> Hi,
>
> It was mentioned in the SOSP paper that in order to run the OSDB benchmark
> on UML, a small modification was made to the OSDB benchmark. I can't seem
> to find a patch or documentation on what that change was.
The xendemo CD contains the source for the modification
to the osdb benchmark in /usr/local/src/bm.
The major changes in the osdb.c file are to enable us to run
multiple copies of osdb at the same time, repeating just the
multi-user IR and multi-user OLTP phases which are the key
benchmark results. The significant change from the point of view
of running the benchmark over UML is the transformation of the
benchmark's timing code to use "gettimeofday" rather than
"times". The latter returns completely bogus results on UML
because a) the kernel HZ constant is set differently from that
reported by "sysconf", and b) under load, UML looses 'interrupts'
all over the place and hence the time (as reported by e.g. times,
sleep and select) runs slower than real time (making UML look
a lot better than it really is).
Best,
Ian
-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive? Does it
help you create better code? SHARE THE LOVE, and help us help
YOU! Click Here: http://sourceforge.net/donate/
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxxx
https://lists.sourceforge.net/lists/listinfo/xen-devel