Re: What the heck does Transmeta do???

Kragen Sitaker (kragen@pobox.com)
Wed, 23 Dec 1998 17:58:05 -0500 (EST)


On Wed, 23 Dec 1998, Joseph S. Barrera III wrote:
> Speaking of Transmeta... is Henry Massalin still there? Is anyone on this
> list friends with him?

I exchanged email with him once, but that's it. I'm surprised to hear
he was at Transmeta at all.

> I loved this thesis work (Synthesis)

Me too. By the way, it appears that the web page his thesis was
previously on is no longer available, but I think I might have it lying
around on disk in PostScript form. Does anyone want a copy?

Synthesis included a bunch of extremely innovative things; even one of
them would have made a good thesis.
- very efficient run-time code synthesis (instead of attaching a vtable
of pointers to functions, you would attach a table of code)
- adaptive scheduling based on I/O streams, so you can get
real-time-like performance on things like audio, without the
inflexibility of typical real-time methods.
- lock-free synchronization -- the kernel ran on a dual-CPU machine,
but used no blocking synchronization techniques at all in order to
avoid deadlock. Instead, it used compare-and-swap exclusively.

Does anyone know what kind of followup work has been done? Surely
enough people were blown away by his work that there should be a few
Synthesis copycats, right?

-- 
<kragen@pobox.com>       Kragen Sitaker     <http://www.pobox.com/~kragen/>
[around 1998-12-23], it is amazing to watch fear and loathing and greed at
play with the more speculative Internet stocks.  To call this a tulip
craze would be a vast understatement. -- Adam Rifkin, <adam@cs.caltech.edu>