Re: life extension of old computers thanks to opensource Contiki

["Michael J. Mahon" <mjmahon@aol.com>]

Peter de Vroomen wrote:

> It's actually even worse. Most opcodes of modern RISC processors are
> tailored to some specific high-level language (of course, mostly C or C++).
> The high-level language compiler can optimize way better for these
> processors than a programmer can do by hand. 

Actually, there are very few *specific* adaptations to programming
languages in most RISC architectures.  The adaptations, if any, tend
to be to specific data representations--like IEEE floating-point, or
null-terminated strings, or, occasionally, to BCD data.

RISC architectures *should* be based on dynamic frequency, not on
a catalog of "nice features", and this eliminates most of the
traditional language-specific opcodes (since every language needs
load, store, and conditional branch ;-).

Many of the "decorations" that have shown up in the last ten years
are related more to simplifying optimization and implementations
in highly superscalar, deeply pipelined machines--like predicated
execution and branch hinting.

> Typically, IF more optimization
> is needed, the programmer needs to only optimize one or two simple routines.
> Optimizing the rest of the code by hand just doesn't weigh up against the
> time it costs.

This, of course, is generally true, independent of architecture or
language.

-michael

Music synthesis for 8-bit Apple II's!
Home page:  http://members.aol.com/MJMahon/

"The wastebasket is our most important design
tool--and it is seriously underused."