🚀 go-pugleaf

In article <CyKrsp.6zu@icdc.delcoelect.com>,
Thomas V. Myers <tvmyers@koicds01.icdc.delcoelect.com> wrote:
>So name me some 'well known benchmarks' that I can FTP without being either a
>member of the RAB or an ANSI/ISO fellow, etc.  I'm not interested in paying a
>bunch of money in order to run a supposedly 'open systems' benchmark.  I also
>don't trust the vendor to run the benchmark, especially for mass storage
>sub-systems.  Chances are slim-to-none that they use the same host with the
>same patch level of O/S, RAM size, etc. that I'm using.

1. The Byte Unix Benchmarks.
2. IOZONE

Come to mind quickly.  One of my favorite Q&D tests is a simple raw
thruput test.  You need to use BIG numbers and time it:

$ time dd if=/dev/rdsk/0s0 of=/dev/null bs=65536 count=1024

Q&D 64MB sequential READ test that I've found to be a good starting
point when examining the speed of disk drives.

I've found that hooking up different SCSI drives to the _same_ machine
and running the above 'dd' test, reflects a given drives performance
pretty accurately reflects on the other tests also.

I don't think there is a whole lot you can do to optimize a 64MB read
other than stuffing 64MB of cache somewhere :).  You may also find that
replacing your SCSI Host Adaptor buys you more than your Disk subsystem.

Replacing an AHA 1542 with a 1742, dropped my times from the 30 second
range into the 17 second range on the above test.  Benchmark numbers
(even silly ones like 'dd') that are that radically different show a
_me_ whole lot.  The 2742 was only able to turn low 20's.  I'd be
curious to see what the new 29XX's will do, but my box is EISA/VESA.
-- 
Keith Smith        Digital Designs               keith@ksmith.com
5719 Archer Rd.    PO Box 85       Free Usenet News and Internet Mail Services
Hope Mills, NC 28348-0085         All 28K/14K Modems  (910) 423-4216/7389/7391
Somewhere in the Styx of North Carolina ...         14K-V.32/28K-V.34/28K-V.34