home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.os.coherent
- Path: sparky!uunet!cs.utexas.edu!torn!mcshub!csx.cciw.ca!hcp
- From: hcp@csx.cciw.ca (H.C. Pulley)
- Subject: Re: Slow IDE perf. fixed!
- Organization: Canada Centre for Inland Waters
- Date: Mon, 21 Dec 1992 14:46:39 GMT
- Message-ID: <1992Dec21.144639.20931@csx.cciw.ca>
- References: <476354@lith.UUCP>
- Lines: 47
-
- In article <476354@lith.UUCP> kestas@lith.UUCP (Kestutis Sliupas) writes:
- >Hi:
- > I would like to report that mkfs on my IDE drive increased
- >the performance 4-500%.
- >
- > A couple of months ago, I measured my IDE performance as
- >@27Kbytes/sec. for read/write. (Seagate ST-1144A, 8M Memory, 386-25)
- >This was not acceptable. I posted to the net and support@mwc.com,
- >and received many suggestions.
- >
- > The solution was to re-mkfs my file system while modifying
- >the 'n' (virtual cylinder size) and 'm' (free list interleave)
- >parameters. After much experimentation and angst, I found that
- >using n=210 and m=6 increased the drive speed to 149 (read) /
- >131 (write) Kbytes/sec. The most important parameter was 'm'.
- >These parameters may not be optimal for every IDE drive; they
- >worked for me. This needs to be determined for each drive.
- >
- > I used a borrowed IDE drive as a backup unit, and a floppy
- >based O/S for the root partition mkfs. HAVE A BACKUP! Hard drive
- >performance was measured using iozone v1.14b by Bill Norcott
- >(norcott_bill@tandem.com) with a 4Mb. filesize. I mailed it to
- >mwc; they may have it available for D/L.
- >
- > I want to thank the following people, in no particular order,
- >for their help and advice:
- > norm@mwc.com (Norman Bartek)
- > bob@dalek.mwc.com (Bob Hemedinger)
- > abner@mwc.com (Abner John Snell III)
- > doug@htrails.mwc.com (Douglas Peterson)
- > and anyone else who I missed.
-
- Thanks for following this stuff up Kestas!
-
- I will be experimenting on my Seagate ST1102A and Maxtor 7120A drives over the
- holidays.
-
- I suggest that we set up a database of good m and n values for various drives.
- We should put this into the FAQ (which, BTW, I haven't seen for a long
- time...).
-
- Harry
- --
- hcp@csx.cciw.ca |This message released|It takes all kinds,
- hcpiv@grumpy.cis.uoguelph.ca |to the PUBLIC DOMAIN.|and to each his own.
- ----------------------------------+---------------------|This thought in mind,
- Stay away from the DOS side, Luke!|Un*x don't play that.|I walk alone.
-