February 14, 2008 9:00 PM PST

Secret recipe inside Intel's latest competitor

Secret recipe inside Intel's latest competitor
Related Stories

Silent start-up readies to take on Intel in notebooks

February 6, 2008

Cell chip: Hit or hype?

February 9, 2005
It works like an Intel chip, but looks like the Cell processor.

That's one way of describing the energy-efficient multiple core processors being devised by secretive Montalvo Systems. The Santa Clara, Calif.-based company has come up with a design for a chip for portable computers and devices that--when finished and manufactured--will theoretically be capable of running the same software as chips from Intel or Advanced Micro Devices.

Montalvo's chips, however, will fundamentally differ from the latest Core or Opteron processors from Intel and AMD in that the cores on its chip won't be symmetrical, i.e. identical to each other. Instead, Montalvo's chips will sport a mix of high-performance cores and lower-performance cores on the same piece of silicon, similar to the Cell chip devised by IBM, Toshiba, and Sony, according to sources close to the company.

By merging asymmetrical cores onto the same piece of silicon, Montalvo can cut power consumption by dishing applications that don't require a lot of computing firepower onto less-powerful, more energy-efficient cores. Applications could conceivably also be shuttled to low-power cores after their need for high-performance elapses: Microsoft Outlook, for instance, requires a burst of performance during the launch phase but far less once it's running.

Asymmetrical cores can also provide better performance on applications such as video if programmed for that purpose, say proponents of the architecture. The Cell processor became the first chip to successfully champion this idea. The Cell consists of a primary microprocessor core and an array of "synergistic processing elements" that can be programmed to perform discrete tasks like managing networking or video streaming.

Cell chips have primarily been used inside Sony's PlayStation 3, but IBM has inserted Cell chips in some server blades. Toshiba plans to put the chip inside TVs and may put it inside PCs. (While the initial Cell comes with eight synergistic cores, chips can be made with fewer.) Mercury Computer Systems has also adopted Cell for some computers.

Montalvo has not stated whether it has adopted an asymmetrical core to save power, boost performance on media applications, or both. In fact, the company doesn't say anything at all. The closest it has come to a public statement are shirts handed out to employees saying that the company can't say what it is up to. Montalvo declined to comment for this story.

The somewhat different, asymmetrical nature of Montalvo's chip in part helps explain why investors have put more than $73 million into the Sisyphean task of taking on Intel. Montalvo wants to land its chips into all sorts of portable computers: notebooks, handheld devices such as the OQO, and ornate smartphones. Several companies, however, have tried this and failed because of the daunting nature of trying to compete against Intel. Cyrix, Transmeta, Rise--none of them ever lived up to its advance billing. Only AMD has survived, and AMD has lost more money that it has made in its 30-year plus existence.

Montalvo is funded by people who've tangled or been entangled with Intel before too. NEA-IndoUS's Vinod Dham, who sits on Montalvo's board, was one of Intel's chief chip architects during the Pentium era. He then went to NexGen, which designed an Intel-compatible chip, and then AMD when it bought NexGen.

Montalvo's CEO is Matt Perry, who also served as chief executive of Transmeta, which once tried to take on Intel in notebooks but now largely concentrates on technology licensing. Peter Song, Montalvo's chief architect, earlier founded a company called MemoryLogix, which tried to build low-power Intel-compatible chips. Other current and former employees include Greg Favor (formerly of NexGen and AMD) and Mike Yamamura. (CNET Networks blogger Peter Glaskowsky is chief systems architect for Montalvo and is listed as a co-inventor on two published Montalvo patent applications, but he was not involved in any way in this story. CNET is the publisher of News.com.)

Some of Montalvo's patent applications can be viewed here.

Although it has designed a chip, Montalvo has not yet produced a chip based on its designs.

See more CNET content tagged:
Transmeta Corp., portable computer, multi-core processor, AMD Opteron, AMD

12 comments

Join the conversation!
Add your comment
Hello?!?!
VIA has been competing against Intel sucessfully for years. In fact the OQO device mentioned in this article has a VIA chip in it NOT Intel!
Posted by c7fanboy (6 comments )
Reply Link Flag
VIA fan boy
haha nice catch

but didn't via basically bought cyrix, killed their line of chips.

has via made money in the chip department?
Posted by basraw (310 comments )
Link Flag
Definition of success....
Is it really successful to have a line of products that virtually no one has heard of or cares about?
Posted by benjwah (195 comments )
Link Flag
VIA's offerings for limited market
ULP chips are good for embedded systems, closed-top boxes and computing appliances that don't need high horsepower, but for desktop computing they're only barely adequate.

Its too bad they never pushed the symmetric processing CPU's of awhile back, where you could add up to 128 of them.
Posted by savagesteve13 (104 comments )
Link Flag
Turing rules
Montalvo Systems
What to make of these guys as you say secrative but Also on the same wave trying to find their Neiches as much as everyone else.
First thing to make clear there is room for them potentially the future and groups should eye this company with calculated risk.

Right lets go into detail here

The Monecular Science and Physics.
It is becomming more advantagious to to work better with Asymetric designs because as you say you can keep things cooler and potentially save energy.
also As you go down past 32nm you hit blur lines of Quantum Distortion. Problems which arent easily solved in terms of the etching Acuracy required from a production stand point.
Designing stuff Asmetrically better gives you more breathing space to wok with in that environment.

Lets put things into perspective here Intel looks to role out 32nm solutions next year.

Also the Effort required to Design at this scale is high too.

So you hit a production bump on the road while the market finds solutionsto all this.

At 32nm its got Quite far but the Quantum Computers Arent Very Capable yet of modeling Such Physics well enough to aid solutions yet.

This means people will have to potentially adopt other methods and there will be much competition to get in on the act during the bump stage to make better use and innovative use at the current stage.
This opens the door for two oppertunities for new comers and the Industrial Engineering !!!Dogs.

1) for new technologies like fast flash memory based IC systems for specalised communications chips and other New Data processing concepts.

2) For the underdogs to bit better at the heels.

What you need to do in assesing this is working out the Human workforce Resource dynamics to take it all on.

One Area that is lacking is the Areas of Physics and partially Moneculer science.
Scientists Are Much in demand in this area especially one that can work well to the demands of such a mighty industrial load.

Also Good Wide Adaptable engineers is a bit of a wild card.
It is Unknown how well these guys will be able to take dreams and designs to the peddle and yeild pratical economically viable solutions.

The Management is Vital. Intel Has Some real Managment Beasts of man mind power at their beckon who are the most disiplined engineering industry minds on the planet and even during a Industrial bump their still gonna want to Dominate (History proves this if nothing else).

Wasnt it the CEO of AMD that sadi the Big boys Make microprocessors.

Well one day little childern will and the Big boys will be tuning Crystal arrays and Using Magic like technologies to reach the Stars (Hitachi Technology is simply the the jurney not the destination)

Today how ever its the Toughest and most Mentally rewarding business on the planet and theres a large gravs yard in Tech experts Caravan Parks to prove it.

Intel has yet to demonstrait how well it can scale down using it Hafnium layer added.
down at them there nm's is a potentially tough science.
Infact 32 nm is impressive.

I'm not talking about the impossibility of things here i'm talking about industrial viability.

Just like you don't worry about when oil runs out you worry about when it hits peak production and how well its output overall can cope.

I just noticed India is bumped the oil costs to fund oil groups better just after Oil groups helped the US economy.
So this moment in time isn't actually one of sudden down turn but it is one that requires Effort from all and a Devotion to lead towards a better future.
Posted by wildchild_plasma_gyro (296 comments )
Reply Link Flag
Huh?
Ummmmm..... no comment
Posted by DivingDancer (17 comments )
Link Flag
duude
you are smoking the good stuff wildchild.
Posted by megustansalchichas (153 comments )
Link Flag
Not exactly the right comparison Michael
The Cell processor is a heterogeous processor because the PowerPC core and the SPEs have different instruction sets and capabilities. If you're right about Montalvo, this is an asymmetric processor, but all execute pretty much the same instruction set (x86). The challenge for asymmetry is having the operating system understand which cores to use for which application or power mode.
Posted by kkrewell (41 comments )
Reply Link Flag
montalvo-catchy!
It's difficult to imagine any company who want to go head on with
Intel. However, the concept and development is interesting to say
the least. It sounds like a more solid competitor than the CPU
trovalds guy produced.

Keep giving Intel competitive pressure. We all want that!
Posted by johnedlt (12 comments )
Reply Link Flag
It's a real long shot
The problem is that it's not that hard to create a x86 compatible cpu and the market is huge. So a lot of companies have spent an incredible amount of money doing so.

However that huge market is so seductive it even convinces smart people to ignore the fact that running x86 software and running x86 software better than Intel are wildly different things.

It's a goal whose difficulty is similar to climbing Everest in short pants.

Maybe they'll at least end up with some decent IP like transmeta that will have some residual value.
Posted by www.hdgreetings.com (39 comments )
Reply Link Flag
Root For The Underdog!
If the Giants can stymie the play stealing Patriots, then I say Montalvo can usurp Intel's grip on the industry! What's capitalism without competition? Just shut up and build the chip! Oh wait, you're already doing that!!! GO MONTALVO!!!!!
Posted by ideletejunkemail (9 comments )
Reply Link Flag
Good luck competing with Intel...
...you'll need it.
Posted by Fil0403 (1303 comments )
Reply Link Flag
 

Join the conversation

Add your comment

The posting of advertisements, profanity, or personal attacks is prohibited. Click here to review our Terms of Use.

What's Hot

Discussions

Shared

RSS Feeds

Add headlines from CNET News to your homepage or feedreader.