Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : Transmeta (TMTA)-The Monster That Could Slay Intel -- Ignore unavailable to you. Want to Upgrade?

To: Bill Fischofer who wrote (84)	1/22/2000 5:39:00 PM
From: Jonathan Edwards	Respond to of 421

First, Crusoe will enable Transmeta to "clone" any specialty microprocessor in software simply by providing an appropriately versioned code morphing engine To reiterate some of my previous posts, I simply don't believe this is true. TransMeta will be able to run applications written for any specialty microprocessor, perhaps, but they will not be able to mimic anything besides the X86 sufficiently closely that an existing design (or an existing design plus small tweaks) could simply replace the chip it was designed for with a TMxxxx. To do so would require changes to the TM hardware (to make it mimic the hardware it's replacing) or some amount of glue around an existing TM to disguise it appropriately. The latter is a particularly inelegant way of doing things, and if I were to pick one word to describe TM's technical accomplishments it would be "elegant"...

To: Bill Fischofer who wrote (84)	1/23/2000 2:39:00 AM
From: axp	Read Replies (1) \| Respond to of 421

Crusoe will enable Transmeta to "clone" any specialty microprocessor in software simply by providing an appropriately versioned code morphing engine Think about this a second. For mainline x86 chips, AMD and Intel are focusing exclusively on getting the instructions to execute faster. And in AMD's case there actually is a background RISC type engine that gets fed with a transformed x86 instruction stream. It's not reasonable to think that a generic code translating engine could ever out-perform a chip designed specifically to run the x86 instructions. There have been a number of forays into advanced emulation/translation schemes recently not the least of which was Digital's FX!32 - a pure software solution. At it's peak it achieved about 10 times the performance of straight "dumb" emulation by translating and optimizing x86 instructions to native the native Alpha RISC instructions. Even this ended up being about half the performance you'd get from compiling the source code directly to Alpha instructions. Further translation performance gains were impeded by the need to preserve the in-memory layout that the x86 instruction stream expected. Sooner or later you run into speed bumps related to system-wide limits such as cache, memory and bus latency. Speed of instruction execution is only useful if you can keep the cpu busy. The difficult of beating AMD an Intel at x86 performance is why the TM market is focusing on the handheld and network appliance market. This is where the low power gives them a huge advantage. Going head-to-head against x86 desktop systems with a generic "code-morphing" chip is a non-starter.