parallel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Benchmark of: -j0


From: Hans Schou
Subject: Benchmark of: -j0
Date: Sun, 12 Feb 2017 17:48:59 +0100

Hi Ole

Should one use -j default or -j0 ?

Is there a benchmark test which shows that -j0 is always faster than default?

Acording to $(grep -c proc /proc/cpuinfo) I have 6 cores. This should imply that my default is -j6.

$ for N in $(seq -w 1000) ; do cp /big.html $N.html ; done
$ time ls -U *.html | parallel --joblog jdefault.log gzip
real    1m2.342s
user    1m24.616s
sys     0m4.352s
$ rm *.gz
$ for N in $(seq -w 1000) ; do cp /big.html $N.html ; done
$ time ls -U *.html | parallel --joblog j0.log -j0 gzip
parallel: Warning: Only enough file handles to run 252 jobs in parallel.
parallel: Warning: Running 'parallel -j0 -N 252 --pipe parallel -j0' or
parallel: Warning: raising ulimit -n or /etc/security/limits.conf may help.
real    1m23.760s
user    1m44.184s
sys     0m4.692s

I read this as the warning tells me if I tweak my system I can have more open files but by default an arbitrary system would perform better with -j default (in my case 6) than -j0.

Am I doing something wrong?

./hans

reply via email to

[Prev in Thread] Current Thread [Next in Thread]