XP Setup hangs with dual PIII-550 on Asus P2B-DS v1.04

[Nir Gilboa]

Hi

I tried memtest86 but that hangs after several minutes too. Also I took
a closer look at the Slot 1's and they are clean and corrosion free.
I had two PII350's sitting around so I shoved those in there instead of
the PIII's and XP loads and is stable.

The next thing I tried is to underclock the system bus to 66 FSB.
That seems to lead to a stable system as well. I only tried it for about
30 minutes but I think the odds are pretty good since it always used
to hang after about 2 minutes or even less.

I guess I will try to run the system bus at 75 or 83 Mhz and see if they
are stable too. Is there any way you guys can think of that I might find
out what's causing the problem at 100 Mhz ?

Also I checked and my board does not have the ACPI fix - R80 is there
and R79's place is empty.

Thanks again for all your help.. i think this is finally getting somewhere.

Nir

 

[Mercury]

Well, the task manager issue was cosmetic for me - made no difference that I
could detect at the time.

Is CPU stepping still an issue with P3's? Could that be it?

 

[Stephan Grossklass]

Hi

I tried memtest86 but that hangs after several minutes too. Also I took
a closer look at the Slot 1's and they are clean and corrosion free.
I had two PII350's sitting around so I shoved those in there instead of
the PIII's and XP loads and is stable.

The next thing I tried is to underclock the system bus to 66 FSB.
That seems to lead to a stable system as well. I only tried it for about
30 minutes but I think the odds are pretty good since it always used
to hang after about 2 minutes or even less.

That basically leaves 4 possible causes:
* The old Vtt issue - it occurred primarily with Katmai core PIIIs.
* Thermal issues - I think one of the PIIIs is an OEM version, does the
cooler make proper contact to the die there and how hot does it get?
The MPS HAL does *NOT* use HLT for some reason, so the CPUs will heat
up quite a bit even with nothing to do. (Try CPUIdle, the "Pro"
version has always supported multiple threads so the current "Extreme"
should do the same.)
* Problems with the core voltage regulators, possibly caused by
dried-out or even leaking associated electrolytics.
* As above, but caused by an underpowered/defective PSU.

Also I checked and my board does not have the ACPI fix - R80 is there
and R79's place is empty.

As to be expected.

Stephan

 

[P2B]

Have you seen it take a very long time booting into W2K/WXP?.

Never - except when certain USB devices were connected, but that's an
entirely different can of worms

If so, this
is caused by the same problem. It didn't happen on the first
boards/BIOS's, but some time latter, there was a BIOS change, that caused
this problem. I went through a dozen P2B variants, and several BIOS
versions, and all would display it in some configurations. It was however
very 'odd' in terms of the patterns that gave it. For example, I had a
system that ran W2K fine, once installed, but would always crash on the
install, if using an IDE CD, but copying the disk to the SCSI HD, and
installing from this worked fine. Another system installed perfectly the
first time, but changing to a latter CD drive, introduced this behaviour.
Five drives latter, from different manufacturers, it evntually worked
again. The problem particularly appeared when there was an IDE CD from
particular manufacturers, but only SCSI HD's.

Perhaps I've never seen it because I've always used SCSI HDs *and* SCSI
CD-ROMs.

There was also a seperate issue, that might apply in this case, relating
to the video card, and the SCSI BIOS. Some video cards caused a problem
with the SCSI BIOS on these machines during boot. There is an area in the
upper memory block, that is incorrectly flagged during bootup, and in some
cases leads to both the SCSI system, and the video, trying to map stuff to
the same address. Some families of video cards, never really worked...

Haven't encountered that either. Closest I've come was a P2B-AE (Sony
VAIO OEM board) with it's drives on an Adaptec 2940U2W that consistently
lost video partway through XP setup. Turned out to be a BIOS conflict,
updating the 2940 BIOS fixed it. I've never run into a video card that
wouldn't work - but it would be interesting to know if the OPs video
card is one that I've used.

Hardly 'cosmetic'. It was caused by the ACPI interrupt continuously
triggering, and the CPU handling these, basically sits 'flat out' doing
little else. It normally did not cause install problems on W2K, but I have
seen it give trouble on XP.



Depends on the benchmark. Most benchmark tests, only use one processor.
Try a compile that supports SMP, and the time needed reduces directly in
line with the change in processor useage.

The benchmark I used did support SMP - at least Task Manager showed 100%
utilisation of both processors while it was running, and the results led
me to conclude the ACPI issue was strictly cosmetic - as claimed by
Asus. It seems reasonable the ACPI hardware issue would cause spurious
interrupts, however my tests found no performance impact as a result.

P2B

 

[P2B]

Nir Gilboa schrieb:




That basically leaves 4 possible causes:
* The old Vtt issue - it occurred primarily with Katmai core PIIIs.

Doesn't ring a bell, can you elaborate?

* Thermal issues - I think one of the PIIIs is an OEM version, does the
cooler make proper contact to the die there and how hot does it get?
The MPS HAL does *NOT* use HLT for some reason, so the CPUs will heat
up quite a bit even with nothing to do. (Try CPUIdle, the "Pro"
version has always supported multiple threads so the current "Extreme"
should do the same.)
* Problems with the core voltage regulators, possibly caused by
dried-out or even leaking associated electrolytics.
* As above, but caused by an underpowered/defective PSU.

I agree entirely, but I'd still clean the slots to be sure. You can't
see oxidation on those contacts. On the other hand, I've never
encountered a board that was stable at 66 but not 100 - cleaning has
only fixed issues at 133+ for me.

P2B

 

[Mercury]

if it was interrupts causing the issue the do you know how that interrupt
would only ever get to the second CPU? Doesn't sound right to me as AFAIK
interrups either go to CPU0 or on smarter / later systems get balanced
across CPU's.......