Answer: pcfs writes just one sector at a time.
This is a known problem, and will be fixed this summer.
Thanks, all!
- Dan Kegel
Replies follow.
Date: Sat, 23 Jan 93 19:50:46 +0100
From: per@erix.ericsson.se (Per Hedeland)
Message-Id: <9301231850.AA28876@super.eua.ericsson.se>
To: dank@blacks.jpl.nasa.gov
Subject: Re: Why are floppies in SS10 so darn slow?
Newsgroups: comp.sys.sun.hardware
In-Reply-To: <saatchi.727734615@bacchus.jpl.nasa.gov>
In article <saatchi.727734615@bacchus.jpl.nasa.gov> you write:
>How come the floppies in our workstations are so slow?
>They seem about 10x slower than the same floppies on MS-DOS machines.
How are you accessing them? While I haven't done any comparisons btwn
the speeds, it is obvious that Sun's 'pcfs' code is awfully slow (for
some reason unknown to me). If you're currently mounting the floppies as
'pcfs' file systems, try out the freely available 'mtools' package
instead - it doesn't make the floppy a part of the file system, and the
commands it provides are messy-dossy, but it is *much* faster and
doesn't require root priviliges to use.
Hope this helps...
--Per Hedeland
per@erix.ericsson.se or
per%erix.ericsson.se@sunic.sunet.se or
...uunet!erix.ericsson.se!per
Date: Sat, 23 Jan 1993 19:47:09 -0600
From: Steve Hanson <hanson@pogo.fnal.gov>
Message-Id: <199301240147.AA18024@pogo.fnal.gov>
To: dank@blacks.jpl.nasa.gov
Subject: Re: Why are floppies in SS10 so darn slow?
Newsgroups: comp.sys.sun.hardware
References: <saatchi.727734615@bacchus.jpl.nasa.gov>
Reply-To: hanson@pogo.fnal.gov
Organization: FERMILAB, Batavia, IL
In comp.sys.sun.hardware you write:
>How come the floppies in our workstations are so slow?
>They seem about 10x slower than the same floppies on MS-DOS machines.
>- Dan K.
Slow at what? you mean writing ms-dos floppies with the PCFS file
system? Or what exactly? If that IS what you mean - PCFS is real slow.
You need to do something faster, like compiling the mtools
package and running it on your Sun.
Other than PCFS, I have never thought the floppies on the Suns were
slow at all.
-- Steve Hanson - FERMILAB, Batavia, Il. hanson@pogo.fnal.gov or hanson@fnal.fnal.govDate: Sat, 23 Jan 93 18:06:00 PST From: David.Robinson@Eng.Sun.COM (David Robinson) Message-Id: <9301240206.AA07166@jetsun.Eng.Sun.COM> To: dank@blacks.jpl.nasa.gov Subject: Re: Why are floppies in SS10 so darn slow? Newsgroups: comp.sys.sun.hardware In-Reply-To: <saatchi.727734615@bacchus.jpl.nasa.gov> Organization: Sun Microsystems, Mountain View, CA USA Cc: Content-Length: 333
In article <saatchi.727734615@bacchus.jpl.nasa.gov> you write: >How come the floppies in our workstations are so slow? >They seem about 10x slower than the same floppies on MS-DOS machines. >- Dan K.
Simple, we read and write one sector at a time in the pcfs, when you do that you burn a lot of rotations.
Fixed in 2.2.
-David
From: Dominic Dunlop <dominic@natcorp.ox.ac.uk> Date: Mon, 25 Jan 93 09:57:02 GMT Message-Id: <963.9301250957@onions.natcorp.ox.ac.uk> To: dank@blacks.jpl.nasa.gov Subject: Re: Why are floppies in SS10 so darn slow? Newsgroups: comp.sys.sun.hardware In-Reply-To: <saatchi.727734615@bacchus.jpl.nasa.gov> Organization: British National Corpus, Oxford University, GB Cc:
Daniel,
In article <saatchi.727734615@bacchus.jpl.nasa.gov> you write: > How come the floppies in our workstations are so slow? > They seem about 10x slower than the same floppies on MS-DOS machines.
For what it's worth, this came up in a developers' workshop session at the recent UK Sun User's Group. The Sun engineers present said this was a known bug, and should be fixed by summer. If you have a software maintenance contract, bug the hot-line about it. You should get put on a list of people who want the patch when it appears. -- Dominic Dunlop
Date: Mon, 25 Jan 93 15:57:57 -0500 From: pbg@cs.brown.edu (Peter Galvin) Message-Id: <9301252057.AA07023@bob.cs.brown.edu> To: dank@blacks.jpl.nasa.gov (Daniel R. Kegel) In-Reply-To: dank@blacks.jpl.nasa.gov's message of 22 Jan 93 20:30:15 GMT Subject: Why are floppies in SS10 so darn slow?
Are you using the "mtools" or mounting the disks? Mounting them is very slow, but the public-domain mtools are quite good and fast. Here is the info I have on them:
=================================== mtools ==================================== ======================== MSDOS floppy read/write tools ======================== ORIGIN viktor@melon.princeton.edu (Viktor Dukhovni) OWNERS rv ARCHS sun FILES: /cs/bin/mcopy /cs/bin/mren /cs/man/man1/mmd.1 /cs/bin/mdel /cs/bin/mtype /cs/man/man1/mrd.1 /cs/bin/mdir /cs/bin/mwrite /cs/man/man1/mread.1 /cs/bin/mkdfs /cs/man/man1/mcopy.1 /cs/man/man1/mren.1 /cs/bin/mmd /cs/man/man1/mdel.1 /cs/man/man1/mtype.1 /cs/bin/mrd /cs/man/man1/mdir.1 /cs/man/man1/mwrite.1 /cs/bin/mread /cs/man/man1/mkdfs.1
Date: Mon, 25 Jan 93 16:52:48 MST From: sacks@panther.mot.com (Dennis Sacks) Message-Id: <9301252352.AA02156@netmgr.panther.mot.com> To: dank@blacks.jpl.nasa.gov (Daniel R. Kegel) Subject: Re: Why are floppies in SS10 so darn slow? Newsgroups: comp.sys.sun.hardware Organization: Info Enterprises X-Newsreader: TIN [version 1.1 PL8]
You don't mention how you are accessing the floppies! the mount /pcfs is slow on any Sun I've seen. I NEVER use it. If I want to access MS-DOG floppies, I use the mtools package which is faster.
If you don't have mtools, it is worth having: basically it gives you mcd, mformat, mcopy, mattr, etc, etc, etc to deal with DOS floppies...
In article <saatchi.727734615@bacchus.jpl.nasa.gov> you wrote: : How come the floppies in our workstations are so slow? : They seem about 10x slower than the same floppies on MS-DOS machines. : - Dan K.
-- Dennis Sacks I'm just a simple cable is sacks@panther.mot.com Your modem ways lighten at Info Enterprises and confirm me.
26 Jan 93 13:49:54 NZD
Date: Tue, 26 Jan 93 13:49:54 NZD From: adm@phys.canterbury.ac.nz (Andrew McGregor University of Canterbury) Subject: Re: Why are floppies in SS10 so darn slow? To: dank@blacks.jpl.nasa.gov Message-Id: <9301260049.AA28760@phys.canterbury.ac.nz> X-Envelope-To: dank@blacks.jpl.nasa.gov Content-Transfer-Encoding: 7BIT
Use track length reads, eg: tar cvfb /dev/rfd0c 36 . (36 blocks = 18k = 1 track)
Andrew McGregor, Physics, University of Canterbury, Christchurch, New Zealand. adm@phys.canterbury.ac.nz
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:07:26 CDT