Transmeta bets on LongRun2

While Intel and Advanced Micro Devices continue bringing low-power mobile chips to market, Transmeta is faced with the possibility of shifting its focus from operations and toward licensing its technologies.

Adding pressure to Transmeta, AMD last week started shipping the 90-nanometer versions of its Mobile Athlon 64 processors, code-named Oakville, to PC makers. Systems based on the chips are expected to be available in the coming months.

Four years after its debut, Transmeta has racked up $591 million in losses and watched rival Intel enjoy the spoils of an industrywide move toward low-power mobile processors. Transmeta, in fact, said earlier this month that it might have to scale back operations to concentrate on its fledgling licensing business if it needs to raise cash in the coming quarters and runs into problems securing credit.

The main benefit as well as the main problem of Crusoe, its first chip, is its code-morphing software. Transmeta used a 128-bit VLIW (very long instruction word) architecture to build Crusoe, but that architecture was not compatible with that of the x86 used by Intel and AMD processors or the vast majority of the world’s PC software. So software was used to translate x86 instructions to Transmeta’s hardware.

Transmeta’s software approach allowed it to use fewer transistors on Crusoe than most x86 chips, cutting the power consumption of the chip. However, PC vendors were looking for more general-purpose performance. Software simply can’t duplicate the raw performance of a well-designed collection of transistors.

Early reviews were not kind. PC vendors were further disappointed by manufacturing delays, as Transmeta tried to make the jump from 180-nanometer process technology to 130-nanometer process technology in 2001. Transmeta relies on third-party foundries to manufacture its designs.

“It’s difficult when you’re a one-product company to execute 100% of the time,” said Kevin Krewell, editor of Microprocessor Report.

Intel countered Transmeta’s Crusoe in 2000 by launching the low-power Pentium III. By 2003, Intel’s Pentium M processor, however, had surpassed Transmeta’s low-power benefits.

According to analysts, Transmeta’s best hope moving forward is the Efficeon chip, a revised version of Crusoe that addresses many of the flagship chip’s performance issues. Efficeon runs faster than Crusoe, uses a 256-bit VLIW architecture, and has a new version of LongRun to boost performance and reduce power consumption.

“Transmeta’s chances in the market depend on how the 90-nanometer (transition) plays out,” said Dean McCarron, principal analyst at Mercury Research.

Transmeta has lined up partners, namely NEC Electronics, which has licensed Transmeta’s LongRun2. Transmeta hopes to harvest its technology by finding other licensees that will tap into its intellectual property, said Arthur Swift, senior vice president of marketing at Transmeta.

LongRun2 allows Efficeon to adjust its clock speed and operating voltage hundreds of times a second to match the workload of applications. This approach of applying power as needed has been a hallmark of Transmeta’s products since the first Crusoe chip, and the company hopes that LongRun2 not only improves its own chips, but also attracts the attention of third-party chipmakers.