Been running ClimatePrediction for a while but got suspicious as it kept
on stopping short of a full run. Read around and got hold of Prime95
which showed there was a definite problem - failed after 1-2 mins. The
board is a Rev2.0 purchased in the middle of the summer and not
overclocked, just running at 200MHz with multiplier of 11. Backing the
freq down to 188 has reduced failures to ~ 2hours but this is not what I
expected.

The Asus Probe utility shows the 5v line at 4.865v when quiescent but
drops to 4.787v under Prime95 load - Oops I thought and got a new 550w
PS to replace the 360W one I was using. No change - 5v shows the same
figures. Checked the BIOS and it agrees with Probe for the quiescent
state. Used a Multimeter on a drive power cable and it shows 5.01v
quiescent. Had a peer at the MB but couldn't see obvious points to check
the supply on the board.

Is it possible I've got a board with an out of spec regulator?
Any other suggestions?

Other specs/history:
Proc: AMD 3200+
Mem: Matched pair 512Mb DDR400 from Mushkin in slots 2 & 3.
Mem running at spec (which is pretty aggressive) with Auto speed setting
(100% reduced stability?)
(Originally these were in 1 & 2 but had memory instability. MemTest86
tests showed slot 1 couldn't hack it at 200/400MHz. MemTest86 now rock
solid on 8hr tests)
Other BIOS settings on defaults.
2 CD/DVD, 1 floppy, 1 ZIP, 3 case fans, 2PS fans, AMD
MB temp (according to Probe) 22C, CPU 45C quiescent, 51C Prime95
--
Dhyan Ahado
(E-Mail Removed)

Dhyan Ahado wrote:

> MB temp (according to Probe) 22C, CPU 45C quiescent, 51C Prime95

Are all these temps measured with Probe? Probe reports the temp of the
socket sensor on the motherboard. MBM5 (latest version) gives both
this socket sensor and the CPU-diode. Under stress the diode temp can
be a lot higher

The highest socket temp I've ever seen with a 2500+, but running @
2200MHz with a raised Vcore of 1.850V is 48C. At that point the diode
showed 62C. Maybe your CPU is just running too hot.

Nanga

In message <(E-Mail Removed)>, Nanga Parbat
<(E-Mail Removed)> writes
>Dhyan Ahado wrote:
>
>> MB temp (according to Probe) 22C, CPU 45C quiescent, 51C Prime95
>
>Are all these temps measured with Probe? Probe reports the temp of the
>socket sensor on the motherboard. MBM5 (latest version) gives both
>this socket sensor and the CPU-diode. Under stress the diode temp can
>be a lot higher
>
>The highest socket temp I've ever seen with a 2500+, but running @
>2200MHz with a raised Vcore of 1.850V is 48C. At that point the diode
>showed 62C. Maybe your CPU is just running too hot.
>
>Nanga

They were - I've now put in MBM5. Diode temp ~55C. On starting Prime5 it
gets up to 66C before the fan speeds up and pulls it down to ~62C. Still
fails!

MBM5 reports 5v line at 4.81 - I assume it's reading the same sensor as
Probe & the BIOS.

On overclockers I saw a comment that anything below 4.9 was asking for
trouble.

--
Dhyan Ahado
(E-Mail Removed)

On Fri, 26 Dec 2003 22:10:11 +0000, Dhyan Ahado <(E-Mail Removed)>
wrote:

>In message <(E-Mail Removed)>, Nanga Parbat
><(E-Mail Removed)> writes
>>Dhyan Ahado wrote:
>>
>>> MB temp (according to Probe) 22C, CPU 45C quiescent, 51C Prime95
>>
>>Are all these temps measured with Probe? Probe reports the temp of the
>>socket sensor on the motherboard. MBM5 (latest version) gives both
>>this socket sensor and the CPU-diode. Under stress the diode temp can
>>be a lot higher
>>
>>The highest socket temp I've ever seen with a 2500+, but running @
>>2200MHz with a raised Vcore of 1.850V is 48C. At that point the diode
>>showed 62C. Maybe your CPU is just running too hot.
>>
>>Nanga
>
>They were - I've now put in MBM5. Diode temp ~55C. On starting Prime5 it
>gets up to 66C before the fan speeds up and pulls it down to ~62C. Still
>fails!
>
>MBM5 reports 5v line at 4.81 - I assume it's reading the same sensor as
>Probe & the BIOS.
>
>On overclockers I saw a comment that anything below 4.9 was asking for
>trouble.

FWIW, mine shows 4.892 to 4.919 for +5v in MBM5 and PC Probe, with a
digital voltage meter on an empty IDE cable I get +4.86 to +4.87 at idle
and +4.83 to +4.84 with prime95 running. I've had this Antec
400W/A7N8X/Barton combo for just over 7 months now and it's never
crashed/rebooted or locked up. My CPU die temp hit about 64C-65C under
full load this past summer.

Ed

Dhyan Ahado wrote:

>> Are all these temps measured with Probe? Probe reports the temp of the
>> socket sensor on the motherboard. MBM5 (latest version) gives both
>> this socket sensor and the CPU-diode. Under stress the diode temp can
>> be a lot higher
>>
>> The highest socket temp I've ever seen with a 2500+, but running @
>> 2200MHz with a raised Vcore of 1.850V is 48C. At that point the diode
>> showed 62C. Maybe your CPU is just running too hot.

> They were - I've now put in MBM5. Diode temp ~55C. On starting Prime5 it
> gets up to 66C before the fan speeds up and pulls it down to ~62C. Still
> fails!
>
> MBM5 reports 5v line at 4.81 - I assume it's reading the same sensor as
> Probe & the BIOS.
>
> On overclockers I saw a comment that anything below 4.9 was asking for
> trouble.

I've seen 5V lines drop to 4.6V with no loss of stability but there's
a first time for everything :-). The A7N8X derives his core voltage
from the 5V line and as long as the Vcore is high enough and doesn't
fluctuate too much there shouldn't be much problems.

Maybe your CPU needs a little more Vcore but your temps are already
quite high. I don't know what hetasink/fan you're using but it looks
like the fan is temperature controlled. Can you fix it so it always
blows at full speed? (Just for testing)

Nanga

In message <(E-Mail Removed)>, Ed
<(E-Mail Removed)> writes
>
> [...]
>FWIW, mine shows 4.892 to 4.919 for +5v in MBM5 and PC Probe, with a
>digital voltage meter on an empty IDE cable I get +4.86 to +4.87 at idle
>and +4.83 to +4.84 with prime95 running. I've had this Antec
>400W/A7N8X/Barton combo for just over 7 months now and it's never
>crashed/rebooted or locked up. My CPU die temp hit about 64C-65C under
>full load this past summer.
>
>Ed
>

Interesting that the external line shows lower than the sensor -
suggests some inaccuracy in the sensor. Then again, given that these
sensors only produce an 8 bit value, a single bit resolution is probably
about the difference.

However does suggest my 4.78 is low. I'd best try a poke at the MB with
the multimeter. I guess the question is, if that is the problem, what
can I do about it? Probably nothing. Bit late to send the board back -
I've been running it since June.

--
Dhyan Ahado
(E-Mail Removed)

In message <(E-Mail Removed)>, Nanga Parbat
<(E-Mail Removed)> writes
>Dhyan Ahado wrote:
>
>>> Are all these temps measured with Probe? Probe reports the temp of the
>>> socket sensor on the motherboard. MBM5 (latest version) gives both
>>> this socket sensor and the CPU-diode. Under stress the diode temp can
>>> be a lot higher
>>>
>>> The highest socket temp I've ever seen with a 2500+, but running @
>>> 2200MHz with a raised Vcore of 1.850V is 48C. At that point the diode
>>> showed 62C. Maybe your CPU is just running too hot.
>
>> They were - I've now put in MBM5. Diode temp ~55C. On starting Prime5 it
>> gets up to 66C before the fan speeds up and pulls it down to ~62C. Still
>> fails!
>>
>> MBM5 reports 5v line at 4.81 - I assume it's reading the same sensor as
>> Probe & the BIOS.
>>
>> On overclockers I saw a comment that anything below 4.9 was asking for
>> trouble.
>
>I've seen 5V lines drop to 4.6V with no loss of stability but there's
>a first time for everything :-). The A7N8X derives his core voltage
>from the 5V line and as long as the Vcore is high enough and doesn't
>fluctuate too much there shouldn't be much problems.

The instability only really shows up in Prime95 (and I suspect
climateprediction). I didn't mention that I pushed Vcore up a couple of
notches - it's reading 1.71 in MBM5.

>
>Maybe your CPU needs a little more Vcore but your temps are already
>quite high. I don't know what hetasink/fan you're using but it looks
>like the fan is temperature controlled. Can you fix it so it always
>blows at full speed? (Just for testing)
>
>Nanga

The heatsink/fan is the standard one coming with the 3200+. Reviews
suggest this is adequate if not overclocking. It's plugged into the MB
which is apparently driving it OK. I haven't currently got an adaptor to
fit the connection - I could make one up but it won't be for a fortnight
as I'm off to Japan tomorrow!


--
Dhyan Ahado
(E-Mail Removed)

On Sat, 27 Dec 2003 02:48:42 +0000, Dhyan Ahado <(E-Mail Removed)>
wrote:

>In message <(E-Mail Removed)>, Ed
><(E-Mail Removed)> writes
>>
> > [...]
>>FWIW, mine shows 4.892 to 4.919 for +5v in MBM5 and PC Probe, with a
>>digital voltage meter on an empty IDE cable I get +4.86 to +4.87 at idle
>>and +4.83 to +4.84 with prime95 running. I've had this Antec
>>400W/A7N8X/Barton combo for just over 7 months now and it's never
>>crashed/rebooted or locked up. My CPU die temp hit about 64C-65C under
>>full load this past summer.
>>
>>Ed
>>
>
>Interesting that the external line shows lower than the sensor -
>suggests some inaccuracy in the sensor. Then again, given that these
>sensors only produce an 8 bit value, a single bit resolution is probably
>about the difference.
>
>However does suggest my 4.78 is low. I'd best try a poke at the MB with
>the multimeter. I guess the question is, if that is the problem, what
>can I do about it? Probably nothing. Bit late to send the board back -
>I've been running it since June.

I had a 5 month old Sparkle 250W PSU I started having troubles with, it
was fine with my CPU at 650MHz but was unstable at anything faster and I
noticed both the 12V and 5V were a bit low, 12v was going down to 11.7-
11.65v, the 5v line wasn't as bad maybe 4.8x something. I opened up the
PSU and adjusted the POTs inside to get both the 12v and 5v as close as
possible to spec while the PC was running, after that I was able to run
the CPU at 800MHz again without lockups.... But I would only go inside a
PSU as a last resort, you could blow up the whole thing and/or
electrocute yourself to death! ;p

Ed

In message <(E-Mail Removed)>, Ed
<(E-Mail Removed)> writes
> [...]
>
>I had a 5 month old Sparkle 250W PSU I started having troubles with, it
>was fine with my CPU at 650MHz but was unstable at anything faster and I
>noticed both the 12V and 5V were a bit low, 12v was going down to 11.7-
>11.65v, the 5v line wasn't as bad maybe 4.8x something. I opened up the
>PSU and adjusted the POTs inside to get both the 12v and 5v as close as
>possible to spec while the PC was running, after that I was able to run
>the CPU at 800MHz again without lockups.... But I would only go inside a
>PSU as a last resort, you could blow up the whole thing and/or
>electrocute yourself to death! ;p
>
>Ed
>

Good idea - I qualified as an electrical engineer originally so I know
how to take reasonable precautions!

It doesn't need to go up much and the external devices should cope with
a slight rise. However I suspect the on board regulators may be pulling
it down so increasing the supply will not actually increase on board
level.

--
Dhyan Ahado
(E-Mail Removed)

On Sat, 27 Dec 2003 11:49:18 +0000, Dhyan Ahado <(E-Mail Removed)>
wrote:

>Good idea - I qualified as an electrical engineer originally so I know
>how to take reasonable precautions!
>
>It doesn't need to go up much and the external devices should cope with
>a slight rise. However I suspect the on board regulators may be pulling
>it down so increasing the supply will not actually increase on board
>level.

I guess you could replace the regulator(s) if you're up to it, or if
they are getting really hot put a heatsink on them to see if that helps
stability.

Cheers,
Ed