|
Transmeta was founded within 1995 by Dave Ditzel as a US-based corporation that designed VLIW code morphing microprocessors. Up to now, it has produced deuce x86 compatible CPU architectures: the Crusoe and Efficeon processors. These CPUs stand appeared within ultra-personal Laptops, Blade servers, Tablet PCs, & potentially a silent desktop, in which the moo power consumption and heat dissipation offer better benefit.
History
Transmeta marketed their micro-processor technology when inordinately innovative & radical in the online-power market section. It got hoped to become two power & performance leaders in the x86 space. Yet initial reviews of a Crusoe indicated the performance fell significantly short of projections. [http://www.vanshardware.com/articles/2003/07/030715_Transmeta/030715_Transmeta.htm] Moreover, the market did non have however in a period of Crusoe development -- Intel & AMD experienced significantly build up speeds & began to location increasing concerns just about power consumption -- so the ipo, Crusoe, was speedily cornered into a sale-volume, microscopic form factor (SFF), sale-power section of the market.
This forced a rapid re-project of the technology, marketed when the Efficeon processor. A Efficeon claimed to stand twice a performance of a original Crusoe CPU at the equivalent frequency. Nevertheless, a performance was however light relative to a competition, & the complexness of the chip experienced increased significantly. This increased size & power consumption can keep around diluted the key market benefit Transmeta's chips experienced antecedently enjoyed all over a competition.
Transmetthe has listed a total of industry leading light like Linus Torvalds and Dave Taylor. At a start, its purpose was saved secret, however part because it got such talent amongst its staff, the industry was constantly abuzz sustaining hearsay additionally to 'conspiracy theories' resulting around first-class press relations (PR).
Linus Torvalds left Transmeta in June 2003 to dedicate himself to the further development of the Linux kernel. the virtually all valuable plus held by Transmeta at a present instance is their patent portfolio, & rumour persist that a company will exist as bought by a big streaming video player like AMD or even Microsoft for this reason.
When of January 2005 the company announced the strategical restructuring out of existence a chip product company to an noetic property company. That is, instead of selling chips a company might okay, sell technology for have by more chip makers. Around February 2005, there was wild speculation that AMD can choose out Transmeta. Around March 2005 the company announced that it was laying off 68 population, allowing 208 employees. Astir half of a unexpended employees were to functiin on propagating the LongRun2 power optimisation technology inside Sony products. Sony was reported to become the key licencee of this Transmeta technology.
In May 31, 2005, Transmeta announced the sign language of plus acquire & license agreements by having Hong Kong's Culture.com Technology Limited (Culturecom; 文化傳信) led by Chu Bong-Foo, the artificer of the Cangjie method and one of the origination fathers of modern Chinese computing. This dealing is subject to the technology exportation license from either the U.S. Department of Commerce and some other required agreements.
Transmeta has agreed to sell its Crusoe business line to Culturecom. Culturecom might besides license Transmeta's Efficeon to produce & sell its products within China.
In August 10 2005, Transmeta announced its first ever profitable quarter.
Technology
A actual TM processors come within-sequentially VLIW cores. To execute x86 code, the pure programs-depending instruction translator dynamically compiles or emulates x86 code sequences, using execution-hotspot guided heuristics. When similar technologies existed (WABI for Sun, FX!32 for Alpha) in the early 90s, the TM approach has placed the lot higher bar for compatibility—respire to execute wholly x86 videos from either initial boot as much as a latest multimedia system videos—when retaining virtually all of its core performance.
Transmeta claims many technical indicator advantages to this approach:
As a market leaders Intel and/or AMD would extend the core x86 instruction placed, Transmetthe can quickly update their product by owning the package update like than requiring a respin of their devices.
Performance & power may be tuned within computer software to meet market needs
It would become the comparatively elementary interest fix devices project or even manufacturing flaws in the devices utilizing software system workarounds.
More instance can be spent concentrating in enhancing a capabilities of the core or even reducing its power consumption while forgoing worrying astir 16 years of backward compatibility to the x86 architecture.
a processor may emulate multiple more architectures, potentially even at the equivalent period. (At its initial Crusoe launch, Transmeta demonstrated pico-Java and x86 running intermixed on the native devices.)
Before Crusoe release, rumour indicated Transmetthe was relying in these advantages to prepare a interbred PowerPC and x86 processor. Notwithstanding, Transmeta would ab initio concentrate alone on the pleasantly on line-power x86 market.
Viability
Transmeta wasted good deal believability & endured important criticism due to the unfortunate initial Crusoe showing with big discrepancies between projections & actuals for each performance & power. On a one h&, the power figures showed a sensible improvement on top the Intel and AMD offerings. Nonetheless a prevent user own household budget (we.e. battery life) simply showed the marginal overall improvement. [http://www20.tomshardware.com/mobile/20010215/index.html] Number one, a Code Morphing Programs (CMS) concerted by using cache architecture by artificial means inflated comparisons between benchmarks & real-globe applications. This is due to the insistent nature & severity of benchmarks and their little footprints. the CMS software system overhead might own actually been a key drive of lot moo performance for numbers of real-globe applications; a elementary VLIW core architecture could not compete in computationally-intensive applications; & a southbridge interface was limited by its low bandwidth for even graphics or more I/O-intensive applications. A select few standard benchmarks potentially failed to start, questioning its claim of to the full x86 compatibility. [http://www.vanshardware.com/articles/2003/07/030715_Transmeta/030715_Transmeta.htm]
the Efficeon addressed several of the Crusoe defect & showed about a 2x real-globe improvement above Crusoe. Additionally, its die size was substantially little than either a Pentium Four or even a Pentium M, while likened in the equivalent run technology. Efficeon's die size withwithin 90nm is 68mm^2, which is 60% of the Pentium Four in 90nm, at 112 MM^2, using each processors possessing the 1MB L2 cache.
a notion of selling the product into a specific thermal envelope was occasionally non understood per mass of reviewers, world health organization tended to compare Efficeon against the gamut of x86 microprocessors, disregardless of power consumption or even application. 1 such lesson of this criticism suggests a performance however significantly lagged Intel's Pentium M (Banias) and AMD's Mobile Athlon XP. [http://www.vanshardware.com/reviews/2004/04/040405_efficeon/040405_efficeon.htm]
For the thermal envelope where Efficeon was designed to compete (7W & 12W), there are uncorroborated claims that its frequency far exceeded anything else in the market, at Unity.Cinque GHz & 7W, after a Centrino at a period can just work in a 7W envelope when its frequency was reduced to One.1GHz. This claim besides true considers single CPU frequency & doesn't assume more super important factor out overall performance, like core CPI (clocks per instructiin), or even memory performance & bandwidth, which keep close at h& variable impact on different benchmarks and technique configurations.
|