Path: utzoo!attcan!uunet!ncrlnk!ncrcae!ece-csc!mcnc!rutgers!apple!bionet!agate!web-1b.berkeley.edu!c60a-1cu
From: c60a-1cu@web-1b.berkeley.edu (Drew Dean)
Newsgroups: comp.arch
Subject: Re: Sun slowness
Message-ID: <16882@agate.BERKELEY.EDU>
Date: 12 Nov 88 02:39:17 GMT
References: <181648@<1988Oct28> <28200222@urbsdc> <360@auspex.UUCP> <52@sopwith.UUCP> <424@auspex.UUCP>
Sender: usenet@agate.BERKELEY.EDU
Organization: University of California, Berkeley
Lines: 24

In article <424@auspex.UUCP> guy@auspex.UUCP (Guy Harris) writes:
> >|s/the CPU/memory - most of the slowness I've seen on Suns has been due
> >|to paging.
> >
> >Yes, but what is the paging due to?
>
>Big software.  What did you expect - paging due to a slow CPU?
I/O bandwidth is the next workstation frontier...The Suns in the undergraduate
labs around here often end up being used in a timesharing mode, and when the
system gets busy (we have 60 Sun 3/50's on 4 fileservers), I've waited
7 minutes (and then given up), to get a login prompt from X Windows....Now
each 3/50 is (for integer operations) about as fast as a Vax 11/780 (note:
I said about...I mean anywhere from roughly .6 - 1.8 times as fast).  That's
still plenty of CPU -- but everything goes over the Ethernet...just not
enough bandwidth....If there was money to put a 40Mb SCSI drive on each 3/50
(for paging and common programs), I think that we'd see a large speedup when
the load gets up.....



Drew Dean
Internet: c60a-1cu@web.berkeley.edu
UUCP: ...!ucbvax!web!c60a-1cu
FROM Disclaimers IMPORT StandardDisclaimer;