Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!husc6!necntc!ames!ptsfa!ihnp4!homxb!mtuxo!mtune!mtung!pgf
From: pgf@mtung.ATT.COM (Paul Fox)
Newsgroups: comp.sys.att
Subject: Re: interleave and disk performance (7300 now)
Message-ID: <1004@mtung.ATT.COM>
Date: Wed, 7-Oct-87 10:06:42 EDT
Article-I.D.: mtung.1004
Posted: Wed Oct  7 10:06:42 1987
Date-Received: Sat, 10-Oct-87 17:40:58 EDT
References: <2310@sphinx.uchicago.edu> <710@neoucom.UUCP> <2119@ihlpe.ATT.COM> <6699@sgi.SGI.COM>
Reply-To: pgf@mtung.UUCP (gws-Paul Fox)
Organization: AT&T ISL Middletown NJ USA
Lines: 24

Speaking of interleaves, I thought I'd pass this tidbit on to all you 
performance hounds...   

The default filesystem interleave on the 7300/3B1 is 1.  I don't know 
what the low-level interleave is.  If the filesystem interleave is 
changed to 4 (either for the whole disk by modifying the mkfs 
parameters in the floppy boot code, or for the freespace only, by 
doing as fsck -S (is that right?  rt*m)) then data transfer rates go 
up quite a bit.  I.e., copying /unix to /dev/null takes about half the 
time it used to.  

Note that UNIX filesystem interleaves decay with time for the 
non-static part of the filesystem, since the freelist is not sorted,
and blocks slowly get shuffled.  Also, since it is a paging 
system, not all of an executable is read at once, so it's not clear 
what the win is for commands.  Data access behaviour affects disk 
performance a lot, of course.  

Anyway, I thought I'd mention it, in case anyone wanted to play with it.  Let
us know your results...

-- 
			Paul Fox, AT&T Information Systems, Middletown NJ.
			  [ihnp4|vax135]!mtung!pgf (201)957-2698