Crashes when using 2 DIMMs

  • Thread starter Thread starter spasmous2
  • Start date Start date
S

spasmous2

I have two memory DIMMs (DDR333 256MB). When I use either one
separately my system runs fine, however with both together I get
random crashes from 5 minutes to a couple hours apart, independent of
the task. Even running Memtest86 on bootup crashes out, so I dont
think its a Windows 2000 issue.

I was wondering if I should play with the voltage settings to see if
that help. Although I don't know how to - there is no voltage setting
the the BIOS. My system is a HP d330 Pentium IV. The chipset is Intel
865.
 
spasmous2 said:
I have two memory DIMMs (DDR333 256MB). When I use either one
separately my system runs fine, however with both together I get
random crashes from 5 minutes to a couple hours apart, independent of
the task. Even running Memtest86 on bootup crashes out, so I dont
think its a Windows 2000 issue.

I was wondering if I should play with the voltage settings to see if
that help. Although I don't know how to - there is no voltage setting
the the BIOS. My system is a HP d330 Pentium IV. The chipset is Intel
865.

There probably isn't a Vdimm option in the BIOS on a board like that.

Blue Black Blue Black

\________/ \________/

| |

Channel 1 Channel 2

What happens if you put your pair of DIMMs in a blue-black pair of
Channel 1 ? Do you still have a problem ? If you swap the DIMMs, such
that the Black DIMM goes into the Blue slot of Channel 1 and vice versa,
do you still have the same symptoms ? Testing the channels individually,
is a less stressful test for the Northbridge.

When I had a stick of memory fail here, I noticed that the BIOS really
doesn't like defective memory in the lower address space. My system
would start (even though the BIOS is supposed to test the memory), when
the bad memory was up high. But if the defective memory was in the first
640K (which the BIOS might have reserved for its usage), or the memory
was in the area where memtest86+ code loads, then those would be good
reasons for not being able to actually run the memtest program. In my
case, shuffling around my collection of DIMMs, such that the bad memory
was pushed out of harms way, allowed memtest to boot and run (and show
me that a memory chip on one of the DIMMs was completely dead). The single
channel configurations, give you a chance to "move the bad memory around".
(On the motherboard I was debugging, I was actually using three DIMMs
at one point, and playing the shuffle game with them, to try to get
something working.)

When you place matched DIMMs in the two Blue slots (or in the two Black
slots), that is "Dual Channel" mode. That way of doing things is preferred,
since it gives higher memory bandwidth. It is also more demanding of the
Northbridge as well. Trying the "Channel 1" and "Channel 2" tests above,
is not the final configuration I'd want for the motherboard, but those
tests might shed more light on what is broken. I'm hoping this is bad
RAM, because normally the 875/865 family is reasonably good at
stock speeds. The 865PE gets a little upset if you run the memory
at speeds like DDR500, and then you start to see video artifacts
from the video card. But at stock DDR400 or DDR333, it should be
rock solid. I had more trouble with my processor, than anything else.
For some reason, it wouldn't overclock worth a damn.

Paul
 
There probably isn't a Vdimm option in the BIOS on a board like that.

              Blue Black   Blue Black

              \________/   \________/

                   |            |

                Channel 1    Channel 2

What happens if you put your pair of DIMMs in a blue-black pair of
Channel 1 ? Do you still have a problem ? If you swap the DIMMs, such
that the Black DIMM goes into the Blue slot of Channel 1 and vice versa,
do you still have the same symptoms ? Testing the channels individually,
is a less stressful test for the Northbridge.

When I had a stick of memory fail here, I noticed that the BIOS really
doesn't like defective memory in the lower address space. My system
would start (even though the BIOS is supposed to test the memory), when
the bad memory was up high. But if the defective memory was in the first
640K (which the BIOS might have reserved for its usage), or the memory
was in the area where memtest86+ code loads, then those would be good
reasons for not being able to actually run the memtest program. In my
case, shuffling around my collection of DIMMs, such that the bad memory
was pushed out of harms way, allowed memtest to boot and run (and show
me that a memory chip on one of the DIMMs was completely dead). The single
channel configurations, give you a chance to "move the bad memory around"..
(On the motherboard I was debugging, I was actually using three DIMMs
at one point, and playing the shuffle game with them, to try to get
something working.)

When you place matched DIMMs in the two Blue slots (or in the two Black
slots), that is "Dual Channel" mode. That way of doing things is preferred,
since it gives higher memory bandwidth. It is also more demanding of the
Northbridge as well. Trying the "Channel 1" and "Channel 2" tests above,
is not the final configuration I'd want for the motherboard, but those
tests might shed more light on what is broken. I'm hoping this is bad
RAM, because normally the 875/865 family is reasonably good at
stock speeds. The 865PE gets a little upset if you run the memory
at speeds like DDR500, and then you start to see video artifacts
from the video card. But at stock DDR400 or DDR333, it should be
rock solid. I had more trouble with my processor, than anything else.
For some reason, it wouldn't overclock worth a damn.

    Paul


It's the weirdest thing. Last night I put both memory sticks in slots
1 & 3 to confirm the problem by running memtest86. In the morning it
was chugging along - 31 passed tests. So I booted up the OS and it's
been stable all day. This after 2 weeks of crashes...

I am starting to doubt whether they really happened ;)
 
spasmous2 said:
It's the weirdest thing. Last night I put both memory sticks in slots
1 & 3 to confirm the problem by running memtest86. In the morning it
was chugging along - 31 passed tests. So I booted up the OS and it's
been stable all day. This after 2 weeks of crashes...

I am starting to doubt whether they really happened ;)

It might be as simple as a stray cornflake (or other breakfast cereal)
in one of the slots. When you tested with just one module, did you
always put it in the same slot?
 
It might be as simple as a stray cornflake (or other breakfast cereal)
in one of the slots.  When you tested with just one module, did you
always put it in the same slot?

Nope, I was really inconsistent. It pays to be careful first time
round huh! It must have been a cornflake. All is well now - thanks.
 
Back
Top