home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.os.msdos.misc:7206 comp.os.linux:25554 comp.sys.ibm.pc.hardware:37221
- Newsgroups: comp.os.msdos.misc,comp.os.linux,comp.sys.ibm.pc.hardware
- Path: sparky!uunet!boulder!romeo!drew
- From: drew@romeo.cs.colorado.edu (Drew Eckhardt)
- Subject: Re: SCSI performance / Adaptec vs Seagate
- Message-ID: <1993Jan27.035844.14556@colorado.edu>
- Sender: news@colorado.edu (The Daily Planet)
- Nntp-Posting-Host: romeo.cs.colorado.edu
- Organization: University of Colorado at Boulder
- References: <C1FwFo.G63@jti.com>
- Date: Wed, 27 Jan 1993 03:58:44 GMT
- Lines: 45
-
- In article <C1FwFo.G63@jti.com> richb@jti.com (Richard Braun) writes:
- >Using a simplistic disk performance test program provided by a drive
- >manufacturer, I came up with the following surprising result:
- >
- > The 8-bit Seagate ST02 is nearly 3 times faster than the
- > 16-bit bus-mastering Adaptec 1542A.
- >
- >The program measures transfer rate, average seek time, track-to-track
- >seek time, and a so-called "performance index". With the drive
- >low-level formatted for the Seagate's disk geometry, I get 3.5ms
- >track-to-track time and a 30 performance index. Reformatted for the
- >Adaptec, the numbers are 4.9ms and 11.6.
- >
- >Why might this be? Is it the different head/sector mapping (64 vs. 63
- >heads, I think), or is the Adaptec just plain slower? Is it possible/helpful
- >to optimize the controller/device driver to match the physical disk
- >characteristics (15 heads, 1632 cylinders)?
- >
-
- Seek tests done via BIOS for IDE and SCSI disks are completely
- bogus. A track->track seek may not go off of the current track if
- the logical # of sectors per track > the real #.
-
- SCSI supports a seek command, but a disk doesn't have to do ANYTHING when
- it sees it. The BIOS on the controller doesn't have to do ANYTHING.
-
- As far as transfer rate, not using the hardware handshaking on the
- Seagate (enabled with the 0ws jumper), being "safe" and checking
- the SCSI status bits, you won't be able to transfer at > ~600K/sec.
-
- I know the adaptec has no problems above those speeds.
-
- Benchmarks need to be taken with a grain of salt, and need to
- be relevant.
-
- Finally, the Adaptec has more intelligence than the seagate, so you
- won't waste serious amounts of time in the kernel manually twiddling
- the SCSI bus. This will leave more time available for user processes,
- and your interactive performance on a real OS will be VASTLY superior
- to what the Seagate can give you.
-
- --
- Boycott AT&T for their absurd anti-BSDI lawsuit. | Drew Eckhardt
- Condemn Colorado for Amendment Two. | drew@cs.colorado.edu
- Use Linux, the fast, flexible, and free 386 unix |
-