ARM announces 2GHZ -Dual Core Processor

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

cbn

Lifer
Mar 27, 2009
12,968
221
106
I want to see Windows 7 Phone on a nice 8-10 inch tablet. Hell, I would be happy with a normal Windows 7 tablet. Waiting on you HP!!!

Maybe MS can win with really good open hardware support for very aggressive ramping of ARM designs in larger form factors? Quad core ARM at high frequencies with support for Turbo multiplier, hyperthreading etc

Could such a strategy also be beneficial for Windows 7 Smartphones? (ie, A program that was originally written for Windows 7 ARM based notebook could become a smart phone program as die shrinks permit)

But if they did this wouldn't they have to change the way they collect fees in order to avoid competition with their own x86 Operating systems.
 

Nemesis 1

Lifer
Dec 30, 2006
11,366
2
0
I don't think many here understand how big this really is. They are stuck in the x86 world with little experience with other platforms so they equate Ghz performance with intel or AMD. They don't understand that the two platforms do things in really different ways so it is hard to compare the two. A notebook based on this cpu and outpacing Atom is very possible.





The MALI system for video is very capable. Things like this are the reason I like ARM so much. ARM is like lego blocks for cpu design. You can assemble the cpu in the way that benefits you the most rather than having to adapt your product for what the cpu manufacturers like intel or amd produce.

http://www.arm.com/products/mu...ics/mali_hardware.html



Are you sure about that. I Don't Care seems to have a good grasp of it . Its you who doesn't seem to have a grasp .
What are you puting this chip up against Atom x86 in its present state ?

When this chip is real , something you can buy. What Intel Atom chip will it be going against

Bob
 

Idontcare

Elite Member
Oct 10, 1999
21,110
64
91
About 2 weeks ago, ARM signed a deal with TSMC to develop Physical IP on their 28nm and 20nm process.

How much could this speed up ARM's time to market for new designs?

ARM pretty much does this with every foundry. Glofo, UMC, IBM. They have to as a matter of supporting the product roadmaps of their customers.

To be sure it speeds up the timeline to verify new designs, but since the speedup was already baked into the timeline and roadmap for 28nm and 20nm products it is not so much a "moves them ahead" milestone but rather is more of a "would have been a problem if this didn't happen" type of project milestone.
 

Phynaz

Lifer
Mar 13, 2006
10,140
819
126
About 2 weeks ago, ARM signed a deal with TSMC to develop Physical IP on their 28nm and 20nm process.

How much could this speed up ARM's time to market for new designs?

You're fucking kidding me right? How hard did you have to look to revive a year old thread?
 

cbn

Lifer
Mar 27, 2009
12,968
221
106
You're fucking kidding me right? How hard did you have to look to revive a year old thread?

http://www.bit-tech.net/news/hardware/2010/07/21/arm-and-tsmc-buddy-up/1

ARM currently holds a near-monopoly in the smartphone sector: almost every portable device uses some variant of its processor technology due to unrivaled performance-per-watt characteristics which keep power usage and heat output to a bare minimum. One area in which the company has traditionally lacked, however, is manufacturing: positioning itself as a pure-IP company, ARM licenses its designs to external manufacturers - unlike rivals Intel and AMD, which by and large produce their own processors.

While this allows ARM more freedom to innovate with its designs, it can increase costs and time to market. This latest partnership with TSMC will see ARM being given unprecedented access to TSMC's not-inconsiderable manufacturing capabilities - something which could give ARM the capacity it needs to take Intel and AMD on in the netbook market and win.

I just read this article yesterday. It is a little bit too old to post as News.

And as far as resurrecting this thread goes why do you have a problem with that? Isn't that what the Anandtech search engine is for? (ie find old threads first before starting new ones)
 

tommo123

Platinum Member
Sep 25, 2005
2,617
48
91
yep :)

makes for easier reading, keeping everything linear

now, a quesition,

i have a 1ghz snapdragon in my phone, and apparently by years end (maybe early next year), we'll have 1.2-1.5 dualies in HTC phones. aren't we going to hit a major battery life problem?

or are these dual core chips smaller than the current crop?
 

cbn

Lifer
Mar 27, 2009
12,968
221
106
ARM pretty much does this with every foundry. Glofo, UMC, IBM. They have to as a matter of supporting the product roadmaps of their customers.

To be sure it speeds up the timeline to verify new designs, but since the speedup was already baked into the timeline and roadmap for 28nm and 20nm products it is not so much a "moves them ahead" milestone but rather is more of a "would have been a problem if this didn't happen" type of project milestone.

http://en.wikipedia.org/wiki/ARM_architecture

ARM Ltd does not manufacture and sell CPU devices based on its own designs, but rather, licenses the processor architecture to interested parties. ARM offers a variety of licensing terms, varying in cost and deliverables. To all licensees, ARM provides an integratable hardware description of the ARM core, as well as complete software development toolset (compiler, debugger, SDK), and the right to sell manufactured silicon containing the ARM CPU. Fabless licensees, who wish to integrate an ARM core into their own chip design, are usually only interested in acquiring a ready-to-manufacture verified IP core. For these customers, ARM delivers a gate netlist description of the chosen ARM core, along with an abstracted simulation model and test programs to aid design integration and verification. More ambitious customers, including integrated device manufacturers (IDM) and foundry operators, choose to acquire the processor IP in synthesizable RTL (Verilog) form. With the synthesizable RTL, the customer has the ability to perform architectural level optimizations and extensions. This allows the designer to achieve exotic design goals not otherwise possible with an unmodified netlist (high clock speed, very low power consumption, instruction set extensions, etc.). While ARM does not grant the licensee the right to resell the ARM architecture itself, licensees may freely sell manufactured product (chip devices, evaluation boards, complete systems, etc.). Merchant foundries can be a special case; not only are they allowed to sell finished silicon containing ARM cores, they generally hold the right to remanufacture ARM cores for other customers.

Those architectural level optimizations and extensions ( I have bolded in the body of the quote) would be patentable by either the semi-conductor designer and/or fab right?
 

Idontcare

Elite Member
Oct 10, 1999
21,110
64
91
Those architectural level optimizations and extensions ( I have bolded in the body of the quote) would be patentable by either the semi-conductor designer and/or fab right?

Yes, absolutely. TI does this for example in their OMAP processors. That is how you can have so much diversity in the ARM world across so many suppliers and yet the suppliers can maintain the kinds of margins needed to justify their efforts to make such optimizations.

This is also the exact reason why Intel's effort to get Atom qualified at TSMC has utterly failed to gain traction so far. Intel insists on retaining the follow-on patents created by potential customers.

Intel is smart for wanting this but foolish for thinking any would-be customer would actually agree to their terms.
 

cbn

Lifer
Mar 27, 2009
12,968
221
106
yep :)

makes for easier reading, keeping everything linear

now, a quesition,

i have a 1ghz snapdragon in my phone, and apparently by years end (maybe early next year), we'll have 1.2-1.5 dualies in HTC phones. aren't we going to hit a major battery life problem?

or are these dual core chips smaller than the current crop?

Battery life should be similar if the reduction in xtor size is enough to compensate.

However, other people understand this much better than I do. (I'm not sure if the reduction in power from one node size to the next is completely linear. It may be that power optimizations through the OS and additional features on the newer chip also help).
 

hans007

Lifer
Feb 1, 2000
20,212
18
81
by the time this chip comes out, amd / intel will have some 22nm chip competing against it. i'd like to see it come out like say today, and it could probably do fine running some arm build of ubuntu 11 or something, but it just wont happen.

if anything the companies that make the good netbooks wont go near it because its a chicken and the egg problem. they wont want to pour money into it, to have it not run windows. at least most wont. maybe hp would, they seem to have the balls to randomly try stuff compared to say dell but still it would be hard to justify the investment, if some 22nm windows compatible chip is just around the corner.
 

cbn

Lifer
Mar 27, 2009
12,968
221
106
by the time this chip comes out, amd / intel will have some 22nm chip competing against it. i'd like to see it come out like say today, and it could probably do fine running some arm build of ubuntu 11 or something, but it just wont happen.

if anything the companies that make the good netbooks wont go near it because its a chicken and the egg problem. they wont want to pour money into it, to have it not run windows. at least most wont. maybe hp would, they seem to have the balls to randomly try stuff compared to say dell but still it would be hard to justify the investment, if some 22nm windows compatible chip is just around the corner.

Intel is ahead of GF/TSMC on process nodes, but for some reason they really lag behind with atom.

For example, Moorestown (atom without PCI bus) will built on 45nm even though it will be released in 2011.

Cedar Trail (atom for netbooks) will be the first 32nm atom chip. It will also be released in 2011.
 

bryanW1995

Lifer
May 22, 2007
11,144
32
91
I read this over on EETimes too, did anyone else notice they are stating clockspeed and power-consumption numbers for a design that has yet to be implemented into silicon?



This is the kind of sensationalism pre-release headline hype that just grinds my gears.

"W00t! 2GHz 1.9W Osprey A9 in the house bitches!"?


? the osprey A9 has not actually been fabricated in reality; silicon for the device does not exist, any and all claims of clockspeed and power consumption are the product of wishful hoping of Ted, our designated "water the plant every other week" guy who can't really be trusted with anything but making awesome headlines

(tongue in cheek of course, they do know the designed capabilities and can reasonably project power-consumption with the assumption that the fab process is hitting all the spec'ed parametrics, but touting design expectations as if they are verified functional silicon realities is just shoddy marketing gimmicks in my book)


my favorite part is the tsmc 40nm reference. maybe reuters has been asleep for the past 18 mos. could somebody tell them that tsmc can't make enough 40nm wafers for their current clients???
 

bryanW1995

Lifer
May 22, 2007
11,144
32
91
I think its you that's living in a different world. AMD and Intel has the advantage of being in a fierce competition(that took lots of companies down with) and public scrutiny with hardware reviewers. These comments are similar to people that touted their horns that RISC was irreplaceable in servers. First chip that challenged that view was the Pentium Pro, by the time AMD and Intel were releasing Athlons and Pentium IIIs, most of the single and dual processor advantages of RISC vaporized.

Part of the demise is of course volume. But another is because of the light that's shone on AMD and Intel all the time. With ARM you gotta just take manufacturer's claims because there are no standard way of benchmarking(how do you benchmark on a phone?!?). Unlike PCs that have multiple component manufacturers that has a standard, ARM platforms are all proprietary and the comparisons are always Apples vs. Oranges.

You know manufacturers are sometimes are right, but even they sometimes don't know how do do things properly. Intel with Netburst, and for a smaller example AMD's C&Q and performance degradations due to OS interaction.

-You know for same amount of transistors at the same process technology, more transistors=more power
-Everything equal, more performance=more transistors=more power

-x86 decoder has been known to take minimal amount of area. Even ARM has decoders, its not like they don't have them!

Unless these guys know miracle technological advancements the server RISCs don't, I don't see how they are so much better than x86. ISA differences are minimal nowadays.

ARM might have some advantage, but its usually not the idea that makes the product, but the implementation. Just like Intel is learning to get to low power, the ARM guys are figuring out high performance. Both aren't quite there yet.

sure if everything is equal, but everything here is clearly NOT equal. x86 generally sucks, however it has a ton of momentum behind it. arm will have problems, as mentioned, not from lack of performance but from lack of software.
 

DrMrLordX

Lifer
Apr 27, 2000
23,006
13,113
136
Interesting choice of thread-necromancy. So, this announcement was made nearly a year ago . . . how much closer are we today to seeing 2 ghz dual-core A9s in production silicon?
 

cbn

Lifer
Mar 27, 2009
12,968
221
106
Interesting choice of thread-necromancy. So, this announcement was made nearly a year ago . . . how much closer are we today to seeing 2 ghz dual-core A9s in production silicon?

I am not sure, but lets say the drive for higher end ARM software actually occurred?

What kind of target would make the most sense? Atom or whatever the future equivalent of CULV arrandale ends up being?

My Very humble guess is that GF (or TSMC) would be more capable of releasing small high power chips than it would be larger low leakage silicon (similar to Intel CULV arrandale). I just don't know how far ARM can be pushed. However, I did see some discussion of 3 Ghz designs being possible on other websites.

P.S. Sorry for the thread Necromancy. I can see it has caused some problems.
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,787
136
Intel is ahead of GF/TSMC on process nodes, but for some reason they really lag behind with atom.

For example, Moorestown (atom without PCI bus) will built on 45nm even though it will be released in 2011.

Cedar Trail (atom for netbooks) will be the first 32nm atom chip. It will also be released in 2011.

Process tech wise, Intel says their LP/SoC process is a year behind the HP process. However, the embedded and ultra portable devices do not have a established market, so the device manufacturers are extremely careful. That leads to extra delay from when the platform is available.

Imagine that on PC. You can't buy CPUs/GPUs/Motherboards/etc seperately, but rather as a whole system from vendors like Dell/HP and such. If they delay the systems, it won't matter if Intel introduces the hardware earlier, because we can't get our hands on them. That's actually true in laptops. It doesn't mean the next generation systems will come earlier either, because the manufacturers want to make profit on the current gen systems. For example I waited 7 months or so extra to get my Viliv S5.


All of this wouldn't be a problem if everything was firm. But Atom on CE devices like TV and on smartphones, they have nothing. Even MeeGo won't be fully ready until October of this year or so.
My Very humble guess is that GF (or TSMC) would be more capable of releasing small high power chips than it would be larger low leakage silicon (similar to Intel CULV arrandale).

IMO the opposite is true. The chips that TI and Altera goes to fab at TSMC are small die, low speed, and low leakage devices

sure if everything is equal, but everything here is clearly NOT equal. x86 generally sucks, however it has a ton of momentum behind it. arm will have problems, as mentioned, not from lack of performance but from lack of software.

As long as Intel doesn't aim Atom at super ultra low end/cheap devices in which the CPUs sell for <$3 they won't have much of a problem.
 
Last edited:

ModestGamer

Banned
Jun 30, 2010
1,140
0
0
That was years ago, the code is so very different now. I wouldn't want to be the one challenged with the task of porting windows to ARM, not even for several millions dollars in salary.


windows is really mostly a gui with alot of add ons streamlined into it. at the end of the day I would think the compiler could be easily manipulated to build the kernel for ARM. its all the add on's and debugging that would really eat time.
 

aphorism

Member
Jun 26, 2010
41
0
0
I think its you that's living in a different world. AMD and Intel has the advantage of being in a fierce competition(that took lots of companies down with) and public scrutiny with hardware reviewers. These comments are similar to people that touted their horns that RISC was irreplaceable in servers. First chip that challenged that view was the Pentium Pro, by the time AMD and Intel were releasing Athlons and Pentium IIIs, most of the single and dual processor advantages of RISC vaporized.

Part of the demise is of course volume. But another is because of the light that's shone on AMD and Intel all the time. With ARM you gotta just take manufacturer's claims because there are no standard way of benchmarking(how do you benchmark on a phone?!?). Unlike PCs that have multiple component manufacturers that has a standard, ARM platforms are all proprietary and the comparisons are always Apples vs. Oranges.

You know manufacturers are sometimes are right, but even they sometimes don't know how do do things properly. Intel with Netburst, and for a smaller example AMD's C&Q and performance degradations due to OS interaction.

-You know for same amount of transistors at the same process technology, more transistors=more power
-Everything equal, more performance=more transistors=more power
-x86 decoder has been known to take minimal amount of area. Even ARM has decoders, its not like they don't have them!
x86 has a lot of overhead not in area but in terms of circuit design, it is going to use a lot of power is very difficult to design. this will end up costing more pipeline stages or higher power consumption. x86 has to decode a lot more information too. instruction lengths must be decoded because of variable instruction lengths so it is impossible to know where the next instruction is in memory. this is a major issue with x86.

also transistors on the same process dont necessarily consume the same amount of power.
Unless these guys know miracle technological advancements the server RISCs don't, I don't see how they are so much better than x86. ISA differences are minimal nowadays.

ARM might have some advantage, but its usually not the idea that makes the product, but the implementation. Just like Intel is learning to get to low power, the ARM guys are figuring out high performance. Both aren't quite there yet.
you sound like a victim of technical marketing. all RISC ISA's are much better than x86. the problem stems from x86's origin. i could go into detail but i dont really feel like explaining everything.

http://www.realworldtech.com/page.cfm?ArticleID=RWT021300000000
While the integer performance gap between the best RISC and CISC processors has closed over the last thirteen years, the deep and fundamental differences between the two architecture design concepts have not. The "RISC and CISC are converging" viewpoint is a fundamentally flawed concept that goes back to the i486 launch in 1992 and is rooted in the widespread ignorance of the difference between instruction set architectures and details of physical processor implementation. Modern out-of-order execution x86 and RISC processors *do* have very similar organization in their back end execution engines, both of may which contain 40 or more physical renaming registers. While RISC data paths are driven directly by RISC instructions, x86 data paths are similarly driven by sequences of simple, shallowly encoded microcode-like control words called micro-ops, or provocatively, RISC-ops.
essentially anything that x86 does that is RISC-like can be done natively at no cost for RISC.
 

cbn

Lifer
Mar 27, 2009
12,968
221
106
IMO the opposite is true. The chips that TI and Altera goes to fab at TSMC are small die, low speed, and low leakage devices

I see what you are saying. Most companies involved in ARM manufacturing have specialized in low power IP. Could High Power IP with ARM possibly find a niche?

For example, In the Anandtech iPad review atom looks roughly twice as fast a 1 Ghz Apple A4 silicon with respect to loading webpages.

http://www.anandtech.com/show/3640/apples-ipad-the-anandtech-review/17

As long as Intel doesn't aim Atom at super ultra low end/cheap devices in which the CPUs sell for <$3 they won't have much of a problem.

I know I am getting ahead of myself, but based on what you are saying "high power" ARM (which may or may not be easier to develop than smart phone silicon) could find higher profit margins in the server and laptop space?
 
Last edited:

cbn

Lifer
Mar 27, 2009
12,968
221
106
http://www.cio.com/article/602053/9_Real_iPad_Alternatives_?taxonomyId=3234

Here are some example of Tablets slated for launch in Q4.

It looks like a competition between Windows 7 Home Premium/atom vs Tegra 2/Android.

With Nvidia using only 1Ghz cores and Android OS still in its infancy I think we are seeing compromises being made. At the moment, I am skeptical if this will be enough to gain a foothold.

But maybe I am underestimating the Google UI? Will that be enough to compensate? Or are all these $250 netbooks floating around just plain better for simple tasks?
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,787
136
you sound like a victim of technical marketing. all RISC ISA's are much better than x86. the problem stems from x86's origin. i could go into detail but i dont really feel like explaining everything.

http://www.realworldtech.com/page.cfm?ArticleID=RWT021300000000

essentially anything that x86 does that is RISC-like can be done natively at no cost for RISC.

So how much power differences do you think exists between x86 and RISC? Let's see what you think of it. 50%? 2x? 10x?
 

cbn

Lifer
Mar 27, 2009
12,968
221
106
http://www.computerworld.com/s/arti...GHz_processor_coming_to_Android?taxonomyId=15

In this article (from June 2010), the CEO of Motorola mentions a 2Ghz smartphone by the end of this year.

But at that level of power would these CPUs begin to need heat sinks?

From what I have seen (browsing through iFixit teardowns of various iPhones today) heat sinks don't exist on low power ARM. In fact, they apparently run so cool that the RAM can be stacked either below or on top of the CPU (not sure which) to save PCB space?