Date: Fri, 31 Oct 1997 22:37:50 -0500 (EST) From: Chuck Robey <chuckr@Glue.umd.edu> To: "Jordan K. Hubbard" <jkh@time.cdrom.com> Cc: smp@freebsd.org Subject: Re: Some SMP timing tests. Message-ID: <Pine.BSF.3.96.971031223222.11985L-100000@picnic.mat.net> In-Reply-To: <26870.878356552@time.cdrom.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, 31 Oct 1997, Jordan K. Hubbard wrote: > I'm not quite sure what to make of the following data, my suspicion > being that I'd have to actually add a killer I/O system to my test > machine in order to *truly* see the effects of SMP on the run time of > a make world since things look pretty I/O bound here, but nonetheless, > having spent 3 days collecting the data, I figured it was worth at > least posting a quick message about here. :-) > > Test machine was a dual P6/200 with Tyan 1668 motherboard, 64MB of > memory, an Adaptec 2940UW controller and IBM DCAS-34330W 4.3GB 5400 > RPM drive. Source tree used for testing was from 3.0-971029-SNAP. > > Two identical kernels were prepared, one with SMP support and one > without, for each run a "throw away" make world being done first > before timing a series of make -j<n> worlds, with n going from 1 to > 20. Each run started from a fresh reboot, no other activity going on > during the time of the runs. Just so I understand (I think I might try it) you had a complete make world in the can, so you rebuilt the minimum (no make clean) right? I have two disks (both with swap, one with /usr/src). I have room to put the /usr/obj on the other disk, but I'm not clear on how to do that ... would a simple ln -s /usr2/obj /usr/obj work (usr2==2nd disk) ? > The most interesting thing about these numbers was that at "high job > counts", where one would expect performance to start to actually > degrade due to having too many compiles competing for various system > resources, performance did not fall as expected. This leads me to > believe that our make actually artificially limits the parallelism > number to somewhere below 20. I haven't bothered to look into make's > code more thoroughly in verifying this, but that's certainly what it > looks like. [lots of postscript deleted] ----------------------------+----------------------------------------------- Chuck Robey | Interests include any kind of voice or data chuckr@glue.umd.edu | communications topic, C programming, and Unix. 213 Lakeside Drive Apt T-1 | Greenbelt, MD 20770 | I run Journey2 and picnic, both FreeBSD (301) 220-2114 | version 3.0 current -- and great FUN! ----------------------------+-----------------------------------------------
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.3.96.971031223222.11985L-100000>