Name of package: cernlib-2006
1st run with value of -j 3 on china
real 27m9.709s
user 21m46.437s
sys 5m0.554s
2nd run with distcc on ireland(client) and scotland(server) with a value -j 20
real 26m57.953s
user 18m13.082s
sys 6m34.777s
3rd run with distcc on ireland(client) and scotland(server) with a value -j 8 on fc10
real 27m34.154s
user 18m12.926s
sys 6m33.681s
4th run with distcc on ireland(client) and scotland(server) with a value -j 10 on fc10
real 27m36.767s
user 18m12.984s
sys 6m32.915s
After Serval test runs i started to wonder if distrubiting jobs over distccd was taking an effect on time so i decided to do test run on my home pc.
5th run on my home computer with a i7 quad proccessor with 3gb ddr3 memory with a value -j3
real 18m51.978s
user 15m22.139s
sys 3m42.258s
6th run on my home computer with a i7 quad proccessor with 3gb ddr3 memory with a value -j10
real 17m47.881s
user 14m19.851s
sys 3m48.304s
After Two more test i realized that -j value was not having much of an effect so i started looking at other things and i noticed that ireland was using fedora 9, and the test on my home pc was a bit faster so i wondered if the operating system had an effect, again i started experimenting with a fedora core machine in my case i choose austraila.
7th run on Austrailia (client) scotland(server) -j3
real 26m5.646s
user 20m1.189s
sys 6m31.448s
8th run on Austrailia (client) scotland(server) -j10 between australia and scotland
real 25m56.564s
user 18m35.217s
sys 6m25.473s
After more examination i discoverd the operating system also had very little effect on the ammount of time it took to build the rpm
Wednesday, March 18, 2009
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment