Path: utzoo!mnetor!uunet!cbmvax!daveh
From: daveh@cbmvax.UUCP (Dave Haynie)
Newsgroups: comp.sys.amiga
Subject: Re: 68030 Questions
Message-ID: <3269@cbmvax.UUCP>
Date: 4 Feb 88 23:05:28 GMT
References: <1431@sugar.UUCP>
Organization: Commodore Technology, West Chester, PA
Lines: 27

in article <1431@sugar.UUCP>, peter@sugar.UUCP (Peter da Silva) says:
> 
> In article <3200@cbmvax.UUCP>, daveh@cbmvax.UUCP (Dave Haynie) writes:
>> in article <494@.UUCP>, alex@.UUCP (Alex Laney) says:
>> > you can't use any of the features of it. You know, turning off caches, etc.,
>> > seems to be mandatory with the '020. 

> Well, I'd think that you'd probably want to invalidate the cache when you
> LoadSeg() something... just in case it's LoadSegging it at the same address
> as something that's already in the cache. It's a real long-shot, but it's
> almost certain it's gonna hit someone sometime.

True.  Unless you can be certain that the LoadSeg function itself fills up
the cache.  That would be the simplest way to handle this without having to
make the 68020 a special case.  Certainly in the future, when data caches
and larger instruction caches are being used, the cache will have to be
explicitly dumped in such cases.  I'm not sure that LoadSeg actually does
this or not, but the cache is only 64 longwords long; it doesn't take long
to overrun this.  So I expect that implicit cache clearing works in this
case.  Any OS gurus out there know fer shure?

> -- Peter da Silva  `-_-'  ...!hoptoad!academ!uhnix1!sugar!peter
> -- Disclaimer: These U aren't mere opinions... these are *values*.
-- 
Dave Haynie  "The B2000 Guy"     Commodore-Amiga  "The Crew That Never Rests"
   {ihnp4|uunet|rutgers}!cbmvax!daveh      PLINK: D-DAVE H     BIX: hazy
		"I can't relax, 'cause I'm a Boinger!"