Date: Thu, 13 Jun 1996 12:29:38 BST From: Michael Searle <searle@longacre.demon.co.uk> To: questions@freefall.freebsd.org Subject: Re: Adaptec2940UW vs. BusLogicBT-958 (opinions?) Message-ID: <mDD97B580@longacre.demon.co.uk> References: <199606121658.JAA16509@freefall.freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
owner-questions-digest@freefall.freebsd.org wrote:
> It's not too hard to conceive of hardware which would fill > 10 MB/s.
> But the application is another matter.
> If you want to strip real time uncompressed video, you probably need it.
> If you're in the lab and need to do some kind of data acquisition you
> might need it. But running a busy web/news/mail/other-inet-application
> server doesn't - it needs lots of ios/sec, not lots of bandwidth. And a
> fast scsi channel can request just as many random seeks/sec as a UW bus.
> Again I'll challenge people to run iozone against either a MFS (am
> based) file system or just small files which fit within the disk cache
> system. If you can only write to *ram* at 10-20 MB/sec do you think
> it'll write to a scsi (UW) channel any faster?
IOZONE: Performance Test of Sequential File I/O -- V2.01 (10/21/94)
By Bill Norcott
Operating System: FreeBSD 2.x -- using fsync()
Send comments to: b_norcott@xway.com
IOZONE writes a 1 Megabyte sequential file consisting of
128 records which are each 8192 bytes in length.
It then reads the file. It prints the bytes-per-second
rate at which the computer can read and write files.
Writing the 1 Megabyte file, 'iozone.tmp'...0.679688 seconds
Reading the file...0.023438 seconds
IOZONE performance measurements:
1542732 bytes/second for writing the file
44739242 bytes/second for reading the file
The test completed too quickly to give a good result
You will get a more precise measure of this machine's
performance by re-running IOZONE using the command:
iozone 28 (i.e., file size = 28 megabytes)
Although iozone seems very inaccurate at such high speeds, sometimes I get
33MB/s, sometimes 44MB/s, but never anything in between. I guess that
0.023438 seconds isn't really accurate to 1us, more like 1/128 second (as
the 33MB/s give 0.03125 (1/32) seconds). And this was done on a P166.
--
Michael Searle - searle@longacre.demon.co.uk
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?mDD97B580>