Date: Sat, 11 Sep 2004 00:48:05 +0200 From: "Bjoern Koenig" <bkoenig@cs.tu-berlin.de> To: <freebsd-current@freebsd.org> Cc: sos@deepcore.dk Subject: profiling data of lacking ICH2 (was: poor ATA disk speed with ICH2) Message-ID: <20040910224642.9B70D631D@hoppel.local>
index | next in thread | raw e-mail
Hello again, I built a recent GENERIC kernel (1.413.2.2 2004/09/09 23:10:48 scottl) with high resolution kernel profiling (config -pp). I ran a little shell script in single-user mode which looks like [...] kgmon -rB dd if=/dev/zero of=/usr/out bs=4096 count=262144 kgmon -hpr [...] and created the following output with gprof: http://www.alpha-tierchen.de/dateien/gprof.out.txt (I'm sorry for posting this external link, but I didn't like to append ~15000 lines to this mail ;-) This is my dmesg: http://www.alpha-tierchen.de/dateien/dmesg.txt hw.ata.wc=1 hw.ata.ata_dma=1 The dd command told ~15 MB/s without profiling (~500 kB/s with profiling). There is absolutely no failure with the hard disk and the controller, because after that I installed a 4.10-RELEASE into the same slice and it works pretty fine (~37 MB/s). I repeated this test at another position on hard disk _completely_ ... and as expected I got similar results. I hope the profiling data helps. I would like to see a gprof output of a system which works correctly, otherwise I have no possibility to compare. I have the time and motivation to analyse them and point out the difference. best regards Björn Könighelp
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20040910224642.9B70D631D>