Messages in this thread | | | Date | Wed, 20 Jan 1999 22:33:39 +0100 (CET) | From | "Szabo, Balazs" <> | Subject | Re: Kernelcompile-test results |
| |
|| || So is going worse... I think to know why. || || > 2.2.0 pre5 arca mini patch: || > make zImage 830.03s user 110.95s system 89% cpu 17:30.46 total || || Unfortunately since pre6 if I remeber well it's been avoided to swap if it || was going to be allocated a buffer to avoid a buffer/swap deadlock. So in || > pre5 kernel we get a default penalization... But pre5+mini-patch it's || very interesting anyway, just to let you know why new kernels are || per-default slower ;).
I don't know this is a good test or not, that's why I figured out a new test. I make heavy load with for example kernel compiling with MAKE=make -j, and paralelly starting processes like this:
tick.sh: #!/bin/sh echo "#" >>$1.out sleep $1 exec tich.sh $1
I should start it ifor example with the parameter 1, 2, 4, 8, 16, 32, 64 After the compilation finishes, we have to look the size of these files. The greater the parameter is, the size difference must be smaller in different situations and swapping models. But We should look at the difference in the parameter 1, 2, 4, or maybe 8. This could be a good test for interactive speed. That kernel is better, which produces longer files.
How about this idea? I will try it soon!
dLux -- =< Slay not he that cannot hear, be thankful ye that hath an ear >=
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |