On Tue, 8 Sep 2009 11:13:29 +0200
Schwickerath Ulrich wrote:
> Hello, Arnau,
Hi Ulrich,
> I've seen this before. The benchmark should try to recompile the
> itself but sometimes it doesn't. If it does not, the reason could be
> that one of the sanity checks goes wrong. Maybe you are hit by this
> problem.
>
> Try to run this script on your box a couple of times:
[...]
> On one of our hardware types I found that the CPU time used was
> always accounted as system CPU rather than user CPU (same cpu as
> yours), while on a different system I found that sometimes it works,
> sometimes not. This behavior breaks one of the numerous sanity checks
> in the benchmark suite. The problem has been reported upstream to
> RedHat. (*)
does not seems my case:
i is 1000000 usertime:0.07 systime:0.78
[root@td145 tmp]# ./test.pl
i is 1000000 usertime:0.07 systime:0.75
[root@td145 tmp]# ./test.pl
i is 1000000 usertime:0.07 systime:0.79
[root@td145 tmp]# ./test.pl
i is 1000000 usertime:0.07 systime:0.79
[root@td145 tmp]# ./test.pl
i is 1000000 usertime:0.06 systime:0.77
[root@td145 tmp]# ./test.pl
i is 1000000 usertime:0.07 systime:0.78
Thanks for your reply!
> Cheers,
> Ulrich
>
> (*) if you see this behavior only sometimes, then you may be lucky by
> just retrying to run the benchmark ....
Cheers,
Arnau
|