Transmeta: So Now We Know
The old paradigm was waste watts and transistors. The New paradigm will be waste bandwidth and save watts.George Gilder
Yesterday, after four and a half years as the most secretive company in technology - a company better known for one of its engineers than for any of its products - Transmeta finally held a formal introduction of its products and itself at Villa Montalvo in Saratoga, California.
And the story was mostly about chips. Not about Linux, or Linus.
Well, we knew that. What we didn't know was how little it was about Linux. At least on the surface. (At a deeper level it was very much about Linux, as we shall soon see.) The whole rollout took better than two hours, and Linus only appeared once, to demonstrate how perfectly Linux could run a Quake game on a Transmeta box. (He said "If I lose, don't blame the operating system," then got creamed in a matter of seconds by an opponent on Windows who was coincidentally a Quake developer.)
The presentation was led by David Ditzel, Transmeta's founder and CEO. Also presenting were Jim Chapman, VP of Sales and Marketing; Doug Laird, VP of Product Development; and Mark Allen, President and COO. All four are Chip Guys. Ditzel especially has a long and distinguished career in microprocessor design. He co-authored the RISC concept with David Patterson at UC-Berkeley in 1980, and led Sun Microsystems' SPARC development work. Laird also came to Transmeta from the Sun SPARC management team. Chapman was VP of sales for Cyrix. And Allen's chip background was at nVidia and Cypress Semiconductor.
But Transmeta is equally a software company, and that's where Linux fits in--literally. Only Linux has two attributes that make it the ideal OS for a new class of "mobile Internet appliances" that feature the screen of a laptop (or smaller), the mass of a notepad (or less) and the battery life of a Walkman (or better). Those attributes are: 1) it can fit into Flash ROM; and 2) the market demands it.
Of course, the market also demands Windows. So Transmeta introduced two new microprocessors: one to run Linux and one to run Windows, the TM 3120 and TM 5400, respectively. The Linux version is the smaller and less power-hungry of the two, and is meant to serve as the brains for a new class of consumer electronic goods that browses the Web. The Windows version is meant to compete against Intel and its clones in the lightweight mobile notebook computer market. In other words, Linux hosts the sexier category.
Both versions have little appetite for battery power. Thanks to VLIW (Very Long Instruction Word) processing, "Code Morphing" software translation and LongRun™ power management, Transmeta chips might run all day at up to 700 MHz without breaking a sweat--and hibernate for weeks as well. While the hottest Intel microprocessors might soak up a hundred watts, a Crusoe processor might use a single watt at the same clock speed, and then a few thousandths of a watt (20 mw, they say) in sleep mode.
Doug Laird impressed the audience with videos that demonstrated how Crusoe "learns" what an application is doing, working constantly to reduce battery consumption while giving the application all the processing power it needs. He also showed photographs that made an Intel chip look like you could fry an egg on it, while the competing Transmeta chip looked cool as an ice cube. To be fair, the actual temperatures involved were 221.9° and 118.8°, both Fahrenheit. But between the two was a far more significant difference: the Transmeta chip didn't need a fan or a big heat sink.
The biggest advantage for Transmeta is in the smaller, lighter, "Mobile Linux" systems. A conventional notebook computer spreads power demands across the display, the disk drives and other components. In a Web "pad" or "slate" there are no drives, no fans and in most cases a smaller, less-demanding screen as well. This means that most of the power is budgeted to the CPU. Here, Transmeta has no competition. Intel's Celeron eats batteries too fast, and the company's StrongARM isn't x86 compatible. Nor are PowerPC and other processors that aspire to low power consumption offering much competition. So, when Ditzel said "if it has a battery and a browser, it's going to be built with Crusoe," it seemed a very credible claim. (For a deeper look at these issues, check out analyst Linley Gwennap's report "Transmeta: A Revolution in Action.")
The Q&A that followed was almost surreal. The four Chip Guys sat on high stools, taking turns answering questions and holding the two chips up for the cameras, which were wielded by a dozen paparazzi who treated the tiny goods as if they were corpses of Princess Di. My question, which I was amazed nobody had asked earlier, was "Why the secrecy?" Especially when, I noted, there was no demand for it. Ditzel's answer was deft. Over the years, he said, he learned there is "a difference between hype and buzz," and that Transmeta's strategy was successful at generating the latter while isolating the company from the former.
Naturally, all the Linux folks wondered exactly what "Mobile Linux" is. Ditzel understated the matter in his opening statements, noting with sweet irony, that "we have a certain amount of expertise with the Linux OS." In the engineering press conference that followed the official introduction, Linus let on that Mobile Linux is "a small distribution" both for Transmeta and OEMs. He also seemed to agree that his professional preoccupations with mobile computing have a natural influence on Linux evolution. He had more to say about that in a private interview later in the day. That will be the subject of my next report.
Doc Searls is Senior Editor of Linux Journal
|Non-Linux FOSS: libnotify, OS X Style||Jun 18, 2013|
|Containers—Not Virtual Machines—Are the Future Cloud||Jun 17, 2013|
|Lock-Free Multi-Producer Multi-Consumer Queue on Ring Buffer||Jun 12, 2013|
|Weechat, Irssi's Little Brother||Jun 11, 2013|
|One Tail Just Isn't Enough||Jun 07, 2013|
|Introduction to MapReduce with Hadoop on Linux||Jun 05, 2013|
- Containers—Not Virtual Machines—Are the Future Cloud
- Non-Linux FOSS: libnotify, OS X Style
- Lock-Free Multi-Producer Multi-Consumer Queue on Ring Buffer
- Linux Systems Administrator
- Validate an E-Mail Address with PHP, the Right Way
- Introduction to MapReduce with Hadoop on Linux
- RSS Feeds
- Weechat, Irssi's Little Brother
- New Products
- Tech Tip: Really Simple HTTP Server with Python
- Poul-Henning Kamp: welcome to
1 hour 38 min ago
- This has already been done
1 hour 39 min ago
- Reply to comment | Linux Journal
2 hours 24 min ago
- Welcome to 1998
3 hours 12 min ago
- notifier shortcomings
3 hours 36 min ago
5 hours 13 min ago
- Android User
5 hours 15 min ago
- Reply to comment | Linux Journal
7 hours 8 min ago
9 hours 57 min ago
- This is a good post. This
15 hours 10 min ago
Free Webinar: Hadoop
How to Build an Optimal Hadoop Cluster to Store and Maintain Unlimited Amounts of Data Using Microservers
Realizing the promise of Apache® Hadoop® requires the effective deployment of compute, memory, storage and networking to achieve optimal results. With its flexibility and multitude of options, it is easy to over or under provision the server infrastructure, resulting in poor performance and high TCO. Join us for an in depth, technical discussion with industry experts from leading Hadoop and server companies who will provide insights into the key considerations for designing and deploying an optimal Hadoop cluster.
Some of key questions to be discussed are:
- What is the “typical” Hadoop cluster and what should be installed on the different machine types?
- Why should you consider the typical workload patterns when making your hardware decisions?
- Are all microservers created equal for Hadoop deployments?
- How do I plan for expansion if I require more compute, memory, storage or networking?