SUNNYVALE, Calif. — Chromatic Research Inc. is rebuilding its concept of a media processor after deciding that its Mpact chip won't be sufficient to drive volume sales. The move signifies a realization that the still-promising media processor concept has a long way to go before it can become a substantial business.

Layoffs at Chromatic have begun, and a spokesman confirmed the company is trimming down to 150 employees. Chromatic employed roughly 300 as of January.

Chromatic officials remain convinced that media processors can grow into a high-volume business, and employees already have begun work on developing a new media processor architecture intended to correct specific miscues suffered by Mpact. "This new architecture will solve the issues that Mpact media processors have faced by better fitting industry design cycles, by dramatically simplifying and opening the software environment, and by offering superior performance and integration," the company said in a statement.

The company believes the most recent Mpact parts still have "life left" in them, the spokesman said. Accordingly, Chromatic will continue selling and supporting Mpact, although all future development on the architecture has stopped. Details of the new architecture won't be available until next year.

Officials also said Chromatic has secured extra funding from unspecified sources.

Mpact did find some success as a driver for PC DVD-ROM, and Chromatic officials had previously touted design wins from Gateway 2000 Inc. and Compaq Computer Corp. Still, early pioneers of the media processor — a single chip for handling graphics, video, audio, telephony and other multimedia functions — are waiting longer than anticipated for the technology to reach mainstream use. Those without a ready flow of cash are finding the going difficult.

The problems have more to do with business than with technology, said Gerry Kaufhold, an analyst with In-Stat Inc. (Scottsdale, Ariz.). While Mpact 2 appeared to be a good part, it had to compete against cheaper fixed-function ICs.

"Moore's Law has to make one more turn of the crank before the media processor technology becomes a mainstream technology," Kaufhold said. "By the year 2001, it's going to happen."

In fact, several companies believe in the media processor concept. Startups continue to emerge to address the idea, and large firms continue to invest in it: STMicroelectronics NV, for example, was a partner of Chromatic's and has put money into similar projects with other companies, Kaufhold said. "That's my ace in the hole. STMicroelectronics wouldn't be looking at media processors if they didn't see it as a threat to fixed-function chips," Kaufhold said.

Chromatic was on the right track by making Mpact programmable, Kaufhold said, because it's impossible to predict which functions are going to be demanded once media processors do come into vogue. In fact, customers will probably have conflicting demands, meaning the only way to mass-produce the chips identically will be to make them programmable, he said.

Whatever the long-term prospects, generating sales is a challenge in the short term. The key, especially for smaller firms, is to find one source of revenue steady enough to keep the company alive while it waits for the market to develop, Kaufhold said.

Philips Semiconductors, while continuing to develop and market its TriMedia processor, found just such a customer base in the digital-television market. In fact, a second TriMedia road map has been created specifically for the needs of DTV vendors.

Philips' upcoming TM-2 chip will meld TriMedia with some of the system functions necessary for digital TV. Likewise, TV manufacturers and other electronics vendors are talking with Philips to see if they can tap TriMedia for other applications.

The TM-2 parts will fall into two camps, both of which will receive all 18 digital TV formats. The TM-2700 will be able to display all 18 formats and is targeted for the TVs themselves, while the TM-2600 will output only standard-definition TV signals and is being targeted at set-top boxes.

Because TriMedia was conceived as a media processor, the chip has processing power left over when handling something like TV. That gives it a flexibility that system vendors find appealing, according to Philips officials, and Phiilps is talking with TV manufacturers about features or applications that could exploit that extra capacity.

Customers are attracted to Trimedia by "not only the flexibility but also the horsepower," said Neil Mitchell, Philips' DTV marketing manager for TriMedia. Potential uses for that power include the processing of modem information or the handling of extra data sent along with a broadcast.

No serious front-runners have emerged among those extra functions, because TV makers are not deciding on such plans yet, Mitchell said. "The initial target, I think, for most customers is to get products on the market for basic audio and video for high-definition TV," he said.

Other major players in media processors have included Samsung, which dropped its program last year, and large Japanese chip makers, which have targeted specific appliances rather than trying to make a more generically appealing part.