SANTAROS@UWF.BITNET (09/03/89)
I am forwarding this personal mail and my personal reply for comp.sys.amiga
readers.
<<<<<< Original Letter Follows >>>>>>>>
Date: Sun, 27 Aug 89 17:28:58 -0400
From: William Dimm <wdimm@lehi3b15.csee.Lehigh.EDU>
Message-Id: <8908272128.AA16499@lehi3b15.csee.Lehigh.EDU>
To: SANTAROS@UWF.BITNET
Subject: Re: CLtd. Controllers (KRONOS)
Hmm.... A question: Is Kronos available for the A1000 yet?
I am currently using an CLtd. interface with a Conner Peripheral CP340.
I am getting around 100K/s reads (don't remember the write speed)
according to diskperf. I have an interleave of 3 - a friend hooked
it up to a logic analyzer & said that 2.7 would be optimal, so I am
using 3. So... why is it going so slow for me when you report
much faster speeds? The drive has 29ms access, & can read 1MB/s
from the media, & 2MB/s from the buffer (which unfortunately is small -
only 2k, or was it 4k?). Any guesses?
--
------------------------------------------------------------------------
// Bill Dimm wdimm@lehi3b15.csee.Lehigh.EDU wcd0@lehigh.BITNET
\X/ "It's not the facts but the explanation that counts." - unknown
------------------------------------------------------------------------
<<<<<<< END OF BILL'S LETTER>>>>>>>
<<<<<< My response starts here !!! >>>>>>>
Date: Sat, 02 Sep 89 22:07:35 CDT
From: SANTAROS@UWF
To: wdimm@lehi3b15.csee.lehigh.edu
Subject: RE: CLtd Controllers (slow CONNER HDRIVE)
Hey sorry for the slow response to your letter, I got bogged down in
mail/school/BBS stuff. Anyway, I really cant fig why your CLtd controller has
such slow speeds, however, have you tried higher interleaves (or slower). I
know you have tested it on an analyser, but was that just the drive? The
AmigaDOS and CLtd controller software (CLtd.device) also affect transfer rate.
In actuallity that is why you can change the interleave! You should do the
painful task of starting at interleave of 1:1 and formatting say 10 cylinders
with AmigaDOS. Then test with your favorite speed program. I use DevSpeed
from CLtd since it tells me how fast CLtd.device and the drive get along, but
DiskPerfa or DPerf2 will work just as well.
As far as buffers go, I set CLtd's SCSI-DOS DevSetUp file to CLtd's default of
9 buffers (I think). In the MountList, however, I have found that using too
few or too many buffers can slow things up! The best combo is somwhere around
the blocks per track entry in the MountList. This seems right since we want
to buffer at least one track of the drive for fast access.
The conner should produce around 250-300K reads with old CLtd unless there is
somthing about that drive I am not aware of.
Oh yea, make sure that you have BLOCKREADS and ADAPTIVEBUFFERS on in DevSetUp.
The conner should be able to handle these, although my Rodime could not on
the old CLtd controller.
If there are any other questions or you still can't get it to work send me
mail and call CLtd about it. They are willing to help. Be patient cause
sometimes they don't understand what you mean the first call.
Talk to ya later!
(A copy of original letter and this one sent to comp.sys.amiga)
<><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><>
John Preston Preston Enterprises Pensacola FL (904)968-2232 CBM DEV
><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><
<<<<<<<< END MY PERSONAL RESPONSE >>>>>>>>>
I hope this might help others with simmilar problems
Oh yea, one thing I left out of my response, CLtd. is working on the
A1000 controllers and they should/are/will be ready very soon!
Later, John Preston