Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!usc!rpi!masscomp!peora!tarpit!bilver!jwt!john
From: john@jwt.UUCP (John Temples)
Newsgroups: comp.unix.sysv386
Subject: Re: IDE Drives with Interactive Unix SysV 3.2  2.02
Keywords: interactive hard disk
Message-ID: <1991Mar13.035309.22322@jwt.UUCP>
Date: 13 Mar 91 03:53:09 GMT
References: <1955@kuling.UUCP> <27d742a0-4fa.2comp.unix.sysv386-1@vpnet.chi.il.us> <1991Mar12.131829.7403@hyper.hyper.com>
Organization: Private System -- Orlando, FL
Lines: 20

In article <1991Mar12.131829.7403@hyper.hyper.com> lam@hyper.UUCP (Edmund C. Lam,,) writes:
>The drawback with IDE comes from the fact that the interface limits
>your data transfer rates to an observed maximum of 400K/s.  

How did you measure these transfer rates?  When I run something like
Coretest under raw DOS, I see transfer rates of around 1.1 MB/sec off
of a Conner drive.  But since Coretest is reading < 64KB blocks in its
test, and the controller has a 64KB cache buffer, all I'm seeing is the
read rate from the cache.  I would assume this is the "interface
speed,", i.e., the speed at which data can be moved from the controller
to the CPU, irrespective of the speed at which data can be moved from
the drive to the controller. 

Under UNIX (and DOS), I see "application" transfer rates which are much
lower -- below 200KB/sec for reads and 50KB/sec for writes.  I don't
know if this is a limitation in the drive-to-controller transfer rate,
or something else.

-- 
John W. Temples -- john@jwt.UUCP (uunet!jwt!john)