Return to BSD News archive
Path: euryale.cc.adfa.oz.au!newshost.carno.net.au!harbinger.cc.monash.edu.au!munnari.OZ.AU!news.ecn.uoknor.edu!news.cis.okstate.edu!newsfeed.ksu.ksu.edu!news.mid.net!mr.net!news.sgi.com!news.msfc.nasa.gov!newsfeed.internetmci.com!news.ac.net!news.serv.net!news.serv.net!michaelv
From: michaelv@MindBender.serv.net (Michael L. VanLoon)
Newsgroups: comp.unix.bsd.freebsd.misc
Subject: Re: dual cpu stuff...
Date: 26 Aug 1996 07:10:22 GMT
Organization: Michael L. VanLoon
Lines: 80
Message-ID: <MICHAELV.96Aug26001022@MindBender.serv.net>
References: <4vcsn4$7ql@cantina.clinet.fi> <4veq47$cc@anorak.coverform.lan>
NNTP-Posting-Host: mindbender.serv.net
In-reply-to: brian@awfulhak.demon.co.uk's message of 21 Aug 1996 11:58:15 +0100
In article <4veq47$cc@anorak.coverform.lan> brian@awfulhak.demon.co.uk (Brian Somers) writes:
Mika Ruohotie (mickey@cantina.clinet.fi) wrote:
: i'm about to get a dual cpu pentium board in few weeks time, and realized
: the other day it has only 512 pipeline cache, do i get slower performance
: when i put two p166/p200 on the board?
I'd get a dual PPro board - no problem with the cache then!
It should never be slower than a single P166; put it that way.
: i remember reading that over p133 the cpus prefer 512kb cache _each_,
: ofcourse, i was unable to figure out is 512 still enough since the
: cpu's are "interlaced", right?
"Interlaced"? I don't think that's a word that has ever been applied
to SMP... ;-)
They "share" the bus. Each runs out of its internal cache when it
can, and then "arbitrates" for the bus, in basically the same way a
PCI SCSI bus-mastering controller arbitrates for the bus. The bus
interface determines how long each CPU can burst on the bus before it
has to get off and let the other CPU use it. Consult
http://www.Intel.com/ if you want more info.
: anyway, i start with one p133, i think...
: how well the current multiple cpu freebsd can take advantage from the
: multiple cpus? i heard that the new nt4 will only hand max 40%
: advantage...
Well then you heard wrong. There is not one magical number that it
maxes out at. It depends on a million little things. What kind of
applications are running, the number of threads, the speed of your
RAM, the amount of cache, the speed of the processors relative to bus
speed, the way the applications are written that you run most, etc....
Some of these mixes may give you a 140% efficiency with two CPUs, some
may give you 180%. It just depends...
NT is abismal when it comes to SMP... it's not worth comparing with
anything. OS/2 is reported to run stuff faster on say a single P133
(no SMP support) than NT on a dual P133 !
That is complete misinformation perpetuated by Sun in their on-going
smear campaign to try and one-up Microsoft at hostile marketing.
And if you believe that OS/2 can run stuff faster on a single P133
than NT on a dual, I have some motherboards I'd like to sell you. Did
you know "gullible" isn't in the dictionary?
FreeBSDs performance (as far as I know, but I'm not talking
"authoritively") may give close to 100% improvement - but it
really depends on your application. If you've got a lot of user-
level cpu intensive code you'll get max benefit. If you've got
lots of kernel-level code, the benefit will be reduced.
Actually, FreeBSD's implementation is very inefficient. It's very
early alpha-quality code. This isn't to put down the people working
on it -- I'll bet they'd say the same thing. There are some really
sharp people in that mix. It's just that it takes time to get all the
right pieces in place for a truly well-tuned SMP kernel, and FreeBSD
is just getting started.
: how about the cache? i know my compiling time went down 24% on single
: cpu... from 256 async to 512 pb, with p133...
The PPro has gotta be better !
Well, of course it is... What does that have to do with dual Pentium
motherboards?
--
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Michael L. VanLoon michaelv@MindBender.serv.net
--< Free your mind and your machine -- NetBSD free un*x >--
NetBSD working ports: 386+PC, Mac 68k, Amiga, Atari 68k, HP300, Sun3,
Sun4/4c/4m, DEC MIPS, DEC Alpha, PC532, VAX, MVME68k, arm32...
NetBSD ports in progress: PICA, others...
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -