Transmeta sell a family of RISC/VLIW processors called crusoe that have a built-in compiler/interpreter ("Code Morpher") that can convert 80x86 instruction set, in realtime, into its own RISC/VLIW instruction set, and hence is able to execute PC software.

Much of this technology was created for optimising a computer language called Self, and further developed by HP's project called Dynamo.

Because RISC/VLIW processors have a simpler structure and many less transistors, the Transmeta processor needs significantly less power; and because RISCy instructions generally run more quickly than 80x86 instructions it can start to compete; in theory even run more quickly.

Using less power is important for lap-tops and in server farms- the lap-top's batteries can last much longer (about twice, since the display takes considerable power), and the air conditioning requirements are far lower (more than ten times) respectively.

One downside to this technology is that the compiler takes time to run; when programs start to run the compiler has to convert the instructions and this takes time.

This leads to a practical issue- benchmarks usually don't run for long, and hence will make the processor seem slower than it really is.

Additionally, as there is only a finite cache for the converted programs, the program can need recompilation more than once; creating worse performance for a short time as you switch applications.

The story of the Transmeta Corporation is, in some ways, symbolic of the story of the Internet bubble of the late nineties.

Some point in early 1995, Dave Ditzel, of RISC-fame and formerly from SPARC started up Transmeta as its CEO. Paul Allen, one of the original founders of Microsoft and ludicrously rich, provided huge sums of venture capital.

Transmeta really made the headlines when, in 1997, they managed to attract one prominent employee -- Linus Torvalds.

Transmeta at this point was shrouded in mystery. Nobody knew what the company was doing. It became a bit of a running joke in the online community -- were Transmeta making graphics hardware, or some kind of robot? Some people even denounced the whole company as a hoax.

A link to the Transmeta website appeared at the bottom of the Linux kernel archives at kernel.org, but the page only contained the message 'coming soon....'. A quick peek at the HTML source code revealed a message saying something like "Sorry, there is no secret message here."

At the end of 1998, Transmeta filed patents on Code Morphing technology, which effectively acted as an emulator to allow a VLIW - type chip to run x86 code.

In November 1999, I had another look at the Transmeta website and was excited to find this hidden in the source code:

---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.---

Yes, Transmeta were making a chip, known as the Crusoe which for a variety of technical reasons was perfect for portable devices. It could adjust its power usage between processor-ticks and it ran cooler so didn't need a fan. This enabled extended battery life, the major concern of notebook users' lives. It also did some damned clever things with an x86 software emulator layer, providing in theory the ability to make the Crusoe mimic different processors.

They also built their own version of Linux -- to a collective sigh of "so that's why they wanted Linus" -- called Midori, designed to be small, fast, and ideal for Internet appliances.

So it wasn't giant robots but it was still pretty cool, and as usual the analysts started saying "The game is up for Intel" (false), "The portable Internet device is the future" (possibly true) and "Everyone should invest in Transmeta" (definitely false).

Some companies, like Hitachi and Fujitsu, quickly announced their intensions to use the Crusoe in their notebook PCs, though it took a bit of time to get the chips out to OEMs.

Transmeta went IPO on the 6th of November, 2000 (Nasdaq: TMTA) with an initial price of $21 per share. By the end of that day's trading, they were going for $45.25 each.

These were the last of the heady days of the technology boom. People no longer believed the cliches like "anything seems possible" or "the way we do business has changed forever". Instead, world-weary venture capitalists tried their hardest to make one last quick buck before the bubble burst.

It burst. The Crusoe chips were good, but not $50-a-share good. The handheld Internet market, emergent in late 2002, was fairly non-existent in 2000. And perhaps most importantly, investors tired of high tech in general, and new high-tech in particular. Transmeta shares fell gracefully from $50 in November 2000 to as low as $2.50 a few days before everything changed. In October 2002 they bottomed out at $0.74, a measly 1.5% of their maximum value.

Transmeta today is based in Santa Clara, California and employs about 300 staff. It's smaller and a bit more realistic about its goals, but is still functioning, and as the market shifts more towards handhelds and wireless networking, perhaps we shouldn't write them off just yet.

http://www.transmeta.com/about/team/ditzel.html
http://www.tekpress.com/Archives/1999/Mar/transmeta.html
http://investor.transmeta.com/faq.cfm
http://www.redherring.com/ipo/2000/1108/ipo-transmeta110800.html
http://midori.transmeta.com/
http://quotes.nasdaq.com

Log in or registerto write something here or to contact authors.