???? Hardware-accellerated python ????

[Alex Martelli]

"Steven Adams" <adams_s at lab.eng.usyd.edu.au> wrote in message
news:8mrpqu$bik$1 at news1.wire.net.au...
> what about using FPGA's? Taht would alleviate some of the production time
> lag between design and time to market - with the benefit that as newer,
> faster designs came out you could 'refit' your 'processor' in the machine.
>
> DISCLAIMER: I only know a little about FPGA's (i.e. they exist and general
> idea of their use).
>
> would this work?

Generally, the price/performance ratio of various sorts of compromise
technologies (gate arrays, either field-programmable or not, &tc &tc)
versus fully-custom designs is not all that hot (given the rapid design
technologies there are today -- warning: it's almost 20 years since I
last designed an integrated circuit, and though I've done what I could
to follow the trends since, I'm obviously a semi-detached observer!-).

Of course, there are all sorts of niches, but they tend to be in the kind
of fields for which you can't really find a packaged solution -- roughly,
IC's characterized by specific-roles, NOT specialized-clones of CPU's,
RAM, etc.  I see no reason on the technology horizon why a CPU that is
specially-designed (or even just specially-microprogrammed) to
execute some particular 'semantically rich bytecode' would have any
better success today than any of the umpteen similar attempts in the
past (such as Intel's greatest flop ever, the iApx 432 -- remember?).


Alex