Newsgroups: comp.lang.c
Path: utzoo!henry
From: henry@utzoo.uucp (Henry Spencer)
Subject: Re: Memory Allocation (was Re: binary data files)
Message-ID: <1989May5.171923.20607@utzoo.uucp>
Organization: U of Toronto Zoology
References: <10946@bloom-beacon.MIT.EDU> <12546@ut-emx.UUCP> <10202@smoke.BRL.MIL> <11486@ulysses.homer.nj.att.com>
Date: Fri, 5 May 89 17:19:23 GMT

In article <11486@ulysses.homer.nj.att.com> kpv@ulysses.homer.nj.att.com (Phong Vo[drew]) writes:
>... Worse than that
>is wasted time due to the number of page faults when data are accessed frequently.
>This is a subtlety that is frequently missed when people run tests on malloc.

If your system is properly designed, the effect is not to run up the page
fault rate, but to increase the demand for physical memory.  (If the
system is not properly designed, or is short of physical memory, you have
problems that malloc cannot fix.)  Frequently-accessed data should be in
physical memory and hence should not fault.
-- 
Mars in 1980s:  USSR, 2 tries, |     Henry Spencer at U of Toronto Zoology
2 failures; USA, 0 tries.      | uunet!attcan!utzoo!henry henry@zoo.toronto.edu