home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!mcsun!uknet!gdt!uwe-bristol!csdsun!r_voisey
- From: r_voisey@beans.uwe.ac.uk (R Voisey)
- Newsgroups: comp.sys.acorn
- Subject: Re: 36Mhz ARM 3
- Message-ID: <1992Nov16.101404.3426@csd.uwe.ac.uk>
- Date: 16 Nov 92 10:14:04 GMT
- References: <1992Nov14.224813.17259@mnemosyne.cs.du.edu>
- Sender: usenet@csd.uwe.ac.uk (uwe nntp usenet poster)
- Reply-To: r_voisey@csd.bristol-poly.ac.uk
- Organization: The University of the West of England (The Magic Roundabout)
- Lines: 25
- Nntp-Posting-Host: usenet@csd.uwe.ac.uk (uwe nntp usenet poster)
-
- In article <1992Nov14.224813.17259@mnemosyne.cs.du.edu>, mhussain@nyx.cs.du.edu (manar hussain) writes:
-
- |> Apparently the choice of around about 24 Mhz for the clock speed of the
- |> ARM 3 is surprisingly enough more than just a mere coincidence.
- |> Essentially what is really important is how quickly in practice any
- |> application a user wishes to use actually runs. In fact beyond a certain
- |> level - about 25Mhz when considering what is in an A5000 apart from an AR
- |> 3 !! - any increase in clock speed will have a minimal effect on apparent
- |> speed. Obvious exceptions would be particularly processor intensive apps.
-
- Of course, it had nothing to do with VLSI swapping the ARM onto a less
- advanced (ie thicker silicon) production rig to free up their good stuff
- for military applications..
-
- Of course, you are correct when you say that the performance increase over
- clock speed is not linear (due, largely, to bus limitations). However, I
- seem to recall Aleph One claiming that the optimum price/performance for an
- ARM-3 on an 8Mhz bus (this was a while ago) was around 36Mhz.
-
- Bob
-
- --
- Robert J. Voisey : r_voisey@csd.uwe.ac.uk
- -- Probug Software Developments --
- 162 Staplehill Rd, Fishponds, Bristol, Avon, England
-