The problem first began when
I upgraded my system with a FIC PA-2013
motherboard, which has a 100 MHz front-side bus, from the PA-2010 I
had
used previously. The 2010 has a 66 MHz FSB, and the 2940 was perfectly
stable when used with it.
The issue manifests itself
as mysterious and random hard lockups of the
system -- in other words, everything freezes for no apparent reason,
often
with the disk access LED solidly lit.
It was this visual indication
that led me to suspect the Adaptec card. I
tried the obvious solutions, such as upgrading the card's BIOS to current
level, and moving it to a different PCI slot. Neither one helped.
What finally cured the problem
was replacing the 2940 with an Initio
INI-9520UW. I had spoken to Initio's Support folks before settling
on their
card, and had asked them specifically if they had any problems with
100 MHz
FSB motherboards.
Not only had they not shown
any problems with such boards, but I was
told that the Engineering department had, as an experiment, overclocked
a 9520
to 118 MHz without any problems. If true, it's quite an accomplishment,
and
I have no reason to suspect Initio of stretching the truth at this
time.
My system has been absolutely
solid ever since the replacement, even
when I did things that I knew for a fact had caused lockups with the
Adaptec
(such as opening Protel '99 right after using Netscape, and running
the
SETI@Home client).
My (albeit unofficial) conclusion:
If you're looking at a mom-board with
a front-side bus faster than 66 MHz, you'd be better off with a SCSI
card
other than Adaptec.
He had problems with the 2940UW when upgrading from a FIC
PA-2010 to a PA-2013.
He explains the 2010 had a 66 MHz. FSB, while the 2013 has a 100
MHz. This depends on the processor you insert on the board, but
more importantly: The PCI-clock doesn't change, so nothing SHOULD change
according to the Adaptec 2940UW. On the PA2010 the PCI-clock is
simply 66 MHz. FSB divided by 2, so 33 MHz. On the PA2013 we have another
story, 100 MHz. FSB, but it is divided by 3, so 33.3 MHz.
That itself shouldn't give any troubles. However, more and more we
see another problem with super-socket 7 chipsets: incompatibility.
My
guess is that the MVP3 chipset (the chipset the PA2013 uses) isn't
so
compatible VIA wants us to believe. Furthermore, due to board-design,
a
lot of motherboards simply cannot deliver the power to the PCI and
AGP-cards, because some AGP-VGA-cards is withdrawing a lot of
current from the motherboard. Fluctuations in the power-usage could
result
in instability in voltage and frequency. Timing-critical components
like a high-end SCSI-card (the 2940UW is!) cannot handle this and a
crash
occurs...
More typical: When you buy an ASUS VR3800 AGP-VGA card, the retail
box states it doesn't work correctly on an ALI Aladdin V chipset.
Again, a super-socket 7 chipset.
I never heard of an Adaptec 2940xxx giving troubles on a motherboard
equipped with an Intel BX chipset (Pentium II/III/Celeron). But I've
heard of a
lot of problems with all sorts of PCI-cards with mainboards with a
MVP3
or Aladdin V chipset.
So my conclusion: The problem is not the Adaptec card, but the PA2013
mainboard. Changing to a different SCSI-card can help you overcome
the
problem (if that other SCSI-card is perhaps not so critical), but a
similar problem
can re-occur if -for instance- you replace the VGA graphics card.