Memory testing with Memtest - question

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

Hi!

I syspect some of my RAMs are defected. I'm running Memtest-86 v3.1 and it's
running 5,5 hours and found 11 errors until now.
I would like to know if I'll see when the test is finished because there are
no signs of time left or sg. like that.
Also, will it report exactly witch module is damaged?
Thanks a lot for helping!

bence

 

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

On Wed, 19 May 2004 07:33:44 GMT, "Locsmándi Bence"
<locsmandi.bence@chello.hu> wrote:

>Hi!
>
>I syspect some of my RAMs are defected. I'm running Memtest-86 v3.1 and it's
>running 5,5 hours and found 11 errors until now.
>I would like to know if I'll see when the test is finished because there are
>no signs of time left or sg. like that.
>Also, will it report exactly witch module is damaged?
>Thanks a lot for helping!
>
>bence
>

There is no "finished", it runs an infinite loop, forever.

You didn't need to keep running it though, once it's found errors the test
has run long enough to stop and make some changes. It will not tell you
which module is damaged, but do not assume that one is damaged unless you
ALWAYS see the exact same address showing an error. If it ran for 5.5
hours and only found 11 errors it may be more likely that it was just a
random error, unless there was so much memory that it only had a chance to
run 5 passes. By random error I mean a marginal memory module or
motherboard or bios setting, it can be one of several factors causing
different address errors besides an actual "damaged" module.

Remove all but one module and retest, then swap in the other modules and
test again. Check bios setup for memory-related settings. Return those
to default values if they weren't there already. Visit a web forum or
the manufacturer's website to find recommendations for memory or solutions
if the board is particularly picky about memory. Sometimes a bios update
will address memory stability.

Since you didn't mention your specific memory or motherboard we can only
make generic suggestions.

 

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

Well, thanks for the in-depth answer and sorry not to mention the
parameters.
ASUS P4C800 motherboard (latest BIOS)
P4 3 GHz CPU
4 x 512 MB 400 MHz PC3200 DDR RAM Samsung.

The test is running for 7 hours now and found 12 errors. All of them found
in test 4 (dunno if that means to you sg) and every Failing Addresses are
different. You read at every errors:
Good Bad
bfffffff bfefffff
Hope that it helps to go on.
Thanks again

bence

"kony" <spam@spam.com> wrote in message
news:a65ma01pdgn2irrsrcr0ffhj0smhtlonvt@4ax.com...
> On Wed, 19 May 2004 07:33:44 GMT, "Locsmándi Bence"
> <locsmandi.bence@chello.hu> wrote:
>
> >Hi!
> >
> >I syspect some of my RAMs are defected. I'm running Memtest-86 v3.1 and
it's
> >running 5,5 hours and found 11 errors until now.
> >I would like to know if I'll see when the test is finished because there
are
> >no signs of time left or sg. like that.
> >Also, will it report exactly witch module is damaged?
> >Thanks a lot for helping!
> >
> >bence
> >
>
> There is no "finished", it runs an infinite loop, forever.
>
> You didn't need to keep running it though, once it's found errors the test
> has run long enough to stop and make some changes. It will not tell you
> which module is damaged, but do not assume that one is damaged unless you
> ALWAYS see the exact same address showing an error. If it ran for 5.5
> hours and only found 11 errors it may be more likely that it was just a
> random error, unless there was so much memory that it only had a chance to
> run 5 passes. By random error I mean a marginal memory module or
> motherboard or bios setting, it can be one of several factors causing
> different address errors besides an actual "damaged" module.
>
> Remove all but one module and retest, then swap in the other modules and
> test again. Check bios setup for memory-related settings. Return those
> to default values if they weren't there already. Visit a web forum or
> the manufacturer's website to find recommendations for memory or solutions
> if the board is particularly picky about memory. Sometimes a bios update
> will address memory stability.
>
> Since you didn't mention your specific memory or motherboard we can only
> make generic suggestions.

 

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

Locsmándi Bence wrote:

> Hi!
>
> I syspect some of my RAMs are defected. I'm running Memtest-86 v3.1 and it's
> running 5,5 hours and found 11 errors until now.
> I would like to know if I'll see when the test is finished because there are
> no signs of time left or sg. like that.
> Also, will it report exactly witch module is damaged?


No need to continue running the test. If you have more than one RAM
module, remove one and run memtest86 again until you've determined which
stick is defective.

If you can't pinpoint the problem in that manner, then perhaps there is
a problem with your motherboard or your memory timings are too aggressive.


-WD

 

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

Thanks for the answer, Will!
I ran the test again with only 1 module in the PC at a time. No defect.
If I think right memtest runs 7 kinda tests and then it restarts with test
1, etc.
How do you mean "your memory timings are too aggressive"? All the possible
settings are set to Auto.

b.

"Will Dormann" <wdormann@yahoo.com.invalid> wrote in message
news:T9Nqc.3329$Uq1.2457@fe2.columbus.rr.com...
> Locsmándi Bence wrote:
>
> > Hi!
> >
> > I syspect some of my RAMs are defected. I'm running Memtest-86 v3.1 and
it's
> > running 5,5 hours and found 11 errors until now.
> > I would like to know if I'll see when the test is finished because there
are
> > no signs of time left or sg. like that.
> > Also, will it report exactly witch module is damaged?
>
>
> No need to continue running the test. If you have more than one RAM
> module, remove one and run memtest86 again until you've determined which
> stick is defective.
>
> If you can't pinpoint the problem in that manner, then perhaps there is
> a problem with your motherboard or your memory timings are too aggressive.
>
>
> -WD

 

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

Locsmándi Bence wrote:

> Thanks for the answer, Will!
> I ran the test again with only 1 module in the PC at a time. No defect.
> If I think right memtest runs 7 kinda tests and then it restarts with test
> 1, etc.
> How do you mean "your memory timings are too aggressive"? All the possible
> settings are set to Auto.

Either slow down the timings manually in the BIOS or decrease the clock
speed for the RAM. I've heard of people having trouble with your
particular board being able to run the RAM at full speed successfully.



-WD

 

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

Well, I understand now what you mean but is there a way in my mobo's BIOS to
do that? What do you mean "timing"?
I see these available settings under "Chipset" BIOS menu item:

Configure DRAM Timing by SPD? [Enabled]
Performance Acceleration Mode? [Enabled]
DRAM Idle Timer? [Auto]
DRAM Refresh Mode? [Auto]
- DRAM CAS# Latency? [these are not visible since the above items are set to
"Auto"]
- DRAM RAS# Precharge?
- DRAm RAS# to CAS# Delay?
- DRAM Precharge Delay?
- DRAM Burst Lenght?
(all of these parameters don't mean too much to me, eg. I don't understand
what they set)

Anyway, I haven't mentioned so far that this PC works as an audio
workstation. No games, no internet, etc. Maybe it counts for you to go on
succesfully.
Well, today I found that my daw (PC) has 2 types of RAMs. 2 pairs. Each one
have many of the same chips on them (Samsung K4H560838E-TCCC), but 2 pieces
of them seem to be ECC "enabled" since they have 1-1 small more chips on
them. Hope it's clear. Tried to change the order of them (not forgotten that
the same ones have to be installed in the blue or black sockets, as the
manual says).
Thanks, really, for trying me to discover this problem. Since it's a
resource hungry DAW (audio worksation) it is really important for me to be
able to use Hyperthreading with my CPU.
Best regards

bence

"Will Dormann" <wdormann@yahoo.com.invalid> wrote in message
news:5TQqc.8$ea6.3@fe1.columbus.rr.com...
> Locsmándi Bence wrote:
>
> > Thanks for the answer, Will!
> > I ran the test again with only 1 module in the PC at a time. No defect.
> > If I think right memtest runs 7 kinda tests and then it restarts with
test
> > 1, etc.
> > How do you mean "your memory timings are too aggressive"? All the
possible
> > settings are set to Auto.
>
> Either slow down the timings manually in the BIOS or decrease the clock
> speed for the RAM. I've heard of people having trouble with your
> particular board being able to run the RAM at full speed successfully.
>
>
>
> -WD

 

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

Locsmándi Bence wrote:

> Well, I understand now what you mean but is there a way in my mobo's BIOS to
> do that? What do you mean "timing"?
> I see these available settings under "Chipset" BIOS menu item:

From another googled post:

-----
Mushkin tech support recommended the following BIOS changes:


AI Overclock Tuner changed from "Standard" to "Manual"
which allowed me to change the DDR reference voltage to 2.75 from "Auto"

then in Chipset, Advanced

disable "Configure DRAM timing by SPD

which allowed me to change the following:

CAS = 2
RAS to CAS = 3
RAS =2
precharge delay = 6

disable "Spread Spectrum"

I had also previously disabled "Legacy USB Support"
------


-WD

 

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

Locsmándi Bence wrote:

...
> Well, today I found that my daw (PC) has 2 types of RAMs. 2 pairs. Each one
> have many of the same chips on them (Samsung K4H560838E-TCCC), but 2 pieces
> of them seem to be ECC "enabled" since they have 1-1 small more chips on
> them. Hope it's clear. Tried to change the order of them (not forgotten that
> the same ones have to be installed in the blue or black sockets, as the
> manual says).
....

You might really wish to investigate this point further, as not all
systems work with mixed ECC and non-ECC memory modules - you may also
wish to consult Dr Google if you are in doubt... or your mainboard manual.

Roy


--
This posting is provided "AS IS" with no warranties
and confers no rights.

 

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

Hi Ray!
Thanks for answering!
I don't think my prob comes from the ECC-nonECC issue since I tried to
install only the ECC modules and also the nonECC modules alone, I mean no
mixed ECC and nonECC modules in the PC at the same time. The problem was the
same.

bence

"Roy Coorne" <rcoorne@hotmail.com> wrote in message
news:2h2sgjF8e1hgU1@uni-berlin.de...
> Locsmándi Bence wrote:
>
> ..
> > Well, today I found that my daw (PC) has 2 types of RAMs. 2 pairs. Each
one
> > have many of the same chips on them (Samsung K4H560838E-TCCC), but 2
pieces
> > of them seem to be ECC "enabled" since they have 1-1 small more chips on
> > them. Hope it's clear. Tried to change the order of them (not forgotten
that
> > the same ones have to be installed in the blue or black sockets, as the
> > manual says).
> ...
>
> You might really wish to investigate this point further, as not all
> systems work with mixed ECC and non-ECC memory modules - you may also
> wish to consult Dr Google if you are in doubt... or your mainboard manual.
>
> Roy
>
>
> --
> This posting is provided "AS IS" with no warranties
> and confers no rights.

 

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

Hi Will!

Tried these settings but PC won't start.

b.

"Will Dormann" <wdormann@yahoo.com.invalid> wrote in message
news:NVVqc.137$eO6.15@fe2.columbus.rr.com...
> Locsmándi Bence wrote:
>
> > Well, I understand now what you mean but is there a way in my mobo's
BIOS to
> > do that? What do you mean "timing"?
> > I see these available settings under "Chipset" BIOS menu item:
>
> From another googled post:
>
> -----
> Mushkin tech support recommended the following BIOS changes:
>
>
> AI Overclock Tuner changed from "Standard" to "Manual"
> which allowed me to change the DDR reference voltage to 2.75 from "Auto"
>
> then in Chipset, Advanced
>
> disable "Configure DRAM timing by SPD
>
> which allowed me to change the following:
>
> CAS = 2
> RAS to CAS = 3
> RAS =2
> precharge delay = 6
>
> disable "Spread Spectrum"
>
> I had also previously disabled "Legacy USB Support"
> ------
>
>
> -WD

 

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

Well, it was my fault, now it starts. Slight improvement.
Thanks.

b.

"Locsmándi Bence" <locsmandi.bence@chello.hu> wrote in message
newsa_qc.134956$O9.1256@news.chello.at...
> Hi Will!
>
> Tried these settings but PC won't start.
>
> b.
>
> "Will Dormann" <wdormann@yahoo.com.invalid> wrote in message
> news:NVVqc.137$eO6.15@fe2.columbus.rr.com...
> > Locsmándi Bence wrote:
> >
> > > Well, I understand now what you mean but is there a way in my mobo's
> BIOS to
> > > do that? What do you mean "timing"?
> > > I see these available settings under "Chipset" BIOS menu item:
> >
> > From another googled post:
> >
> > -----
> > Mushkin tech support recommended the following BIOS changes:
> >
> >
> > AI Overclock Tuner changed from "Standard" to "Manual"
> > which allowed me to change the DDR reference voltage to 2.75 from "Auto"
> >
> > then in Chipset, Advanced
> >
> > disable "Configure DRAM timing by SPD
> >
> > which allowed me to change the following:
> >
> > CAS = 2
> > RAS to CAS = 3
> > RAS =2
> > precharge delay = 6
> >
> > disable "Spread Spectrum"
> >
> > I had also previously disabled "Legacy USB Support"
> > ------
> >
> >
> > -WD
>
>

 

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

Locsmándi Bence wrote:

> Well, it was my fault, now it starts. Slight improvement.
> Thanks.

Ok then you can try my earlier recommendation to run the RAM at a
reduced clock speed.


-WD

 

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

In article <8x2rc.79$ea6.67@fe1.columbus.rr.com>,
wdormann@yahoo.com.invalid says...
> Locsmándi Bence wrote:
>
> > Well, it was my fault, now it starts. Slight improvement.
> > Thanks.
>
> Ok then you can try my earlier recommendation to run the RAM at a
> reduced clock speed.
>

On an older nForce2 board (A7N8X) I backed off from CL
2.0 to CL 2.5... (yet another thing to try).

 

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

Toshi1873 wrote:

> In article <8x2rc.79$ea6.67@fe1.columbus.rr.com>,
> wdormann@yahoo.com.invalid says...
>
>>Locsmándi Bence wrote:
>>
>>
>>>Well, it was my fault, now it starts. Slight improvement.
>>>Thanks.
>>
>>Ok then you can try my earlier recommendation to run the RAM at a
>>reduced clock speed.
>>
>
>
> On an older nForce2 board (A7N8X) I backed off from CL
> 2.0 to CL 2.5... (yet another thing to try).

Also good advice, Toshi. It does fall into the category of my
suggested "less aggressive memory timings", but thanks for spelling it
out for the OP.


-WD

 

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

You mean CAS Latency?

b.

"Toshi1873" <toshi1873@nowhere.com> wrote in message
news:MPG.1b16a2c428f2ad209898f7@news-50.giganews.com...
In article <8x2rc.79$ea6.67@fe1.columbus.rr.com>,
wdormann@yahoo.com.invalid says...
> Locsmándi Bence wrote:
>
> > Well, it was my fault, now it starts. Slight improvement.
> > Thanks.
>
> Ok then you can try my earlier recommendation to run the RAM at a
> reduced clock speed.
>

On an older nForce2 board (A7N8X) I backed off from CL
2.0 to CL 2.5... (yet another thing to try).

 

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

In article <_T4rc.138962$O9.134700@news.chello.at>,
locsmandi.bence@chello.hu says...
> You mean CAS Latency?
>

Yes, CL 2.0 / CL 2.5 are CAS Latency settings.