It looks like 2000 is shaping out to be a fascinating year for the technology space. The year kicked off with a bang when AOL announced it was acquiring Time-Warner, changing the Internet landscape by combining .com with a .bam. Today, a second major landscape change happened in the computer industry, as Transmeta, aka Silicon Valley’s most secretive company, announced what it had been working on for the past few years.
Many of the media seemed to miss the story, however, treating it as just another company announcing a new computer chip. What they missed, largely, is the revolution heralded by Transmeta’s code morphing software. What it does is translate X86 instruction into Transmeta’s native code.
The revolution behind this lies mainly in the fact that this architecture could theoretically allow machines running Crusoe (the Transmeta line of chips) to act not only like x86 machines running either Windows, Linux, BSD, BeOS, or Solaris for x86 but also run code designed for any other chip. All they would have to do is throw in a different version of the morphing software on top of the chip (say, for example a G4 software layer) and run the OS running on other machines.
Logically, what Transmeta has done is simply create the most versatile chip in the industry, allowing it to morph into all the most popular computer chips out there through software emulation.
And because the first translation layer they built is the x86 one, it means that their chip can now power Windows and Linux boxes as well as any Intel chip.
In other words, what Transmeta announced today is that they had built a chip that is what Sun is trying to build with the Java chip, and in the process they have created what is essentially a virtual machine that addresses directly the chip, completely bypassing the OS.
Caching for better performance
Another interesting part of the announcement of an adaptive algorithm in their chip set that caches the most often use instruction sets. As a result, the performance of the chip is greatly increased because the Transmeta software caches the information that repeats itself. In the demo they gave at their press conference, they showed how running a DVD movie on a Pentium III processor compared to running the same movie with the same DVD reader on a Crusoe chip. The difference was stunning. While on the Pentium III, the idle time drop from roughly 90% to roughly 50% for the duration of their running the application, the Crusoe chip first saw a spike to less than 50% for a couple of minutes and then went back into idle mode as all the necessary instructions to run the DVD players had been cached in memory.
The bottom line here is that they have essentially created an adaptive multi-tasking environment on a chip.
Built-in power management
The other thing that was interesting was their approach to power management. For starter, the Crusoe chips do not call on all areas of a processor all the time. Because of the adaptive nature of their memory caching system, they show a major performance increase at a lower consumption rate. According to the Transmeta web site,
When an application doesn’t need peak processor performance, the model TM5400 can save power by reducing its clock speed and voltage. Using Transmeta’s LongRun power management technology, software continuously monitors the demands on the processor, dynamically and smoothly adjusting the processor’s speed to exactly what is needed to run the application.
You probably keep more than one application open on your computer at a time (for example, as I’m writing this, my email client is running in the background, as well as a few browser windows from which I’m getting information) but most of those being in the background, they should take major amounts of CPU. A Crusoe enabled machine would realize that and reduce the voltage consumption of those tasks, therefore increase the battery life on your computer (this is, of course, aimed at the Mobile market Transmeta is going after).
On a regular machine, the chip knows either how to turn a process on or off but not how to degrade it to lower power consumption. As a result, Transmeta has built a better mousetrap.
Upgradeable through the Internet?
Because the core of the chip resides in a software layer that sits above the hardware level, it is possible to create a chip that is upgradeable through the Internet. A few years ago, the first ROM-upgradeable modems came out, allowing users to move from a 28.8kbps modem to an 56k modem by just downloading software. Theoretically, Crusoe chips should be upgradeable in much the same way, which makes for an interesting world where hardware and software become closer.
But why does Linus Torvalds work there?
The question on the mind of hordes of Linux users was why would Linus work at a chip company? Well, today, the answer was given. Along with the announcement of the new chip, Transmeta announced Mobile Linux, a version of Linux designed for systems without hard disks, such as Mobile Internet devices (for example, Web pads, palmtops, etc…)
The principal enhancements for Mobile Linux will be in power management and in the reduction of the memory footprint. So that’s what Linus has been working on all this time, as well as being part of the team that built the morphable software. All and all, one can say that it was really smart of them to add him to their roster of talent, as it will fire up the Linux community to develop for this new version of Linux (and yes, Mobile Linux will be open source).
In what may be a departure from the industry’s playbook, Transmeta decided to announce the products they had on hand. Yes, the chips are available today (they mentioned that IBM would manufacture some of them) and no, they won’t answer any questions about the future. In other words, this is no vaporware but a very real set of products. This could change the way companies around the industry release their products.
The losers: Palm Computing, Sun, Intel, Motorola and AMD
All and all, this announcement changes the landscape for a couple of major players. For starters, Palm Computing might find itself in the middle of a very big battle. Because of the low power consumption of the Crusoe chips and the introduction of Mobile Linux, it is now possible to envision Palm-sized devices that can run audio and video. This represents a major threat to Palm Computing’s installed base.
Second among the losers today are the chip companies. Because they can run the x86 set, Transmeta’s chips are a potential replacement for Intel and AMD chips in the laptop market. A lot of the problems those chips have encountered is that the higher the processor speed, the hotter the chips run, and the larger the fans that need to be used to cool them. If Crusoe holds up as well as they demoed, Transmeta could capture a major lead in the laptop market. Add the morphing software architecture and Transmeta could also become a competitor to chip makers Sun and Motorola.
Sun is further hurt in the fact that this is a virtual chip that takes away from the concept of virtual machines that Sun pushed with Java. As a result, Sun’s hope of covering the market with Java devices as well as their JINI-everywhere strategy may have to be rethought.
The winners: Linux, Microsoft, Consumers
Linux is a clear winner with the introduction of Mobile Linux, which will strengthen Linux’s gain in the computer market by offering it as an alternative to any other OS in the hand held market.
Microsoft, interestingly, also becomes a winner with this. With Intel making some moves to back alternative operating systems, Microsoft was finding itself in the difficult position of being somewhat subservient to its biggest ally. Yes, there’s AMD but Crusoe is yet another competitor to Intel, which only strengthens Microsoft’s position in dealing with the chip manufacturer. Also, Microsoft gains some advantage here because its Windows OS can now run at better performance levels on handheld devices. Yes, it will have to battle Mobile Linux but this could make the main Windows line (Windows 9x and 2000) an alternative platform in the portable device market, which means that Microsoft could now ditch their efforts on WinCE and still have a card in that new market if they want to.
Consumer also stand to win from this announcement as it heralds a new age of mobility. Soon, a lot of low powered mobile devices will hit the market: Portable DVD players and wireless Internet devices now have more of a chance in the marketplace since they won’t consume as much in batteries. I think we’ll all be happy to see those new devices hit the market.
All and all, it was yet another day when a major announcement changed the computer landscape. Since that’s two major announcements in as many weeks, I wonder what’s going to happen next week.