PDA

View Full Version : ARM Cortex A9 MPCore



Greenwidth
September 17th, 2009, 11:00 AM
This might shake things up a bit if the claim that "the processor offers five times the power while only drawing comparable amounts of energy" as Intel's Atom, are true.

Works with Linux, but at the moment not with Windows:

http://www.pcpro.co.uk/news/351619/arm-launches-attack-on-intels-netbook-stranglehold

hessiess
September 17th, 2009, 11:21 AM
Of course an ARM chip will be more powerful and use less power than an eqiv X86 chip. ARM was designed to be low power/high performance. x86 on the other hand is an antique bloated mess of an architecture.

If computer architectures are going to improve, x86 needs to go away.

3rdalbum
September 17th, 2009, 01:50 PM
Works with Linux, but at the moment not with Windows

No, not with Windows. The point of ARM is to get away from x86, isn't it?

sim-value
September 17th, 2009, 01:52 PM
No, not with Windows. The point of ARM is to get away from x86, isn't it?

About that ....

And i actually am Really tired from this stupid X86 Crap ....

cb951303
September 17th, 2009, 01:53 PM
No, not with Windows. The point of ARM is to get away from x86, isn't it?

you're forgetting windows mobile.

hessiess
September 18th, 2009, 09:04 AM
you're forgetting windows mobile.

Mobile/CE is a completely different operating system, none of the useural `desktop' applications would work without porting and recompiling. Apple did themselves a favour when they switched to a UNIX style OS (OSX) breaking backwards compatibility and flushing out all of the old junk. Windows is carrying around a lot of legacy garbage from its single user OS days which is seriously messing up the architecture of the OS. A lot of configurations are global which should be user specific and a substantial number of programs will not run without administrator assess.

Ditching the current `Desktop' versions of Windows would be a breath of fresh air for the computing industry.

gnomeuser
September 18th, 2009, 10:47 AM
This might shake things up a bit if the claim that "the processor offers five times the power while only drawing comparable amounts of energy" as Intel's Atom, are true.

Works with Linux, but at the moment not with Windows:

http://www.pcpro.co.uk/news/351619/arm-launches-attack-on-intels-netbook-stranglehold

Actually ARM was recently in meetings with Microsoft on the subject of porting Windows to the ARM platform. Of course just having Windows doesn't make all the applications and libraries that aren't portable run overnight but it is a signal to Intel that their precious Wintel castle isn't safe forever.

ARM also joined the Linux Foundation this week, I suspect they will be pouring effort into making Moblin run on ARM so Netbook vendors can have that as a choice alongside the upcoming Google ChromeOS and naturally Ubuntu (and now openSUSE also has an ARM port thanks to GSoC).

ARM are really making an important effort, they are serious about getting a bite of the lowend spec computer market. It's a perfect time for them, people are looking to save money and save on energy. ARM presents a cheap high powered alternative for use in nettops and netbooks into a market that is increasingly going in that direction. The power is far sufficient for most users to do their work.

I think ARM has a bright future in personal computing and all the cash they have made from their existing market is going to do them a world of good in meeting their goals.

3rdalbum
September 18th, 2009, 11:19 AM
you're forgetting windows mobile.

No, I just forgot to mention Windows Mobile :-D

Swagman
September 18th, 2009, 11:33 AM
I wanna know what the delay is getting ARM netbooks out the door.

I suspect Redmond has already stuck its oar in to delay release until a win port is available.

argor
September 18th, 2009, 12:49 PM
I wanna know what the delay is getting ARM netbooks out the door.

I suspect Redmond has already stuck its oar in to delay release until a win port is available.

the pandora will come out in about month or 2

hessiess
September 18th, 2009, 12:54 PM
the pandora will come out in about month or 2

They said that last year;)

argor
September 18th, 2009, 06:06 PM
They said that last year;)
jep :lolflag: but now production is at full run they are just waiting for case moulds to finis production

Dr. C
September 18th, 2009, 09:07 PM
Mobile/CE is a completely different operating system, none of the useural `desktop' applications would work without porting and recompiling. Apple did themselves a favour when they switched to a UNIX style OS (OSX) breaking backwards compatibility and flushing out all of the old junk. Windows is carrying around a lot of legacy garbage from its single user OS days which is seriously messing up the architecture of the OS. A lot of configurations are global which should be user specific and a substantial number of programs will not run without administrator assess.

Ditching the current `Desktop' versions of Windows would be a breath of fresh air for the computing industry.

Give up binary backwards compatibility with x86 Windows applications and drivers and one gives up the main if not the only advantage of Microsoft Windows. Compatibility with the vast number of propriety Windows applications and driver's. So Microsoft is dammed if they do and dammed of they don't when it comes to porting Windows 7 to ARM. GNU / Linux can just port and migrate with the exception of the odd propriety bits. Ubuntu already runs on ARM.

gnomeuser
September 18th, 2009, 09:39 PM
jep :lolflag: but now production is at full run they are just waiting for case moulds to finis production

[citation needed]

Firestem4
September 18th, 2009, 10:08 PM
I'm really excited to see ARM edging into mainstream netbook/notebook/desktop market.It will be nice to actually use some technology that is not antiquated and not on the last limbs of innovation.

Only so much more can be done to the X86 architecture till we start seeing diminishing returns (which we already are, but more at a marketing standpoint). The biggest jump in X86 processor technology has always been the ability to fit more on less. 4 cores in a 45nm die!. AWESOME!. But how much smaller can we make them? 45nm is already incredibly small. They can't fit much more without having to redesign the architecture. (Which changes everything about the form factor as well).

moster
September 18th, 2009, 10:27 PM
I think windows mobile is not real contender bacause Microsuc** would be pushing it already big time. But they do not.

hobo14
September 19th, 2009, 09:38 AM
I wanna know what the delay is getting ARM netbooks out the door.

...

I'm wondering exactly the same thing.

I can't wait to get my hands on a decent ARM netbook.

You have to assume that they'd at least LIKE to have them out in time for christmas.

hessiess
September 19th, 2009, 09:44 AM
I'm really excited to see ARM edging into mainstream netbook/notebook/desktop market.It will be nice to actually use some technology that is not antiquated and not on the last limbs of innovation.

Only so much more can be done to the X86 architecture till we start seeing diminishing returns (which we already are, but more at a marketing standpoint). The biggest jump in X86 processor technology has always been the ability to fit more on less. 4 cores in a 45nm die!. AWESOME!. But how much smaller can we make them? 45nm is already incredibly small. They can't fit much more without having to redesign the architecture. (Which changes everything about the form factor as well).

Its not 4 cores on a 45 nanometre die, its 4 cores on a die which is around 10mm square(variable) which is made from transistors which are 45nm in size. A RISC architecture like ARM should be smaller in its silicone requirements simply because theare is less stuff to implement, smaller = more cores in the same space on silicone.

JDShu
September 21st, 2009, 03:04 AM
I just read up on this and I'm a little bit confused...

So this new ARM processor will use as little power as an Intel Atom, but has a much higher performance, or am I missing something? Isn't that kind of too good to be true? Whats the basic theory behind this?

hobo14
September 21st, 2009, 03:58 AM
I think it's more accurate to say it outperforms the single core Atom (but not the double core Atom?) and takes less power doing it.

ARM has excellent performance / power consumption ratio, that's pretty much been their focus up to now.

EDIT: A temporary answer to your questions, before someone who can explain it better does so:

ARM is RISC, and has always been designed to take power conusmption into account.
x86 is massively bloated CISC that has to support tons of backward compatibility, and Intel didn't even think about power consumption until after the P4.

CISC is a hardware nightmare (although it's not bad from a purely software point of view). x86 processors now have a RISC layer "below" their CISC, and hardware conversion from the CISC we all know and hate, to their RISC.

Someone who knows more about it than I do will be able to give you a much better answer ;)

Firestem4
September 21st, 2009, 04:09 AM
I just read up on this and I'm a little bit confused...

So this new ARM processor will use as little power as an Intel Atom, but has a much higher performance, or am I missing something? Isn't that kind of too good to be true? Whats the basic theory behind this?

The reason behind this is in the fundamental difference between the CPU's. Intel and AMD processors are called CISC processors. CISC stands for Complex Instruction Set Computer. RISC (which all ARM processors are) stands for Reduced Instruction Set Computer.

The key differences between the two are the instructions. CISC machines are able to perform more instructions every cycle. This allows the processor to accomplish more, but it also uses much more power to perform each individual instruction. RISC processors on the other hand only execute a small (reduced) amount of instructions per cycle. This means that it uses less energy because it is "doing' less every cycle. However because of the reduced instructions they are able to make instructions execute very quickly, thereby bridging the gap performance wise.

There are numerous more differences but those are the two biggest that show the fundamental differences between the two architectures. (I may not be completely accurate, so if I am wrong somewhere someone please correct it for me.)

Firestem4
September 21st, 2009, 04:11 AM
Its not 4 cores on a 45 nanometre die, its 4 cores on a die which is around 10mm square(variable) which is made from transistors which are 45nm in size. A RISC architecture like ARM should be smaller in its silicone requirements simply because theare is less stuff to implement, smaller = more cores in the same space on silicone.

I see..thanks for explaining it =)

hobo14
September 21st, 2009, 04:28 AM
The reason behind this is in the fundamental difference between the CPU's. Intel and AMD processors are called CISC processors. CISC stands for Complex Instruction Set Computer. RISC (which all ARM processors are) stands for Reduced Instruction Set Computer.

The key differences between the two are the instructions. CISC machines are able to perform more instructions every cycle. This allows the processor to accomplish more, but it also uses much more power to perform each individual instruction. RISC processors on the other hand only execute a small (reduced) amount of instructions per cycle. This means that it uses less energy because it is "doing' less every cycle. However because of the reduced instructions they are able to make instructions execute very quickly, thereby bridging the gap performance wise.

There are numerous more differences but those are the two biggest that show the fundamental differences between the two architectures. (I may not be completely accurate, so if I am wrong somewhere someone please correct it for me.)

Hmm, this is not right.

Cycle time is set to suit the instruction set (or more accurately, to suit the longest pipeline stage for the instruction set)
so it would be more enlightening to talk in terms of instructions/time rather than instructions/cycles.
All single core processors (no matter CISC or RISC) finish processing (at most) exactly one instruction per cycle.


_____

Most CISC instructions are larger and more complex than RISC instructions.
CISC needs fewer instructions to achieve the same as RISC, but it's instructions take more time.

That may sound like it's six one way, half a dozen the other,

but....

not all CISC instructions are much more complex than RISC, and the complex hardware needed for CISC means it's simple instructions take longer too, without really doing the extra work that it's complex instructions achieve.

BuffaloX
September 21st, 2009, 10:52 AM
The ARM processor was actually introduced in the mid 80īs as a high performance CPU, beating both Motorola and Intel offerings at the time.

Unfortunately it was unsuccessful as a PC platform, because only one vendor backed it.

Fortunately it had some success for embedded systems, and later especially in the mobile phone market, with sales now exceeding 1 billion units per year.

If this CPU makes it as the base for a real computer platform, who knows what the ARM team can do to up performance?

If the smartbooks really takes off, with good Linux reception, we may see much more powerful options in the future.

They could very likely make a 16 core 4Ghz CPU in no time, and outperform Intel at a fraction of the price.
But this would require people to be ready to ditch their old software, which is probably the great barrier they are facing.

3rdalbum
September 21st, 2009, 11:24 AM
Its not 4 cores on a 45 nanometre die, its 4 cores on a die which is around 10mm square(variable) which is made from transistors which are 45nm in size. A RISC architecture like ARM should be smaller in its silicone requirements simply because theare is less stuff to implement, smaller = more cores in the same space on silicone.

I thought 45nm was a reference to the smallest feature size, not the size of the transistors?