home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!dtix!darwin.sura.net!gatech!ukma!rutgers!cbmvax!daveh
- From: daveh@cbmvax.commodore.com (Dave Haynie)
- Newsgroups: comp.sys.amiga.hardware
- Subject: Re: SetCPU
- Message-ID: <35099@cbmvax.commodore.com>
- Date: 14 Sep 92 16:59:10 GMT
- References: <8SEP199215283822@vx9000.weber.edu> <ABAIR.92Sep9002447@parsons.sps.mot.com> <PETERM.92Sep11173544@kea.am.dsir.govt.nz>
- Reply-To: daveh@cbmvax.commodore.com (Dave Haynie)
- Organization: Commodore, West Chester, PA
- Lines: 21
-
- In article <PETERM.92Sep11173544@kea.am.dsir.govt.nz> peterm@am.dsir.govt.nz (Peter McGavin) writes:
-
- >It seems surprising, then, that the OS does not enable data-burst by default.
- >However I suspect there is a reason why it doesn't, i.e, that data-burst
- >slows most programs down slightly.
-
- Wow, someone though before they posted! You win the cupie doll, that's
- absolutely right. Instruction burst is generally a guaranteed win -- the only
- time you don't fetch instructions in order is around jumps. So prefetching
- instructions usually wins. With data, the program's behavior may lead to
- burst acting either favorably or unfavorably. If the extra three longwords
- of a fetch tend to be used, it's a win, if they tend to be discarded, it's a
- loss.
-
- The software folks at the time (mainly Bryce Nesbitt I think) did some testing
- and found data burst to hurt more than help under AmigaOS. That's a general
- observation which is certainly program and OS dependent.
- --
- Dave Haynie Commodore-Amiga (Amiga 3000) "The Crew That Never Rests"
- {uunet|pyramid|rutgers}!cbmvax!daveh BIX: hazy
- "Work like a horse, drink like a fish" - Psychefunkapus
-