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 AMD.
repeat?
Only if they decide to hire Linus …
Yep. What we have is going to take the world by storm but we can’t talk about it yet! And now Transmeta would be completely forgotten if Linus hadn’t worked there.
It’s called “delusions of grandeur”. 😉
Transmeta is exactly what I thought of as soon as I read the blurb.
So, what we have here is a theoretical competitor, and we won’t know whether asymmetrical cores will pay off until the design is reduced to silicon. In theory, though, it sounds like a interesting idea to have a mixture of low- and high-power consumption cores on a single chip. Of course, I’m not sure what that means for locality and pipelining.
“Small, secretive start-up lays plans to challenge Intel”
Here’s a prediction. A few years after the hype has died down, their business model will devolve into one based on suing Intel for alleged patent infringement.
Ah, this brings back memories. Anyone recall the frenzied excitement when someone found the “secret” message in the HTML of the early Transmeta webpage?
——————————-
Yes, there is a secret message, and this is it: Transmeta’s policy has been to remain silent about its plans until it had something to demonstrate to the world. On January 19th, 2000, Transmeta is going to announce and demonstrate what Crusoe processors can do. Simultaneously, all of the details will go up on this Web site for everyone on the Internet to see. Crusoe will be cool hardware and software for mobile applications. Crusoe will be unconventional, which is why we wanted to let you know in advance to come look at the entire Web site in January, so that you can get the full story and have access to all of the real details as soon as they are available.
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
Here the article implies is that a chip having is inherently like the Cell BE
but since the Cell BE is “asimmetrical” in that the SPE cores perform stream processing on data sent to them by the PowerPC Processing Unit, which should be the only general purpose one that can run PPC code and performs I/O to and from the other cores, it is a completely different thing
At least something NOT to compare a cpu with a mixture of cores having different performance levels ( but all compatible to the same ISA and able to run the same general purpose x86 code, as the ability -referred to in the article- to migrate lightweight, or idle, processes to lower performing cores, would imply )
The author would have done better by citing the relevant patent (“scalable performance multiple core cpu”.. ) than to take the Cell as an example …
Your point is correct (your description of Cell isn’t). However I think the author means it’s like Cell in that there are a number of cores and they are not identical. That much is true but the analogy can’t be taken beyond that point.
It’ll be interesting to see what this processor can do, current x86 processors have to cripple themselves to get a low power rating and even then they are nowhere near low enough for the power ratings needed in smartphones.
If someone can figure out how to get down lower they’d have quite a serious advantage over Intel and AMD (and Via for that fact).
Don’t expect them to be an independent company for long though, Nvidia needs to get into the desktop processor game.
Looks like they forgot about VIA. Ever since they hit the market with their Mini-ITX platform 6 years ago, they’ve been taking the x86 industrial, embedded, and appliance markets by storm. Granted they already had quite the successful chipset business, but tapping into the processor market was quite new for them when they bought Cirix in 1999. They’ve definitely proven themselves to be a worthy competitor in the “bigger isn’t better” markets.
Very good point.
I wonder if Montalvo will produce a processor that draws low enough power to not require a fan, but has enough processing power for desktop use. That would be a seller.
Intel’s 80386 did that.
I meant today’s desktop. Not the 80’s.
The 1988 desktop was really about like the 2008 desktop… except for the TCP/IP stack and the constant whining by users regarding how they’re unhappy with their fonts. Those are definitely facets of the 90’s and continue into the 2000’s.
What we need are not new processor designs… but new users. 😉
slowly. Although it’s true people still want a powerfull processor, they also want a cooler, quiter, more efficient processor. It’s the only way to continue with the rapid increase’s in the number of new computers. Very encouraging news if they can pull it off, and it may force other manufacturers to sit up and rethink thier own paths. If they can get it to market, and if the ultra portable guys want it, they could finally end the need for speed, mines faster than yours era we’ve just come through.Eeepc proved that cheap works, maybe cheap and efficient is next.
Just too much silicon has to be burned on x86 instruction translation. An architecture like ARM has a far better instruction set with a good balance of risc/cisc instructions which have very good instructional density and ability to directly execute. Smaller dies and lower power.
Well maybe they truly have some vast innovation…or maybe innovative patents at least…
The fact they have Transmeta people on board could indicate it won’t be a native x86 but rather use a translation / JIT system like transmeta did.
That’d be one way to get power usage down.