Asus P2B v1.10 reboots @beginning of test #6 in memtest

[Ixnei]

On Mon, 26 Apr 2004 22:18:06 -0400, P2B wrote:
[snip]

Might be worth trying it without G35 to G37 - that's how I have my Asus
S370-DLs wired and they don't have this issue.

It appears that Roland is not using this one, but having the problem
still:

*****

... In particular, quite a few people connected G35 (Vtt) to G37 (vtt
only on tualatin) (which I didn't), ...

*****

Indeed, I didn't notice that until after I posted. My theories haven't
panned out too well so far :-(

At this point, IMHO it makes most sense to isolate the problem to the
board or slot adapter, by swapping in an adapter that's known not to
exhibit the problem - either a Slot-T or modded S370-DL. I have new
Slot-Ts in stock and would be happy to loan one or two if mailing costs
were covered.

Thanks for the offer. I've been tempted to just order on of these
slot-t's anyways, seeing as how they're only $25 shipped. That way, I
would have a nice "reference" to work with on these sorts of things.

<snip>

--
We HAVE been at war with Iraq for 13 years now, bombing their
country on at least a weekly basis.
"U.S.-led sanctions have killed over a million Iraqi citizens,
according to UN studies" - James Jennings
3,000+ innocent Iraqi civilian casualties can't be "wrong"...

 

[Roland Scheidegger]

I'd be interested to see if your board uses the same vendor mix/etc.

I've checked it and it's exactly the same.
I've also looked at the coppermine - tualatin datasheets again. About
the only interesting thing I've found is that Vtt slew rate up to
coppermines was 8A/us, whereas on tualatins it's 13A/us, though the
static current has not increased. That could be problematic, though
adding the small caps (for improving HF performance) I've done should
have helped in that case I guess, and since you also get the problem
with coppermines I don't know if this really could be a problem.
Also, I did not need to connect G35-G37, since the slotket already
connects this (even visible).
I've also found a potentially very problematic pin, X34 - this is
vcccore on mendocinos and (some) coppermines (and it's wired to vcccore
on the soltek sl-02a++), but Vtt on tualatins. However, it doesn't seem
to be used on the tualatin (at least not on the stepping I got),
according to the multimeter at least it doesn't seem to be connected.
I've isolated it just in case without any change, so I did to AF36.
Soldering a 15Ohm resistor to NCHCTRL (N37) to Vtt also didn't do
anything (hey I'm getting desperate).
I was unable to get any replacement capacitors for now, if I'd be
convinced that it really would help I'm sure I could find some shop
which has them (I only found rubycon zl 10V parts as the closest match,
but the 1000uF part is 10mm in diameter and thus not suitable for
replacing the 8mm parts).
I could still try lowering vtt voltage, but somehow I'm not confident it
would help.

:-(

Roland

 

[Ixnei]

I was unable to get any replacement capacitors for now, if I'd be
convinced that it really would help I'm sure I could find some shop
which has them (I only found rubycon zl 10V parts as the closest match,
but the 1000uF part is 10mm in diameter and thus not suitable for
replacing the 8mm parts). I could still try lowering vtt voltage, but
somehow I'm not confident it would help.

:-(

Roland

Just out of curiousity, what video card are you using with your setup?
Also, did you happen to run 3dmark2001?

FYI, test #15 (advanced pixel shader?, where it rebooted on me) is not
performed on older video cards like the geforce2 gts. So if I had used
one of my GF2's instead of a GF3, I wouldn't have gotten a reboot...

I know this is a triviality, as memtest can be used as a signal for the
problem, but ... I may bite the bullet and just swap out the $5 or so
worth of caps on this board. I will add a 1000uF where there is one
missing by the slot1 (CE3), maybe the same for the cap missing next to the
DIMM slots (CE27), and I will also increase the 1000uF cap to 1500uF where
the board indicates a larger drawn (CE12). Will post back later if/when I
do this, and results...

As an aside, I had a DFI AK74-EC socket-462 board with 2 of 3 1000uF caps
near the SDRAM DIMM slots clearly "blown" (bulging). This board would not
even POST with memory in DIMM 1 or 3. It would post from DIMM 2 and
passed memtest on 32MB, but anything bigger and it would fail many tests
misrably. I replaced the two caps, and was able to POST off of DIMM 1
using a 256MB stick, so I am encouraged by this capacitor modding (at
least in the extremes of obviously bad caps)...

--
We HAVE been at war with Iraq for 13 years now, bombing their
country on at least a weekly basis.
"U.S.-led sanctions have killed over a million Iraqi citizens,
according to UN studies" - James Jennings
3,000+ innocent Iraqi civilian casualties can't be "wrong"...

 

[Roland Scheidegger]

Just out of curiousity, what video card are you using with your
setup? Also, did you happen to run 3dmark2001?

Crap, forgot that.

FYI, test #15 (advanced pixel shader?, where it rebooted on me) is
not performed on older video cards like the geforce2 gts. So if I
had used one of my GF2's instead of a GF3, I wouldn't have gotten a
reboot...

I just have a ati radeon sdr, so even if I wouldn't have forgotten it I
couldn't have run this test...

I know this is a triviality, as memtest can be used as a signal for
the problem, but ... I may bite the bullet and just swap out the $5
or so worth of caps on this board. I will add a 1000uF where there
is one missing by the slot1 (CE3), maybe the same for the cap missing
next to the DIMM slots (CE27), and I will also increase the 1000uF
cap to 1500uF where the board indicates a larger drawn (CE12). Will
post back later if/when I do this, and results...

I wouldn't bother with the capacitor near the dimm slot - this doesn't
look like the problem has something to do with ram. Also, the cap
missing near slot1 (CE3) is not for Vtt. Not that it's bad idea to
replace other caps, as long as it isn't for I/O voltage (only needed for
external L2 cache for PII/PIII Katmai cpus) but I still think it's more
likely that vtt is a problem.
Or maybe better input filtering would help? If you look at the sample
application from the hip6019b datasheet, there are 4*1000uF caps for
bulk input filtering, 7*1000uF for Vout1 (Vcc), 5*1000uF for Vout2 (if
the p2b sticks to sample application closely, this is for L2 cache on
PII/PIII Katmai cpus mostly), 4x1000uF for Vout3 (vtt).

Happy soldering & good luck...

Roland

 

[P2B]

Just out of curiousity, what video card are you using with your setup?
Also, did you happen to run 3dmark2001?

FYI, test #15 (advanced pixel shader?, where it rebooted on me) is not
performed on older video cards like the geforce2 gts. So if I had used
one of my GF2's instead of a GF3, I wouldn't have gotten a reboot...

I know this is a triviality, as memtest can be used as a signal for the
problem, but ... I may bite the bullet and just swap out the $5 or so
worth of caps on this board. I will add a 1000uF where there is one
missing by the slot1 (CE3), maybe the same for the cap missing next to the
DIMM slots (CE27), and I will also increase the 1000uF cap to 1500uF where
the board indicates a larger drawn (CE12). Will post back later if/when I
do this, and results...

I'll wager a Slot-T against a P2B v1.10 it won't make a difference ;-)

P2B

 

[Ixnei]

Hi,

just to drop in a few lines: I have the same problem - and of course
I'am
interested if fiddeling around with capacitors will solve it.

Board: Asus P2B (No further letters)
Rev.: 1.10
Cpu: PIII-800 (Copper, 100Mhz FSB, 1,65 V) OS: Linux with Kernel 2.2.14
(Uptime > 2 Years now) VReg: HIP6019BCB Mem: Just upgraded from 2*64K ->
2*128K 100Mhz SRAM Cards: ELSA Winner 1000 VGA, SCSI, ISDN, Network, 2nd
Printer

(Because of the memory upgrade, I run in the memtest prob.)

What type of processor are you using? Is it a true slot1 processor, or a
socket 370 in a slotket? What slotket?

I was having the memtest test #6 problem with a coppermine celeron 900/100
and 600/66 running 256MB ram, both in a 370SPC v1.0 slocket. I managed to
pick up an Asus S370-DL rev 1.02 slotket, and it ran a coppermine celeron
1100/100 with 512MB on the P2B rev 1.10, passing memtest looping for hours
(with fastest CL2 memory timings).

So, there does appear to be a marginality created by the quality of
slotket used. I ended up dumping the board for $25, and moved on
already...

--
We HAVE been at war with Iraq for 14 years now, bombing their
country on at least a weekly basis.
"U.S.-led sanctions have killed over a million Iraqi citizens,
according to UN studies" - James Jennings
3,000+ innocent Iraqi civilian casualties can't be "wrong"...