Almost random system freezes - SUSE10.3 - Diagnostic Help ..

David Haggett wrote:

> On Sat, 09 Feb 2008 19:14:49 +0000, Baron wrote:
>
>>> Not sure about the RAM ones, but none of the components on the board
>>> appear to be swollen. There is one (in a row of 8 next to the CPU,
>>> just behind the edge of the heat-sink that isn't on the board
>>> straight, but it never was, even when I built the PC.
>>
>> Those are the CPU PSU capacitors. I can tel from your description
>> that you have an Intel based mainboard. The caps on those are
>> usually quite robust.

When I said "Intel" based, I meant Intel CPU rather than Intel chipset
based.

> That's good to hear... but if I told you it was an MSI motherboard
> based on a VIA K8T800 chipset, would that change your evaluation? It
> is a good board, with 2 built-in SATA controllers, and would really
> like to keep it.

Microstar have made some good and some bad boards. At the moment they
are at the bottom end of the market. My personal preference is
for "Gigabyte" In general they are good. But like others they have
turned out some bad ones. Now "PC Chips" is one to avoid like the
plague!

>>> You may be right - although as I said, this one isn't old (about 6
>>> months) and wasn't cheap either.

Out of interest, where are you based?

>> Based on the additional info. Try a new PSU.
>
> I'll do that. Has to be a quiet one, though. If it does turn out to
> be the PSU, is there any way of proving it (thinking refund).

Well if it solves your issue that should be proof enough. But I do get
your drift. Local white box merchants tend not to want to know once
they have got your money!

--
Best Regards:
Baron.

On Sat, 09 Feb 2008 23:51:30 +0000, Baron wrote:


>
> Out of interest, where are you based?

I'm in Surrey.

>>> Based on the additional info. Try a new PSU.
>>
>> I'll do that. Has to be a quiet one, though. If it does turn out to
>> be the PSU, is there any way of proving it (thinking refund).
>
> Well if it solves your issue that should be proof enough. But I do get
> your drift. Local white box merchants tend not to want to know once
> they have got your money!

Actually, just found something interesting, but don't know what it
implies. Output of sensors:

VCore: +1.49 V (min = +0.00 V, max = +1.09 V) ALARM
+3.3V: +3.34 V (min = +2.83 V, max = +1.28 V) ALARM
+5V: +5.03 V (min = +0.08 V, max = +0.43 V) ALARM
+12V: +11.98 V (min = +5.84 V, max = +0.49 V) ALARM
-12V: -12.11 V (min = -13.59 V, max = -6.93 V)
-5V: >>> +3.54 V (min = +2.39 V, max = -7.31 V) <<<<< ALARM
V5SB: +5.48 V (min = +0.03 V, max = +0.03 V) ALARM
VBat: +3.54 V (min = +0.38 V, max = +2.24 V) ALARM

As a layman, that -5V rail doesn't look quite right to me. Any opinion?

--
David Haggett
Linux user since 01/01/2003
Email: david<at>haggett<dot>demon<dot>co<dot>uk

On Tue, 12 Feb 2008 10:04:14 +0000, Baron wrote:

>> As a layman, that -5V rail doesn't look quite right to me. Any
>> opinion?
>
> I agree completely ! But I am also concerned about the bracketed
> figures. They just arn't right. Could be that the bit of software you
> are using has not been setup properly ?

I wouldn't rule out the possibility. I just ran sensors-detect, and used
its defaults. On the other hand, the BIOS agrees with the sensor (sort
of - it claims -5V is NONE suggesting).

Regarding VCore, my CPU is an AMD Athlon64 3200+ (I should have mentioned
it earlier in the thread).

I'm buying a new CPU, and the m/c is staying off until its fixed.


--
David Haggett
Linux user since 01/01/2003
Email: david<at>haggett<dot>demon<dot>co<dot>uk

On Tue, 12 Feb 2008 10:04:14 +0000, Baron wrote:

>> As a layman, that -5V rail doesn't look quite right to me. Any
>> opinion?
>
> I agree completely ! But I am also concerned about the bracketed
> figures. They just arn't right. Could be that the bit of software you
> are using has not been setup properly ?

I wouldn't rule out the possibility. I just ran sensors-detect, and used
its defaults. On the other hand, the BIOS agrees with the sensor (sort of
- it claims -5V is NONE suggesting that either sensors, or the BIOS can't
handle the true value).

Regarding VCore, my CPU is an AMD Athlon64 3200+ (I should have mentioned
it earlier in the thread).

I'm buying a new PSU, and the m/c is staying off until its fixed.


--
David Haggett
Linux user since 01/01/2003
Email: david<at>haggett<dot>demon<dot>co<dot>uk

David Haggett wrote:

> On Tue, 12 Feb 2008 10:04:14 +0000, Baron wrote:
>
>>> As a layman, that -5V rail doesn't look quite right to me. Any
>>> opinion?
>>
>> I agree completely ! But I am also concerned about the bracketed
>> figures. They just arn't right. Could be that the bit of software
>> you are using has not been setup properly ?
>
> I wouldn't rule out the possibility. I just ran sensors-detect, and
> used its defaults. On the other hand, the BIOS agrees with the sensor
> (sort of - it claims -5V is NONE suggesting).

Not all boards use -5v. There isn't one coming out of the PSU. There
is a -12v which may be used to provide it. Thats why you need to
confirm those voltages with a meter. Having said that, I don't have a
clue where you would find that one.

> Regarding VCore, my CPU is an AMD Athlon64 3200+ (I should have
> mentioned it earlier in the thread).

In that case 1.49 is close enough to 1.45 not to worry about.

> I'm buying a new CPU, and the m/c is staying off until its fixed.

Nothing wrong with the 3200+. Mine is a 3000+ I've played with a 4600+
and the difference is not great for the extra money.

But I can understand you holding off until you get a new PSU!

--
Best Regards:
Baron.

On Tue, 12 Feb 2008 20:55:30 +0000, Baron wrote:

>> I'm buying a new CPU, and the m/c is staying off until its fixed.
^^^ meant PSU

> Nothing wrong with the 3200+. Mine is a 3000+ I've played with a 4600+
> and the difference is not great for the extra money.

Obviously My ISP didn't honour my supersede


--
David Haggett
Linux user since 01/01/2003
Email: david<at>haggett<dot>demon<dot>co<dot>uk

David Haggett wrote:

> On Tue, 12 Feb 2008 20:55:30 +0000, Baron wrote:
>
>>> I'm buying a new CPU, and the m/c is staying off until its fixed.
> ^^^ meant PSU

Ah, right, typo Trying to confuse her indoors ! Clever..

>> Nothing wrong with the 3200+. Mine is a 3000+ I've played with a
>> 4600+ and the difference is not great for the extra money.
>
> Obviously My ISP didn't honour my supersede
>
No ! Mine doesn't either.

--
Regards:
Baron.

On Tue, 12 Feb 2008 20:55:30 +0000, Baron wrote:

Have installed a new PSU, and have started the system from cold to
graphical login twice without a system hang.

My previous supply was a 520W model with 4 12V rails capable of
delivering 16A each. This supply is only a 500W model, but with 2 rails
capable of delivering 18A each. I also made sure I put the graphics card
onto its own rail.

> Not all boards use -5v. There isn't one coming out of the PSU. There
> is a -12v which may be used to provide it.

I looked it up in the motherboard manual - which seems to suggest the
board expects -5V to be delivered on pin 18 of the 20pin connector.
Doesn't say what it is used for, but I suspect it might not be needed for
this board either - I can't recall whether my original PSU had one, but I
suspect not.

Anyway, hoping that the fix is permanent, thanks for help and advice.


--
David Haggett
Linux user since 01/01/2003
Email: david<at>haggett<dot>demon<dot>co<dot>uk

This message is not archived

Hi David,

David Haggett wrote:
> On Tue, 12 Feb 2008 20:55:30 +0000, Baron wrote:
>
> Have installed a new PSU, and have started the system from cold to
> graphical login twice without a system hang.

That sounds very very promising.

> My previous supply was a 520W model with 4 12V rails capable of
> delivering 16A each. This supply is only a 500W model, but with 2
> rails capable of delivering 18A each. I also made sure I put the
> graphics card onto its own rail.

If you take a look inside you will find that all those rails are wired
to the same points ! 12v to 12v, 5v to 5v etc. The fact that there are
less "rails" (your word) is not relevant unless you run out of power
connections.

>> Not all boards use -5v. There isn't one coming out of the PSU.
>> There is a -12v which may be used to provide it.
>
> I looked it up in the motherboard manual - which seems to suggest the
> board expects -5V to be delivered on pin 18 of the 20pin connector.

Pin 18 is marked as reserved in the ATX Design Guide.

> Doesn't say what it is used for, but I suspect it might not be needed
> for this board either - I can't recall whether my original PSU had
> one, but I suspect not.
>
> Anyway, hoping that the fix is permanent, thanks for help and advice.

Hey I'm glad that it solves, what is for most, a nasty issue.

--
Best Regards:
Baron.