[Gllug] Kernel compile taking forever...!

[Nix]

On Wed, 02 Feb 2005, Tethys announced authoritatively:
> Andrew Halliwell writes:
>>Opteron's a different kettle of fish though.
>>Any penalty you get from increased load times should be more than negated by
>>the fact you've to a boatload of new registers, which of course, should
>>reduce memory requests and thus, increase speed.
> 
> I don't know how much you know about the UltraSPARC, but shortage of
> registers is not one of its failings...

It's better than that. Thanks to software management of the register
window overflow/underflow exceptions, you could theoretically fill the
entirety of accessible storage with `stack' (with the software spilling
them to RAM, disk, the network) with the code outside the part of the OS
that's handling the overflow/underflow exceptions needing no changes
(there's even a backup register window for the overflow/underflow
exception handlers to use. *They*'d have to put up with the terrible
limitations of a finite stack, though.)

No OS on SPARC goes to this extreme of course because it would be
insane. But it's possible. :)

-- 
`Blish is clearly in love with language. Unfortunately,
 language dislikes him intensely.' --- Russ Allbery
-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug