Looking for a history of Transmeta: not just the chip design (Crusoe & Efficeon) but also how they got there, what the original ideas were and what drove them towards VLIW.
I am not looking for a rant about the failures of VLIW but an understanding of the design decisions made at the time and why they thought they were correct.
@bhtooefr that is what I was wondering but, unless I remember wrongly, a substantial part of the Moscow Center was eventually employed by Intel so Transmeta as an offshoot sounded wrong.
@cynicalsecurity @bhtooefr My understanding was that Transmeta's problems weren't really technical in nature, but that they were trying to compete with Intel itself and a bunch of other chip companies with their own fabs, so there wasn't really room for an IP-only company in the x86 space. Is that wrong?
@seanl @cynicalsecurity I'd argue that it's perfectly fine to be IP-only in the x86 space - VIA pulled it off, after all, and AMD is doing it now.
Transmeta's problems were really two-fold.
There was a huge gap in the x86 market of low power, moderate performance - Intel had nothing, AMD and VIA had low power, low performance parts. However, once demand became apparent, Intel could push down into that space.
And, Intel basically paid Transmeta to leave the market.
@cynicalsecurity @seanl As I understand it, Transmeta sued Intel for violating power management patents (a proven strategy - VIA sued Intel for violating Centaur's power management patents almost immediately after buying Centaur, to extract an x86 and chipset license from Intel), Intel countersued for x86 patent violations.
Then, the settlement basically involved Intel paying Transmeta, and Transmeta agreeing to stop making x86 processors.
@seanl @bhtooefr to be perfectly honest I am rather more interested on its origin and the thinking around the design decisions. I have vague recollections of Transmeta’s
end as it attempted to transform itself into a pure IP company (like Arm, of course).
Transmeta did have several low-power design wins: I played with a Fujitsu and a Compaq back in the days. Battery life compared very favourably to Intel-based designs.
@cynicalsecurity My understanding is that it was an offshoot of the work that Moscow Center for SPARC Technologies was doing with the Elbrus VLIW architectures - that might be a path to look at.