[P2B] three questions after

  • Thread starter Thread starter geos
  • Start date Start date
G

geos

hello,

finally I've found a bit of time and modded my p2b-d motherboard with
the support of Paul and P2B posts and instructions. what can I say?
success! :) after 24 hours of choosing the best processors, a little bit
of dancing with solder and conductive paint, I have a "light", easily
removable version of this mod. two p3450@600 run smoothly. thanks guys
for your posts!

I would like to ask about some issues I noticed during this operation.

1) I can see there is a problem with detecting memory which I don't
quite understand. there are two 128MB hynix and one 256MB infineon
memory sticks. when I fill the two first memory slots with 128 sticks
and *third* with 256, bios detects only 384MB. I have to enter and exit
bios with saving changes (although I don't change anything) to have
system detect full capacity on the next reboot. but when I fill the
first two memory slots with 128 sticks and *fourth* with 256 everything
is fine, the bios addresses 512mb. what can it be? it wasn't before when
mobo was running @112MHz. can memory sticks from different producents be
incompatible in some way? (128 sticks are one-sided and 256 is two-sided).

2) the processors slots on my motherboard are a little bit loose and
processors aren't rigidly mounted. until I noticed that it was imposible
to run p3450 @ 600. after creating some sort of "scaffolding" these
processors ran @ 630 with 70% success rate under w2k3ee.. is it possible
that this loose mounting can influence the ability of running processors
with higher frequencies 600MHz and cause no problem with 450MHZ?

3) I don't have any processor terminator for this motherboard. however I
noticed that because of this loose mounting the motherboard detects only
one processor while there are two in the slots. (I wasn't doing that on
purpose, just noticed that as a side effect and was a little bit afraid
of the processors' condition) my question is if it is possible to make a
processor terminator for this motherboard from an unworkable processor?

cheers,
geos
 
geos said:
hello,

finally I've found a bit of time and modded my p2b-d motherboard with
the support of Paul and P2B posts and instructions. what can I say?
success! :) after 24 hours of choosing the best processors, a little bit
of dancing with solder and conductive paint, I have a "light", easily
removable version of this mod. two p3450@600 run smoothly. thanks guys
for your posts!
Click to expand...

Congratulations :-)
I would like to ask about some issues I noticed during this operation.

1) I can see there is a problem with detecting memory which I don't
quite understand. there are two 128MB hynix and one 256MB infineon
memory sticks. when I fill the two first memory slots with 128 sticks
and *third* with 256, bios detects only 384MB. I have to enter and exit
bios with saving changes (although I don't change anything) to have
system detect full capacity on the next reboot. but when I fill the
first two memory slots with 128 sticks and *fourth* with 256 everything
is fine, the bios addresses 512mb. what can it be? it wasn't before when
mobo was running @112MHz. can memory sticks from different producents be
incompatible in some way? (128 sticks are one-sided and 256 is two-sided).
Click to expand...

Unfortunately erratic memory detection is a "feature" of P2B series
boards when running 3 or 4 memory sticks at 124Mhz FSB or higher.
Increasing Vio from 3.2v to 3.4v by changing the value of R50
(http://tipperlinne.com/p2b-dsvio.htm) sometimes helps, but in general
there are only two solutions:

- try your memory in various combinations and positions until you find a
setup that is detected consistently at boot, or

- enter/exit the BIOS on every boot (no need to save). This will
correctly detect three sticks of memory at 133Mhz on almost all boards,
and most will correctly detect four sticks using this trick.

IME it doesn't matter what brand of memory you use, or whether you mix
brands and/or sizes. I normally use Crucial/Micron 256MB sticks, but I
have one P2B-DS board that will only detect 1GB (using BIOS trick) if
the stick in the second slot is a no-name and the rest are Micron. My
primary system runs 768MB (3 x Micron 256MB) @ 140Mhz, it needed
increased Vio to detect them consistently without using the BIOS trick.
2) the processors slots on my motherboard are a little bit loose and
processors aren't rigidly mounted. until I noticed that it was imposible
to run p3450 @ 600. after creating some sort of "scaffolding" these
processors ran @ 630 with 70% success rate under w2k3ee.. is it possible
that this loose mounting can influence the ability of running processors
with higher frequencies 600MHz and cause no problem with 450MHZ?
Click to expand...

Absolutely.

It can be a real PITA to get the CPUs properly seated on these boards,
and it gets worse as they get older. Symptoms range from failure to
POST, through detecting only one CPU when two are installed, to system
freezes while the OS is running. The problem is exacerbated by higher
bus speeds.

First find some cardboard the same thickness as the CPU card edge, soak
it in 99% isopropyl alcohol, and insert it into the CPU slots to clean
them. Repeat until the cardboard comes out clean (the black stuff is
oxidisation, which increases contact resistance). Then use a q-tip
soaked in alcohol to clean the contacts on both sides of the CPU edge
connectors. You can use rubbing alcohol (70%) instead, just make sure
you allow time for the other 30% (water) to evaporate before powering up
the board. You will be surprised how much grunge comes off slots and
CPUs that look clean!

Next, get proper CPU retention mechanisms for your board. Slot-1
processors and adapters will not operate reliably without retention
mechanisms.
3) I don't have any processor terminator for this motherboard. however I
noticed that because of this loose mounting the motherboard detects only
one processor while there are two in the slots. (I wasn't doing that on
purpose, just noticed that as a side effect and was a little bit afraid
of the processors' condition) my question is if it is possible to make a
processor terminator for this motherboard from an unworkable processor?
Click to expand...

No - but the P2B-D/DS will run just fine with an empty CPU slot, you
don't really need a terminator card. IME the only time that's not true
is if you install XP on two processors then remove one - it will freeze
during boot until you put the second CPU back or replace it with a
terminator. I have not seen this behavior under NT4 or W2K.

HTH

P2B
 
about memory! i had same problem and i solve it by buying older memory
double sided! because BX chipset can only recognize double size in full (if
it is 256MB double side it will recognize and if it is 256MB single side or
newer brand it will only recognize 128MB!)
i think that is your problem!
I bought Kingmax memory and it is double sided with 4 chips on each side and
now i have 1GB of memory for almost 2 years and now problem!!!
 
ice said:
about memory! i had same problem and i solve it by buying older memory
double sided! because BX chipset can only recognize double size in full (if
it is 256MB double side it will recognize and if it is 256MB single side or
newer brand it will only recognize 128MB!)
i think that is your problem!
I bought Kingmax memory and it is double sided with 4 chips on each side and
now i have 1GB of memory for almost 2 years and now problem!!!
Click to expand...

hello ice, but that's not the reason. the memory is recognized at full
size at lower frequencies. as P2B explained that "erratic memory
detection is a "feature" of P2B series boards when running 3 or 4 memory
sticks at 124Mhz FSB or higher".

thanks for response,
cheers,
geos
 
P2B said:
Unfortunately erratic memory detection is a "feature" of P2B series
boards when running 3 or 4 memory sticks at 124Mhz FSB or higher.
Increasing Vio from 3.2v to 3.4v by changing the value of R50
(http://tipperlinne.com/p2b-dsvio.htm) sometimes helps, but in general
there are only two solutions:

- try your memory in various combinations and positions until you find a
setup that is detected consistently at boot, or

- enter/exit the BIOS on every boot (no need to save). This will
correctly detect three sticks of memory at 133Mhz on almost all boards,
and most will correctly detect four sticks using this trick.

IME it doesn't matter what brand of memory you use, or whether you mix
brands and/or sizes. I normally use Crucial/Micron 256MB sticks, but I
have one P2B-DS board that will only detect 1GB (using BIOS trick) if
the stick in the second slot is a no-name and the rest are Micron. My
primary system runs 768MB (3 x Micron 256MB) @ 140Mhz, it needed
increased Vio to detect them consistently without using the BIOS trick.
Click to expand...

thanks P2B, that behavior is really strange. after some time of peace
and quiet and recognizing full capacity of memory I plug-off and plug-in
the IDE connector and can not back to the peaceful state, the bios trick
is needed every time now.

I wonder wthat's the reason behind that behavior? what does happen to
the circuits when one just enters and exits the bios and after that the
motherboard can see full memory capacity? can it be avoided by some bios
modification (ie. that influence or bios enrty)?

cheers,
geos
 
geos said:
thanks P2B, that behavior is really strange. after some time of peace
and quiet and recognizing full capacity of memory I plug-off and plug-in
the IDE connector and can not back to the peaceful state, the bios trick
is needed every time now.

I wonder wthat's the reason behind that behavior? what does happen to
the circuits when one just enters and exits the bios and after that the
motherboard can see full memory capacity? can it be avoided by some bios
modification (ie. that influence or bios enrty)?
Click to expand...

Good question - one we have discussed here in the past. Plausible
theories were offered, but none that could be tested to yield a
definitive answer. The BIOS appears to completely re-initialize when you
exit setup, so one would expect the same behavior as at initial power-on.

When you power on, one of the first things the BIOS does is basic
initialisation of the chipset. Next is memory detection. My theory is
the timing of these operations is somehow different at power on vs. exit
setup, such that the memory detection "bug" doesn't occur on setup exit.

I've tried changing just about every available BIOS setting, and even a
few that aren't available (using BIOS patches), but I've never found one
that changes the memory detection behavior.

P2B
 
Back
Top