Date: Sat, 3 Nov 2007 15:53:44 -0800 (PST) From: Jeff Roberson <jroberson@chesapeake.net> To: Josh Carroll <josh.carroll@gmail.com> Cc: freebsd-performance@freebsd.org Subject: Re: ULE vs. 4BSD in RELENG_7 Message-ID: <20071103155130.Y544@10.0.0.1> In-Reply-To: <8cb6106e0711031629j68e181ccra0f16ee6f42ed982@mail.gmail.com> References: <8cb6106e0710230902x4edf2c8eu2d912d5de1f5d4a2@mail.gmail.com> <20071024111105.M598@10.0.0.1> <8cb6106e0710241229i12852d8cq436f4c955ac62c56@mail.gmail.com> <20071024133240.X598@10.0.0.1> <8cb6106e0710251925s2db0117cvcb67321b08d7b2a1@mail.gmail.com> <20071102102331.G544@10.0.0.1> <8cb6106e0711021306w10c48a15s99eab526064ac814@mail.gmail.com> <20071102150028.R544@10.0.0.1> <8cb6106e0711031629j68e181ccra0f16ee6f42ed982@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, 3 Nov 2007, Josh Carroll wrote: >> What would be interesting to know is if the sum of the temperatures is any >> different. 4BSD gets a much more random distribution of load because a >> thread is run on whatever cpu context switches next. ULE will have >> specific load patterns since it scans lists of cpus in a fixed order to >> assign load. So that means it prefers to run on lower numbered cpus if >> they are idle. This should have a side effect of allowing unused cores to >> powerdown more frequently than with 4BSD although I have not verified this >> in practice. > > The sum of the core temperatures when the system is idle is the same > for both ULE and 4BSD. 125 C in my case (31.25 C average). Under load, > the sum (again for both) is 184 C. So you're absolutely correct, while > one core seems to get these short bursts (probably my rrd script doing > its thing), perhaps this is shared over the cores with 4BSD. The > overall temperature remains the same. That's great to know. It would be possible to develop load balancing algorithms that attempt to keep as many cores as possible idling without sacraficing too much performance to improve power usage. This is something I've been thinking about as part of load balancing and topology improvements for 8.0. It's important to note that temperature of individual cores may not be the best estimate of power usage by that core since you would expect heat to travel well across a package leading to elevated temperatures for otherwise idle parts. Jeff > > Josh > _______________________________________________ > freebsd-performance@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-performance > To unsubscribe, send any mail to "freebsd-performance-unsubscribe@freebsd.org" >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20071103155130.Y544>