MAXTOR XT4170 with WD1007 Controller (& SpeedStore)

[Karl Denninger]

In article <10375 at rls.UUCP> randy at rls.UUCP (Randall L. Smith) writes:
>
>For openers, my comments were directed to the WD1007-WA2 and the XT4170-E.

True... and mine were more towards it's replacement, the SE2.

>> Unless you're set up for a secondary ROM entry (which is REAL important if
>> you have a VGA card), in which case it's CC00:5.
>
>Granted.  Presumably, if you set up such a configuration, you would be
>aware of the differing address requirements of unique setups.  The address
>C800:5 is by and large more likely.

Not unless you run monochrome monitors!  Most color systems will be using
CC00:5.....

>>> Somehow by multiplying them together you get the actual surface area
>>> equivalence for the drive.  For a drive with more than 1024 cylinders
>>> (as is XT4170E) use the selection with 63 sectors/track. 
>> 
>> NO!  Doing that makes the system quite a bit slower than if you leave it at
>> the real physical parameters!   You see, the controller has to remapping if
>> you set things up like that.  
>
>But Karl, that is the function of the WD1007-WA2.  Dos and earlier versions
>of Unix cannot (with native drivers) access those extended address ranges. 
>Since all the computing and remapping is done on the controller, this
>should not be slower.

SCO Xenix/Unix, ISC 2.x, & AT&T 3.2 can all deal with > 1024 cylinders.  
And if you can, you should use them.  SCO can even have a DOS partition on 
a drive like this, as long as the DOS partition lives within the first 1024 
cylinders on the disk.

The controller WILL remap the ROM entries to support this configuration as
well.  SCO Unix/Xenix finds the correct parameters, as does ISC.

>If your system can directly address these drives, then more power to you. 
>I too would use direct addressing if it were possible, but alas it is not
>here.  I would venture most cannot and I am not aware of any that do.  I'd
>love to hear (by e-mail) of some that do. 

See above.  Most modern Unixen can work, and work best, in native mode.

>> The WD1007-SE2 can also set up for 36 sectors and still run 1:1 interleave,
>> while the WD1007-WA2 can only do 35 sectors if you want 1:1, not to mention
>> that the SE2 board will also run 15Mhz (760MB) drives at full rated speed.
>>
>>> The real bummer is you have to *manually* enter all the defect list.
>>> Grrrr.  WD says they will fix this eventually.  It's really unthinkable
>>> they didn't provide that from day one.  Oh well.
>> 
>> The SE2 doesn't do this; it reads the defect list from the drive and ALSO
>> allows you to enter manual defects as well.
>
>Well, we're talking about two different controllers in the same line and I
>cannot speak to the SE2.  Your corrections really apply to the SE2 and it
>is my fault for not considering that controller in making my comments. 
>However Karl, all my instructions come from TFM or my own experience so you
>are simply refuting Western Digitals manuals and reccommendations.

TFM and WD's recommendations don't have anything to do with Unix - they're
all oriented for MSDOS and Novell Netware people.  For those users the
recommendations are correct.  For the Unix users they are not terribly
wrong, but they do lead to lower performance than you could obtain
otherwise.

The other problem with the "remapping" is that some OS versions may attempt,
now or in the future, to do some form of cylinder grouping on files.  You
are going to find your performance lowered significantly if you run
remapped and don't have to, since what the system thinks defines a cylinder
is not in reality a single disk cylinder anymore.

>As I said before, this is *exactly* what I am running under Microport v3.0e
>and *no* problems with performance or reliability have occurred.

You won't see a reliability problem.  Performance is subjective, at least
until you see one running optioned the other way.  I've seen some difference
between remap/noremap options; it ends up being about 15% at the filesystem
level.  Nothing spectacular, but nothing to sneeze at either.

--
Karl Denninger (karl at ddsw1.MCS.COM, <well-connected>!ddsw1!karl)
Public Access Data Line: [+1 708 566-8911], Voice: [+1 708 566-8910]
Macro Computer Solutions, Inc.		"Quality Solutions at a Fair Price"