Subject: Re: about to buy a...
To: None <>
From: Rodney W. Grimes <>
List: port-i386
Date: 07/08/1995 00:54:25
> Verily did "Rodney W. Grimes" write:
> > XX. MER DTC-3130 DTC 3130 NCR 53C810 fast SCSI-2 controller         $  81.00
> [...]
> > The DTC board also comes in a BIOS version as DTC 3131, but I don't
> > have a price on that version that is valid any more since I have gone to
> > using motherboards that have the NCR SDMS BIOS built into them.
> > Manufacuters and model numbers are all above, you should be able to price
> > shop and possibly find lower prices, heck even the distributor is endcoded
> > above :-).
> Are these DTC cards supported?  It appears odd that DTC chose to use a
> "public" chipset (i.e. non-proprietary).  No offense intended to our
> DTC lovers out there.

The DTC board looks like anyone else 53C810 based controller.  Since
the one chip pretty much does it all you won't find much difference in
anyones NCR based PCI controller cards.

DTC has been in both camps on the chips they use, sometimes they do there
own (like when they can offer a _significant_ value over standard parts),
somethimes they use off the shelf parts.

I can not speak for what is supported under NetBSD, the cards do work
under FreeBSD using basically the same code that is in NetBSD.

> Since we've swung the conversation to SCSI cards, if this DTC card is
> supported, how does it rate in speed/reliability/problem_free compared
> to Adaptec and Buslogic cards (the ones that actually work with blah MB
> and NetBSD).

Shouldn't really call it ``this DTC card'', we should be talking about
NCR 53C810 based controllers, as I stated above, they all look the same,
about the only difference is a very limit set of models include on board
BIOS support.

I would rate the ``speed'' as equivelent, but they do consume slightly
more PCI cycles than either the aha2940 or the bt946 due to the fact
that they are getting there scsi sequencer code out of host memory
instead of local memory.  This is a very insignificant amount of PCI
bus bandwidth when comprared to the _measured_ tranfer rates in
excess of 6MB/sec to a single drive using iozone that I have done using
the ASUS SC-200.

Reliability, well, any good MTBF formula will show you that reliabilty
is related to component count, and considering that the SC-200 has 3
active IC's and about 22 passive components (capacitors, resistors, 
connectors, jumpers, etc), and 1 crystal vs the bt946 which has 14 active
IC's and about the same passive component count and crystal it is rather
easy to state that the SC-200 has a significant advantage in the
reliability department!

I would also trust my SCSI chip design to a company who has been doing
it for 20 years rather than these PC centric companies that have only
been around for 1/4 as long.  Remeber NCR is in the SCSI chip market,
they supply the likes of Sun, HP, SGI, Mips, etc etc.. where as BusLogic
and Adaptec are purely PC based vendors.

> Also, since this conversation is highly PeeSea specific, why don't we
> move it to port-i386.. (I've made the start, sorry for the crossposting)

I finished the move... :-)  And as far as I am concerned this is the
end of this thread, I won't have much else to say on the subject.

Rod Grimes                            
Accurate Automation Company                 Reliable computers for FreeBSD