Some more data as suggested. I run without nice, with nice -n 5 and nice -n
10, each version twice:
Without nice:
real0m10.662s
user0m9.513s
sys 0m3.448s
real0m10.121s
user0m7.712s
sys 0m2.444s
CPU% according to top reached 118%.
nice -n 5:
real0m22.141s
user0m27.202s
sys 0m1.444s
real0m21.855s
user0m28.674s
sys 0m2.468s
CPU% according to top reached 147%.
nice -n 10:
real0m21.638s
user0m26.878s
sys 0m1.268s
real0m21.654s
user0m27.090s
sys 0m1.176s
CPU% according to top reached 131%.
I do not trust the accuracy of top; I only use it to see relative CPU usage
of processes running *at the same time*. The above results speak for
themselves.
However the cpu usage reported by 'time' shows that nice incurs a hefty
penalty even though the system does nothing else.
I stress that I do not think that the above indicate a problem with nice.
This maybe just how it works. I think I need something other than nice to
tell the system that a process should be given all the spare cycles but
nothing else.
Any suggestions?
Matyas
-
Every hardware eventually breaks. Every software eventually works.
--
To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org