Memtest errors?

[Richard Dower]

I bought sme new Corsair PC3200XLPT TwinX 1GB, and i used Memtest86+, i get
errors on test 3 on one of the modules and errors on test 6 on the other
module.

But how serious are Memtest errors?, does your memory have to pass all
Memtests with zero errors for it to work properly?

 

[Apollo]

I bought sme new Corsair PC3200XLPT TwinX 1GB, and i used Memtest86+,
i get errors on test 3 on one of the modules and errors on test 6 on
the other module.

But how serious are Memtest errors?, does your memory have to pass all
Memtests with zero errors for it to work properly?

If you want your system to be stable then yes, you risk data corruption,
BSOD's etc.

What mobo, vdimm, memory timings? are you overclocking?

Increasing vdimm / relaxing timings will probably cure the errors, but
if you're not overclocking you should be able to run at spd settings.

Provide more info about your setup, I have the same mem running at
220MHz 5,2,2,2, but it needs 3.0 volts of vdimm to get there.

 

[Richard Dower]

If you want your system to be stable then yes, you risk data corruption,
BSOD's etc.

What mobo, vdimm, memory timings? are you overclocking?

Increasing vdimm / relaxing timings will probably cure the errors, but if
you're not overclocking you should be able to run at spd settings.

Provide more info about your setup, I have the same mem running at 220MHz
5,2,2,2, but it needs 3.0 volts of vdimm to get there.

I am running a 3.2GHz CPU, Northwood, when installing the new memory i set
everything to default...no overclocking. My motherboard is a Gigabyte
GA-8KNXP revision 2.1 with the latest FJ bios. Antec 550W PSU, leadtek
6800GT at default settings. Two Western Digital Raptors 74GB running in RAID
0, Audigy 2 ZS.

I set the voltage to 2.8v, thus increasing by +.3v from the default 2.5v, i
set the memory to detect by SPD data. When it booted up i got memory timings
by SPD of 2.5-2-2-5.

Then i started the Memtest86+ runs, i am using the latest 1.26 version.
Right away on test 3 i got errors. Tried everything, changing the modules to
different DIMM slots, i tried the modules in both dual channel mode and as
individual modules.

The first module start having errors at test 3 of the standard test, i
repeated the test many times and got errors in the same place. I took that
module out and tested the other one, i started having errors at test 6.

I tried changing the voltage from 2.5v - 2.8v, still errors. I relaxed the
memory timings and still got errors.

Thus i am at a loss, i contacted Corsair and they gave me an RMA number,
it's possible that it was just a bad batch or something....i asked them to
test the modules and replace them if nesscary.

 

[Apollo]

I am running a 3.2GHz CPU, Northwood, when installing the new memory i
set everything to default...no overclocking. My motherboard is a
Gigabyte GA-8KNXP revision 2.1 with the latest FJ bios. Antec 550W
PSU, leadtek 6800GT at default settings. Two Western Digital Raptors
74GB running in RAID 0, Audigy 2 ZS.

It's been a long long time since I built with Intel, but that rig sounds
nice ;-)

I set the voltage to 2.8v, thus increasing by +.3v from the default
2.5v, i set the memory to detect by SPD data. When it booted up i got
memory timings by SPD of 2.5-2-2-5.

Then i started the Memtest86+ runs, i am using the latest 1.26
version. Right away on test 3 i got errors. Tried everything, changing
the modules to different DIMM slots, i tried the modules in both dual
channel mode and as individual modules.

The first module start having errors at test 3 of the standard test, i
repeated the test many times and got errors in the same place. I took
that module out and tested the other one, i started having errors at
test 6.

Looks like you tested them as fully as possible, I was going to ask if
you were getting the same error for the same dimm in different slots.

I tried changing the voltage from 2.5v - 2.8v, still errors. I relaxed
the memory timings and still got errors.

They should run fine at default voltage for stock speed and timings,
something is definately wrong.

Thus i am at a loss, i contacted Corsair and they gave me an RMA
number, it's possible that it was just a bad batch or something....i
asked them to test the modules and replace them if nesscary.

Best course of action really, you tested them correctly, but it does
seem odd that two modules would both be defective. Have you looked
through google groups searching on that mobo / memory combination,
perhaps someone else has had the same problem as yourself.

Hope you get it going, (then you can look at overclocking it ).

 

[Dave]

Richard,

Don't ever assume that a Gigabyte board is correctly reading the SPD on the
modules. I've read multiple reviews, and I have personal experience with
two Gigabyte boards, where the instituted timings are divergent from the
SPD. Even the latest Gigabyte board I used, supposedly their top of the
line, wouldn't allow manula settings and gave me 2.5 4 4 15(!) at some
frequencies, but never anywhere near the SPD of 2 3 2 5. Set the timings
manually.

Start at 2.5 3 3 7 (if it can't go this low, then I'd return them out of
principle) with the same 2.8 voltage, at the rated frequency, and work your
way down, one number at a time, until you get errors; then move the number
up one to its error free state. If you're still getting errors, it is
possible that you got TWO defective sticks, but unlikely.

 

[Richard Dower]

Richard,

Don't ever assume that a Gigabyte board is correctly reading the SPD on
the modules. I've read multiple reviews, and I have personal experience
with two Gigabyte boards, where the instituted timings are divergent from
the SPD. Even the latest Gigabyte board I used, supposedly their top of
the line, wouldn't allow manula settings and gave me 2.5 4 4 15(!) at some
frequencies, but never anywhere near the SPD of 2 3 2 5. Set the timings
manually.

Start at 2.5 3 3 7 (if it can't go this low, then I'd return them out of
principle) with the same 2.8 voltage, at the rated frequency, and work
your way down, one number at a time, until you get errors; then move the
number up one to its error free state. If you're still getting errors, it
is possible that you got TWO defective sticks, but unlikely.

It is possible, thing is....right now i have 1GB of Corsair PC3200LLPT
2-3-2-6, i have manually set the timings on these modules to 2-3-2-5 and
they work perfect!

Have tested many times with the Windows memory diagnostic and Mestest86+ and
they pass all tests, every time.

Even these modules gave me 2.5-3-2-6 by SPD, so i changed them to 2-3-2-5.
My 1GB of XL modules gave me 2.5.-2-2-5 by SPD at 2.8v

my PC booted with the XL modules, no problem there, the issue was within
memtest86 and the errors i received from the diagnostic utiltity. Thus i
have returned the XL modules to Corsair for testing and possible
replacement.

 

[Bob Davis]

Don't ever assume that a Gigabyte board is correctly reading the SPD on
the modules. I've read multiple reviews, and I have personal experience
with two Gigabyte boards, where the instituted timings are divergent from
the SPD. Even the latest Gigabyte board I used, supposedly their top of
the line, wouldn't allow manula settings and gave me 2.5 4 4 15(!) at some
frequencies, but never anywhere near the SPD of 2 3 2 5. Set the timings
manually.

Where do you get the specs for a given model. I have four sticks of
Kingston KVR400X64C3A/512 and have never seen any references to
specifications for setting the modules manually. I have them set to SPD now
and all is stable.