#EISA SCSI problems
8 messages in this thread
Gary:
Remember the question I had last week re. the EISA SCSI cache ctrl. and large
file paging locking up? Well my friend just called me and said tech support at
Phar Lap just told him this is a known bug or incompatibility of pharlap
virtual control memory manager and EISA SCSI. The bug still exists in their
latest release but they are working on a patch as "highest priority" to fix
this. Maybe you could verify this in explanation to the forum members. My
friend requested I pass this along to you. His system locks up the moment he
pages to the EISA SCSI drive. He also wanted to request if anybody knows of a
work around to the EISA SCSI bug other than more RAM to let us know.
Don,
All of our systems run Eisa Scsi drive controllers and we haven't had any
problems. If there's a particular board we could check for you let us know.
Paul
Thanks for the reply, Paul.
In many cases I've read about on this forum purported "bugs" often turn out to
be cases of improperly installed drivers, defective hardware, or peculiar
combinations of hardware, etc. Hopefully, you can prove this case is not a bug
but another improper installation.
I believe the cache ctrl. is an AMI EISA SCSI with 16 meg installed.
The MB is an EISA ALR Power Cache of circa 1990 vintage with 32 meg
RAM. Video card is a Hercules Superstation XP. His problem surfaces
when rendering requires access to the swap disk. The HD is a 1.2G
Quantum with 1.05G free (all contigous). Windows works fine
including photostyler. He has a device ASPI.sys installed in
config.sys.
Don,
I can come pretty close to the configuration you mentioned. We are
running AMI EISA Motherboards with AMI EISA SCSI cache conrollers,
the controllers are running various amounts of ram. All of our
systems have at least 64 megs of ram on the MB, but we have still
found ourselves paging to disk, We've never had any problems with
them. On the hard disk side we're running arrays of Fujitsu 1.2 and
1.7 gig drives, so except for the ALR machine we're pretty much
running the same sorts of systems. My first looks would be at the
EISA setup fields, to make sure that the MB is accessing everything
correctly and that the controller card is fully configured with the
correct !*.cfg and !*.ovr files. I can't emphasize enough how
important these files are to the system integrity when utilizing an
EISA bus. Also you may want to make sure that if he is utilizing a
SCSI-2 interface that the drive is setup to run under SCSI-2. We've
had people in the past configure their controllers for 32 bit SCSI-2
and still run drives under a standard SCSI protocol. We do have at
least one ALR EISA machine around here, so if you can't get the
problem solved, let me know and I'll see if we can't swap some boards
around here and duplicate your setup more closely.
Paul
Thanks again Paul for your response. It appears to me that this problem is
resolvable but just a few more questions. Are you using any Model 441 AMI
controllers? Would you suspect that an early (1990) vintage EISA ALR bus is
fully EISA compatible? The controller is configured for 32 bit SCSI-2 but he
says he tried it both ways. We're also wondering whether the Quantum 1.2 G
drive is SCSI-2 compatible. Would it work at all if it wasn't?
Don,
No problem on the questions, we're always glad to help.
If your using a full length AMI SCSI-2 16MB caching controller, then we're
using the same one. The Quantum drive (I'm assuming it's a ProDrv-1225S) is
fully SCSI-2 compatible. However one problem that ~may~ be arising is multiple
hardware caches. You have a 512KB cache on the drive itself that allows it to
achieve its 10ms access time and then you have a seperate "huge" cache on the
controller. This configuration has been known to cause problems with some
software and actually slows down most systems. It may be advisable to try
disabling the drive's 512kb cache.
Additionally, if you're using the Advanced SCSI Programming Interface (ASPI)
driver, then make sure that you're using the latest one for the card, and that
you're using the right one. There are several enhanced versions that support
SCSI-2's 32 bit addressing,Single fast mailbox handling, Tagged queing, etc.
Finally, the ALR board sounds like it was definately on the bleeding edge of
1990 EISA standard and the EISA bios on the board may be causing problems. If
all else fails, you may want to call ALR and see if you can't get an updated
bios for a few bucks.
AMI has a BBS at (404)246-8780 HST compatible at 14,400. That's always a good
call for the latest card drivers.
Hope it helps, Paul
hmmmm…. I've seen many many EISA SCSI systems that work fine with Phar Lap. I
think that your friend might have misinterpreted what he heard from the tech
support person at Phar Lap — perhaps the problem relates to a _specific_ EISA
SCSI controller.
I'll make a note to ask someone at PL about this the next time we speak to
them.
– G
Thanks, Gary, I'll be looking forward to anything you find out about this. The
guy is pretty upset since he spent so much$ and can't get it to work. Please
see thread response to Paul Lind(anim8). So far it appears that his problem IS
hardware specific although he is convinced otherwise. Guess you heard this
many times before. It would be great if I could get him to buy my DPT EISA
ctrl. card which I can't seem to find a buyer for. <G>