Date: Thu, 17 Aug 2006 10:57:04 -0400 From: Bill LeFebvre <bill@lefebvre.org> To: freebsd-stable@freebsd.org Subject: Re: TOP shows above 100% WCPU usage Message-ID: <44E483C0.6050209@lefebvre.org> In-Reply-To: <20060817144110.GB88424@dan.emsphone.com> References: <44E1F796.5070105@rogers.com> <20060815172728.GB88051@dan.emsphone.com> <44E204C0.60806@rogers.com> <44E3FE61.6060800@lefebvre.org> <20060817144110.GB88424@dan.emsphone.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Dan Nelson wrote:
> One problem is that method doesn't scale to lots of CPUs. On a Sun
> T2000 a non-threaded process consuming all of one CPU would only report
> 3.12 %CPU in that case (100/32).
I agree. The alternative is having a 10-thread process on such a system
report 1000% cpu utilization, and/or having the sum of the cpu utilization
column be 3200. Not sure which is worse.
>
> I just built top-3.6 on such a system, though, and it does report a
> simple "main(){for(;;);}" process as consuming 100 %CPU. Maybe you're
> thinking of Solaris's own prstat command?
>
Heh. I released 3.6 with new SunOS code that didn't adjust for number of
cpus, and someone flagged the behavior as a bug. So you're right: 3.6
doesn't do it this way. But 3.5 did, and it seems at least some people
prefer it that way.
Bill LeFebvre
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?44E483C0.6050209>