How many passes for memtest86? (Full Version)

All Forums >> [Computer Hardware] >> Memory



Message

Bluehabit -> How many passes for memtest86? (4/10/2009 1:24:12 PM)

If you are trying to rule out RAM for the reason your latest OC is failing how many passes would you advice per stick? Im on pass 2 so far and no errors. 



daviangel -> RE: How many passes for memtest86? (4/10/2009 9:04:39 PM)

I usually only run once cuz in the past that's all it usually takes memtest to find bad ram--it shows up pretty quick in my experience anyways!




=Hellspawn= -> RE: How many passes for memtest86? (4/10/2009 9:18:01 PM)

right if you have a problem it usually finds it fast.... the memtest is extremely demanding...


if you are on the 2nd pass with no errors... you are good..[;)]



chevell -> RE: How many passes for memtest86? (4/11/2009 9:37:04 AM)

Actually it can take 20 hours or more to show an error on Ram that is bad and that's if you are only testing one stick at a time. Not to mention that Memtest often fails to show errors on Ram that is known to be bad.  It's free software and it's not perfect but it's the best one around.



NordicJedi -> RE: How many passes for memtest86? (4/11/2009 4:19:09 PM)

Typically, 4 passes is the mark for memtest.

If your OC is failing, there are better ways to test why you can't get stable. Use Prime 95 small-fft's test to rule out CPU error, and use either the blend test or Large In Place fft's test to rule out memory.





daviangel -> RE: How many passes for memtest86? (4/12/2009 4:52:22 AM)


quote:

ORIGINAL: chevell

Actually it can take 20 hours or more to show an error on Ram that is bad and that's if you are only testing one stick at a time. Not to mention that Memtest often fails to show errors on Ram that is known to be bad.  It's free software and it's not perfect but it's the best one around.

Yeah you need a memory hardware tester(like rst pro2) to be absolutely sure but as this review shows 1 run of memtest(make sure you are using the latest version) pretty much give the same result as the hardware tester.
"We considered memory stable when it was able to run 1 extended pass of Memtest86+, twelve hours of Prime95/3dmark testing and ran without locking up while running all tests on the RST Pro2. People that work in the memory industry feel that if a single error is shown during RST Pro2 testing that the memory has failed testing and needs to be fixed."
http://www.legitreviews.com/article/204/3/



chevell -> RE: How many passes for memtest86? (4/12/2009 11:22:23 AM)

quote:

ORIGINAL: daviangel


quote:

ORIGINAL: chevell

Actually it can take 20 hours or more to show an error on Ram that is bad and that's if you are only testing one stick at a time. Not to mention that Memtest often fails to show errors on Ram that is known to be bad.  It's free software and it's not perfect but it's the best one around.

Yeah you need a memory hardware tester(like rst pro2) to be absolutely sure but as this review shows 1 run of memtest(make sure you are using the latest version) pretty much give the same result as the hardware tester.
"We considered memory stable when it was able to run 1 extended pass of Memtest86+, twelve hours of Prime95/3dmark testing and ran without locking up while running all tests on the RST Pro2. People that work in the memory industry feel that if a single error is shown during RST Pro2 testing that the memory has failed testing and needs to be fixed."
http://www.legitreviews.com/article/204/3/



Right you are about using the hardware based memory testing equipment.

I'am glad to see that somebody has spent as many hours as I have reading about memory. But the single pass about Memtest is absolute BS man.

Testing for 20 hours on a single stick is in fact listed in many fact sheets about using memtest. And yes, even if Ram passes the 20 hour mark it can still be bad! If people would take the time to research this subject you would see that I'am right. I spent one month straight reading the facts about Memtest. I suggest that everyone here google the subject and you will see what I mean.
 
Unfortuneately, those expensive memory testers aren't within reach of the average hobbiest so were all stuck using free software which isn't exactly perfect.

I go by the tried and true method. If a computer locks up after trying all adjustments, even if Memtest says the Ram is good It's not!
 
I have personally fixed many computers using this philosophy.



daviangel -> RE: How many passes for memtest86? (4/12/2009 12:59:03 PM)

Oh I just wanted to add that the cheapest I've seen that memory hardware tester is about $900US on ebay to give people an idea.




=Hellspawn= -> RE: How many passes for memtest86? (4/13/2009 9:18:00 AM)

after a bit of reshearch.... I stand corrected.. looks like 3-4 passes is the norm..



chevell -> RE: How many passes for memtest86? (4/13/2009 10:48:11 AM)

3-4 passes is good for testing an overclock I agree. But if someone attempts to look for a bad stick of Ram using only 3-4 passes you could still easily have bad Ram getting past a 3-4 pass test.

I see a bunch of guys with freezing problems run a few passes with all sticks in and then conclude the Ram is good. Unfortunately, an incomplete testing procedure like that won't fix anything.

I remember when I ran Memtest boot CD for 25 passes which my old Ram passed but I still had freezing problems so the Ram didn't work right and had to be replaced. Although Memtest for Windows and Prime blend would freeze up which were my best clues.

When I say 20 passes I mean on a single stick using the memtest boot CD and looking for a bad stick of Ram not just testing an overclock. All the info I could find insisted that 20 passes was needed to conclude if a stick of Ram was bad or not.

Also, from what I read about this, it's very difficult to find a bad stick of Ram using the boot CD when you are testing more than one stick.



NordicJedi -> RE: How many passes for memtest86? (4/14/2009 1:18:47 AM)

Thanks for the info Chevell. There aren't a lot of instructions when it comes to Memtest and what's considered the proper method for testing. The number of passes may be overkill, but if you're at the point where nothing you can do is working, then this may be your only option of testing.





chevell -> RE: How many passes for memtest86? (4/14/2009 7:46:06 AM)

quote:

ORIGINAL: nordicjedi

Thanks for the info Chevell. There aren't a lot of instructions when it comes to Memtest and what's considered the proper method for testing. The number of passes may be overkill, but if you're at the point where nothing you can do is working, then this may be your only option of testing.


What is amounts to is finding a dead stick of Ram can be a royal PITA. I can't tell you how many times I've seen bad Ram pass hours and hours of not only Memtest but Prime Blend also, yet the freeze ups continued.

At that point I'am left trying to explain to the Op that Memtest often fails at finding bad Ram and that they need to replace their Ram despite what Memtest and Prime blend is telling them. So the trust issue comes into play and the only way to get them to replace the Ram is to insist I know more than what Memtest is telling them. It's not an easy sell sometimes....That is why I usually just go by the symptoms like the random freeze up syndrome.

Most DDR2 is so cheap it's hardly worth taking the time to use Memtest because it's faster and easier to just replace them. But if your using more expensive Ram like high end DDR3, that long testing procedure might be worth while.

It would be nice if Memtest kicked out an error in a few minutes on bad Ram, but that isn't reality for Ram testing.



Tweaked -> RE: How many passes for memtest86? (4/14/2009 8:37:52 AM)

Thank you for enlightening us on this Chevell.  I too was under the impression that running Memtest86 boot cd for about 5 passes with all stick in place was enough to identify bad ram.  As for the BR, NordicJedi PMd me nominating your post.



chevell -> RE: How many passes for memtest86? (4/14/2009 9:25:54 AM)

Thanks for the BR Tweaked!  Nordi that was an outstanding BR request on your part, much appreciated. I just read a BR worthy post from you, now I have to find it again. [;)]

I felt the need to spread the word about the realities of using Memtest. I spend over a month reading up on this subject because there seemed to be so much confusion about what to expect from Memtest.



NordicJedi -> RE: How many passes for memtest86? (4/14/2009 10:27:19 AM)

Np chevell. We may have disagreements from time to time, but a good post is always a good post.



daviangel -> RE: How many passes for memtest86? (4/14/2009 9:35:23 PM)

quote:

ORIGINAL: chevell

3-4 passes is good for testing an overclock I agree. But if someone attempts to look for a bad stick of Ram using only 3-4 passes you could still easily have bad Ram getting past a 3-4 pass test.

I see a bunch of guys with freezing problems run a few passes with all sticks in and then conclude the Ram is good. Unfortunately, an incomplete testing procedure like that won't fix anything.

I remember when I ran Memtest boot CD for 25 passes which my old Ram passed but I still had freezing problems so the Ram didn't work right and had to be replaced. Although Memtest for Windows and Prime blend would freeze up which were my best clues.

When I say 20 passes I mean on a single stick using the memtest boot CD and looking for a bad stick of Ram not just testing an overclock. All the info I could find insisted that 20 passes was needed to conclude if a stick of Ram was bad or not.

Also, from what I read about this, it's very difficult to find a bad stick of Ram using the boot CD when you are testing more than one stick.

You're supposed to test them individually unless they were sold as a matched pair from my  understanding and that's the way I usually do.
Q »    I have two Corsair Memory modules that work fine together but when I added two more modules of the matching part number, they won’t work. A »    We test these modules as they’re sold. In other words, a 2GB kit is tested only as a pair, not with two other modules. We’re not certain how two kits when put together will perform. While it is possible to get 2 separate kits working together, it will most likely require adjustments to be made in your BIOS to get the modules up and running. If possible, we suggest lowering your memory frequency and slightly increasing northbridge voltage. Please consult your motherboard manual for information on how to make these adjustments.

If you’re not comfortable making these adjustments, or if the BIOS doesn’t allow for these adjustments to be made, you might want to consider our 4GB memory kits.
Also, Corsair recommends 3 or 4 passes:

How do I use memtest? A »    There’s a readme.txt file included in the package you download from memtest.org which will explain how to setup the program. We recommend using the default test on each module for three or four passes. This will take a while but it gives the modules the best opportunity to fail.
Why not stick with 3 or 4 then? Because it's completely arbitrary and up to how long you want to wait around testing memory.
Anyways, I stick by my original recommendation since that's what the official memtest FAQ also suggest and I don't have extra time to wasting running extra tests when 1 run will tell me most of the time.
"To conclude, one successful pass of memtest will give you a pretty good
idea that your memory is ok, only in rare cases will there be errors
showing after the first pass. To be sure though simply have the test run
overnight or even for a couple of days depending on the level of importance
of the system."
http://forum.canardpc.com/showthread.php?s=ea8da4299411dc4a03b3549846f9379c&t=28864
But like the FAQ says it's really up to you. Anyways, in the past all my Corsair memory I RMA's and Corsair quickly replaced failed memtest after the 1 run not 3 or 4. After testing the said ram in other machines of course to rule out other factors.

If your system is really that unstable you should be spending that extra time troubleshooting other factors.
- If memtest86+ shows no errors does that mean my memory is not defective?

Of course no answers are definitive, no matter how good memtest86+ is or
will become there is always the possibility that a particular type of error
will go unnoticed. As long as you are having no problems with the system it
will be pretty safe to say that the modules are good. If you are having
problems with the system however you will just have to check elimination of
factors, ie swapping the modules for new ones and/or testing with modules of
a different brand/type. See the question 'What is elimination of factors?'
for more information on this method.



daviangel -> RE: How many passes for memtest86? (4/14/2009 9:45:43 PM)

quote:

ORIGINAL: Tweaked

Thank you for enlightening us on this Chevell.  I too was under the impression that running Memtest86 boot cd for about 5 passes with all stick in place was enough to identify bad ram.  As for the BR, NordicJedi PMd me nominating your post.

Actually, there is a small chance it will never get detected but the majority of the time it will be on the first pass;)

From memtest FAQ:
"
As much as possible of the system memory is tested. Unfortunately memtest86+
can usually not test all of the memory. The reason for this is that todays
processors and BIOSes have become so complex that they require a small
amount of memory to keep accounting data of the processor and BIOS state
respectively. If memtest were to write over these areas the state of the
processor or BIOS becomes invalid and it's behaviour unpredictable. Alas it
is also impossible to relocate these areas in the memory.

This means that a small area of your memory can not be tested by memtest.
If this part of the memory is defective you will know soon enough though as
the processor, parts of the processor or the BIOS simply won't work
correctly if this part of your memory is defective. Do realise though that
in very rare cases memtest will show no errors even though the module is
defective, not because memtest can't detect the error, but because memtest
can't test the area the error is located in.
"
Hence, the need for the hardware tester mentioned if you are really really really need to know i.e. mission critical server?



ericc191 -> RE: How many passes for memtest86? (4/17/2009 3:27:14 AM)

Is memtest86 boot cd better then using the built in windows test? I just spent 2 hours trying to diagnose problems with my ram. Thought it might be bios settings. Finally tried one stick at a time to rule out the bios. Sure enough, second stick was bad.

EDIT: IMO, stay away from Corsair memory. Terrible quality control. They seem more interested in power supplies these days.



daviangel -> RE: How many passes for memtest86? (4/17/2009 3:37:01 AM)


quote:

ORIGINAL: ericc191

Is memtest86 boot cd better then using the built in windows test? I just spent 2 hours trying to diagnose problems with my ram. Thought it might be bios settings. Finally tried one stick at a time to rule out the bios. Sure enough, second stick was bad.

EDIT: IMO, stay away from Corsair memory. Terrible quality control. They seem more interested in power supplies these days.


Yeah since you'll get the latest updates and bug fixes including i7 support. The one that comes with Windows is an older version.



Page: [1]



Forum Software © ASPPlayground.NET Advanced Edition 2.4.5 Unicode

0.453