Date: Thu, 20 Feb 2003 15:23:47 -0600 From: Dan Nelson <dnelson@allantgroup.com> To: Will Saxon <WillS@housing.ufl.edu> Cc: questions@FreeBSD.ORG Subject: Re: network tuning Message-ID: <20030220212347.GR13096@dan.emsphone.com> In-Reply-To: <0E972CEE334BFE4291CD07E056C76ED8CBBC03@bragi.housing.ufl.edu> References: <0E972CEE334BFE4291CD07E056C76ED8CBBC03@bragi.housing.ufl.edu>
next in thread | previous in thread | raw e-mail | index | archive | help
In the last episode (Feb 20), Will Saxon said: > From: Dan Nelson [mailto:dnelson@allantgroup.com] > > That's consistent with a dual-CPU box. The CPU states are for the > > system as a whole, but the CPU usages in the process listing are > > per-process. A single CPU-heavy process will cause its process > > line to hit 100% CPU, but that will only force the User percentage > > to 50%, since there is antoher CPU sitting idle. > > Aha, didn't realize this. We were all kind of wondering if that was > all-cpu or just the one cpu. Well, there isn't a faster processor > available on that platform so we will have to just build something > better, or find something multithreaded. Threading won't help you, since FreeBSD's threads implementation is a single-process model, and if you're just watching one interface you really have a serial process that won't benefit from multiple CPUs anyway (grab packet, add its counts to some arrays, grab next packet). If you're bored you can build ntop with profiling and figure out what part of the code is the killer. -- Dan Nelson dnelson@allantgroup.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20030220212347.GR13096>