home *** CD-ROM | disk | FTP | other *** search
- Path: comma.rhein.de!serpens!not-for-mail
- From: mlelstv@serpens.rhein.de (Michael van Elst)
- Newsgroups: comp.sys.amiga.programmer
- Subject: Re: CHIP RAM speed test results
- Date: 22 Mar 1996 14:37:42 +0100
- Organization: dis-
- Message-ID: <4iuaf6$3rl@serpens.rhein.de>
- References: <38232981@kone.fipnet.fi> <4ico8k$2vau@columba.udac.uu.se> <42116651@phoenix.owl.de> <4iogfh$e6i@serpens.rhein.de> <4iss4q$2762@columba.udac.uu.se>
- NNTP-Posting-Host: serpens.rhein.de
-
- m93kma@sabik.tdb.uu.se (Kristofer Maad) writes:
-
- >Michael van Elst (mlelstv@serpens.rhein.de) wrote:
- >: "Frank Wille" <frank@phoenix.owl.de> writes:
- >: >So MOVE16 is 50% faster, although the A3640 doesn't support burst
- >: >transfers.
-
- >: Sure. The MOVE16 instruction bypasses the cache.
-
- >Why not simply turn off the data cache during copy, then? Could save
- >some cycles.
-
- I don't see how some extra operation can save cycles.
-
- You can only turn off the data cache completely or by manipulating
- the MMU tables on a page basis. Since you do not want to disable the
- data cache completely this is some significant overhead.
-
- Another problem is that this is multitasking unfriendly. You absolutely
- must disable interrupts, otherwise the cache may contain stale data
- and you have to flush it. And this is even more costly than manipulating
- the MMU tables.
-
- You see, the move16 instruction is there for a reason.
-
- --
- Michael van Elst
-
- Internet: mlelstv@serpens.rhein.de
- "A potential Snark may lurk in every tree."
-