Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!cs.utexas.edu!rutgers!rochester!pt.cs.cmu.edu!wb1.cs.cmu.edu!avie
From: avie@wb1.cs.cmu.edu (Avadis Tevanian)
Newsgroups: comp.sys.next
Subject: Re: nlist initialization?
Message-ID: <5133@pt.cs.cmu.edu>
Date: 6 Jun 89 08:12:52 GMT
References: <1499@uw-entropy.ms.washington.edu> <10985@orstcs.CS.ORST.EDU>
Distribution: usa
Organization: NeXT, Inc.
Lines: 25

In article <10985@orstcs.CS.ORST.EDU> johnsot@mist.CS.ORST.EDU (Tim G. Johnson) writes:
>but when I did a lseek into /dev/kmem, I never got the load average.
>If you figure actually get a valid reading, let me know. Thanks.
>(BTW, try looking up _mach_factor too. I dunno what it is really, though.
>/usr/ucb/w will show the mach factor with the -m flag.)
>

Mach doesn't use any floating point internally, so it stores the load average
as a scaled integer, with a scale factor of 1000 (LSCALE in sys/kernel.h).

The Mach Factor is an interesting number I made up a couple of years ago.
It is intended to take into account multiprocessors, and is a measure of
the expected number of processors that would be available to your application
if you were to run it.  For example, if there was one compute bound thread
in a single cpu system, the Mach Factor would be .5 (expect 1/2 of 1 cpu).
If you had 1 thread running on 2 processor system, you could expect to get
1 whole cpu, thus a Mach Factor of 1, if there were 2 threads on that same
system, you could expect to get 2/3rd of a cpu (3 threads would share 2
processors if you launched a new thread), hence a Mach Factor of 0.6666.

-- 
Avadis Tevanian, Jr.    (Avie)
Manager, System Software Group / Chief Operating System Scientist
NeXT, Inc.
avie@cs.cmu.edu or avie@NeXT.com
--