AST co-founder seeks room inside the PC

LOS ANGELES--There is very little room inside a PC these days, both literally and figuratively.

Safi Qureshey, best known as the "S" in the old PC maker AST Research, is trying to get a new chip company off the ground.

(Credit: Ina Fried/CNET News )

Tiny Netbooks leave little physical space for any added components, while brutal price competition means it's just as hard financially to convince PC makers they need something extra.

Still, that's what Safi Qureshey, who co-founded PC maker AST Research almost three decades ago, is trying to do.

His start-up, Quartics, is pitching a chip that would augment the PC's main processor and graphics card with a programmable chip for handling things like Flash movies and video conferencing.

"It's a co-processor. It does not replace anything," Qureshey said in an interview at last week's Windows Hardware Engineering Conference in Los Angeles. Quartics' main targets, he said, are Netbooks and cheap laptops that don't have a lot of horespower to spare.

"We significantly enhance the video quality of a very low-cost laptop and we enhance battery life," Qureshey said in an interview. (For more from the interview, check out the video interview below.)

The company has been trying to get off the ground for a while now, having started in 2003. It has a number of venture backers and but has not publicly announced any PC maker customers.

"We are working very closely with one. We just don't want to preannounce their name," he said.

Quartics' chips are being manufactured in sample quantities, Qureshey said, with production volumes planned for the first quarter of next year.

But the advent of Netbooks makes Qureshey hopeful that such chips are now poised to take off.

Cost will certainly be a key factor. Qureshey said the company hopes to get the volume price of its chip "in the teens" of dollars as opposed to the "twenties" where it is today. Some of that cost, he said, can be offset by using a cheaper main processor or graphics chip, he said.

Microsoft was showing off a sample of Quartics' chip in one of its booths at WinHEC.

[sal-magnone]

So wait, you use a lower horse power processor so you have something to add horse power to? My head hurts.

I know this actually makes sense (go lower power because you still beat the higher cpu with this new chip) but that is not going to change my sense that I suddenly don't want one of these in any of my devices.

[ahickey]

I don't see the market.
By the time they get the price into the teens and the deals done with the manufacturers and the software in place to support the chips built in graphics chips and processors will be faster, more power efficeint and cheaper

[ohdotoh]

The PC of tomorrow will have more processors, and they will be more specialized than those in the PC of today. Whether or not this company will be one of the companies that makes one or more of those processors is an open question, but that is the only way that performance can continue to increase. If were not true that processors will not only increase, but become more specialized, then today' s PC would not even need a video card at all since today's Quad P4 class CPU has more power than a 386 and its VGA card combined.

[inachu]

I thought we moved away from co processors back in the late 1980's?
I guess we can have these new coprocessors help make vista faster.
Or how about a co processor just to make and keep the task bar responsive 100% everytime. Or a co processor just used to help bring boot time to a acceptable level.

[c|net Reader]

The scheme is to lower the total cost of a certain level of usable performance. Assuming the low cost laptops and netbooks are used for web browsing, word processing, and e-mail, there's not a great deal of pressure on the CPU except when doing video and animation. If the coprocessor can make those perform well with a lesser CPU, then the overall cost drops without performance degradation.

When built-in graphics and CPUs get more powerful, this coprocessor may still add value. The point where it doesn't isn't revealed by the scant information in this post.

[maeckg]

Signal processor is likely what Safi is talking about: a processor that is extremely good at audio and video processing, much more efficient that a general processor like the main Intel or AMD CPU. It is not just a mathimatical co-processor like the first PC CPUs needed to crunch spreadsheets and other math intensive applications; it was already integrated into the main CPU in the 486 era.
This could be tied into an embedded graphics subsystem, but he is referring to a programable signal co-processor, which would provide flexible proformance for streaming media that is not just a graphics application. It is an idea that idea that let the Amiga and Atari TT run circles around PCs in media/video work over a decade ago. Intel could have included a signal co-processor in it's chip set specification in the 1990s to make the PC much more powerful with media, but loaded the Pentium with microcode that was inefficiently processed. Shamej, they did not know about YouTube, but multimedia was already exploding.
We have arrived to a time that software and OSs are more able to use multiprocessors to provide real performance gain instead of brute processing power. Intel has realized that efficiency is a premium advantage, especially in increasinly popular laptops and mobile computers.Safi's processor does fit into a better power savings and performance envelope especially for netbooks. Usable performance at low cost is the crux of the netbook. But I hope the media co-processor works its way into all laptops since we use them more and more for streaming media. It provides a further iteration to multiprocessing.