Date: Sat, 1 Jan 2000 16:01:59 +1030 From: Greg Lehey <grog@lemis.com> To: R Joseph Wright <rjoseph@nwlink.com> Cc: freebsd-questions@FreeBSD.ORG Subject: Re: some performance issues Message-ID: <20000101160159.O1528@freebie.lemis.com> In-Reply-To: <386D8ABD.C2894A91@nwlink.com> References: <386C0676.F39EC477@3-cities.com> <386C2354.ABD1ED54@nwlink.com> <386C3173.1D695393@3-cities.com> <386C543D.6E59C9DF@nwlink.com> <19991231104441.C2609@emu.sourcee.com> <386CE8AB.29A140B5@nwlink.com> <386CF9DC.B71A9887@3-cities.com> <386D3D3C.C92D02A3@nwlink.com> <386D5C88.B8257D45@3-cities.com> <386D8ABD.C2894A91@nwlink.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Friday, 31 December 1999 at 21:03:57 -0800, R Joseph Wright wrote: > >>> I recompiled my kernel and changed the flags from 0x80ff80ff to >>> 0xa0ffa0ff. What is a good program to use to see if I can really tell a >>> difference? I tried out some of my bigger programs like star office and >>> gimp. Star Office still took a little while to load, gimp seemed to fly >>> (when it wasn't crashing). >> >> I use "iozone -s 160m" because I have 128MB of memory. The size of the >> file needs to be larger than memory so you aren't benchmarking the >> cache. You can find it in the packages/benchmark area of the CDROM. iozone is not really a storage device testing program. Use rawio (in the Ports Collection) for that: it bypasses the cache. Read carefully the warnings which tell you that you should not use the write tests on a file system which contains data you want to keep. > Kent, I tried that, only for me it worked doing "iozone 160". Here is > my results: > > 11483869 bytes/second for reading > 15155142 bytes/second for writing This is sequential access. You'll never get that in practice. > Is that really slow? It's nowhere near 33MB/second. 33 MB/s is the transfer rate from disk buffer to CPU. The transfer rate off the platter is slower, and your speeds there look pretty good. But remember that they're the ideal case. > I have a Maxtor 7000rpm with UDMA66 capabilities, although I'll have > to wait for -STABLE to support that, and a new motherboard as well. > I may recompile with the old flags just to see the difference. I'd be interested to see the difference, but I don't think it'll be very much. Try both with rawio, and look at the random access results, which are the only ones that count in practice. Greg -- When replying to this message, please copy the original recipients. For more information, see http://www.lemis.com/questions.html Finger grog@lemis.com for PGP public key See complete headers for address and phone numbers 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?20000101160159.O1528>