU's and tends to handle memory mapping differently.

For power, ARM is a very very ~low~ power design. And by low power I mean both electrical usage and processing capability. It simply isn't a powerful CPU, never was designed to be. Also between the two design's x86 is more suited for general purpose processing then ARM is. If you want a low power x86 CPU you don't use something like Atom as Intel deliberately neutered the design to prevent it from competing with the i3's for low power Kiosk / Web systems. You would instead use something like a Via Nano or an AMD Geode. Both CPU's are significantly stronger then an Atom due to them being superscaler architectures.

And before anyone makes a snide comment about Via's not being made, your just not looking in the right places. Nearly every ATM, Point of Sale or Electronic slot machine contains a Via CPU. Their CPU's can run at 1~1.3 GHZ without a fan in both blazing hot weather ~and~ freezing cold weather. They make amazing set-top systems or network attached devices.
 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom