On Mon, Oct 01, 2007 at 07:04:27PM +0200, Thomas Klausner wrote:
> On Mon, Oct 01, 2007 at 07:04:50AM -0600, Eric Haszlakiewicz wrote:
> > Thomas, what shell are you using?  None of the ones I tried produces 
> > output like you have.
> 
> I tried it with bash. I can reproduce it with "ls -laR /var >
> /dev/null", like you suggested.

I noticed this on netbsd-4, while testing out a quad CPU box:

ksh$ pwd
/l/netbsd/netbsd-4/release
ksh$ /usr/bin/time sh -c 'for i in $(find . -name *gz); do gzip -t $i & done; wait'
    7.40s real    25.02s user     2.75s system
ksh$ /usr/bin/time sh -c 'for i in $(find . -name *gz); do gzip -t $i & done; wait'
    7.33s real    26.25s user     1.92s system
ksh$ /usr/bin/time sh -c 'for i in $(find . -name *gz); do gzip -t $i & done; wait'  
    7.46s real    25.36s user     1.79s system
ksh$ /usr/bin/time ksh -c 'for i in $(find . -name *gz); do gzip -t $i & done; wait'
    7.24s real    54.92s user     5.74s system
ksh$ /usr/bin/time ksh -c 'for i in $(find . -name *gz); do gzip -t $i & done; wait'
    7.12s real    91.19s user    11.68s system
ksh$ /usr/bin/time ksh -c 'for i in $(find . -name *gz); do gzip -t $i & done; wait'
    7.31s real    44.51s user     7.61s system
ksh$ /usr/bin/time bash -c 'for i in $(find . -name *gz); do gzip -t $i & done; wait'
    7.95s real    53.45s user    14.49s system
ksh$ /usr/bin/time bash -c 'for i in $(find . -name *gz); do gzip -t $i & done; wait'
    7.23s real    46.67s user     6.55s system
ksh$ /usr/bin/time bash -c 'for i in $(find . -name *gz); do gzip -t $i & done; wait'
    7.53s real    77.91s user    18.84s system

So, something strange is happening, and sh seems fairly immune,
while ksh and bash both get erratic.

-- 
Paul Ripke