Return to BSD News archive
Xref: sserve comp.unix.bsd:7546 comp.os.linux:14972 Newsgroups: comp.unix.bsd,comp.os.linux Path: sserve!manuel.anu.edu.au!munnari.oz.au!sgiblab!swrinde!emory!wupost!gumby!yale!yale.edu!jvnc.net!nuscc!ntuix!eoahmad From: eoahmad@ntuix.ntu.ac.sg (Othman Ahmad) Subject: Re: IDE faster than SCSI-2 Message-ID: <1992Nov6.012510.12371@ntuix.ntu.ac.sg> Organization: Nanyang Technological University - Singapore X-Newsreader: TIN [version 1.1 PL6] References: <MAD.92Nov5204000@lettuce.math.keio.ac.jp> Date: Fri, 6 Nov 1992 01:25:10 GMT Lines: 30 MAEDA Atusi (mad@math.keio.ac.jp) wrote: : The file size of one megabyte is too short to get meaningful result. : You are just measuring the speed of copying to/from buffer cache. You may be right there but remember that Julian's machine runs at 50Mhz, whereas mine at 33Mhz. buffer cache from 386bsd cannot be as large as 1 Mbyte. At least we are testing the efficiency of the use of buffer cache. The latest version of iozone actually flushes the cache but I do not like this, because that is not how we use unix file system. It does not measure realistic situation. If you want a figure excluding the disk/buffer cache, take the read figure. If it were really all from buffer cache, then the write figure should be much higher, equivalent to memory bandwidth, i.e. 25Mbyte/second. Assuming 4 cycles of 50Mhz, 32 bit per cycle. : : You should give numbers at least twice as large as your RAM size. : Actually I did that, but the figures are not informative. Not relevant to our usual use. It confused more than it explains. I advise people to ignore these "unrealistically high load". However I'll do it anyway. -- Othman bin Ahmad, School of EEE, Nanyang Technological University, Singapore 2263. Internet Email: eoahmad@ntuix.ntu.ac.sg Bitnet Email: eoahmad@ntuvax.bitnet