My pc failed MemTest-86 v3.2 -twice- what now?

my fresh build is having serious problems so i decided to run memtest86 -the problems are so severe that i was even surprised i was able to download it... after 11 minutes of letting it run, it's failed test 6 twice with these errors:

Tst Pass Failing address Good Bad err-bits count chan
----- ------ ------------------------------- ------------ ------------- ------------ --------- -------
6 0 00017527454 - 373.4MB 00800000 00800800 00000800 1
6 0 00017aa3454 - 378.1MB 20000000 20000800 00000800 1

i waited for test 6 to cycle again and this time there were no additional errors after a total run time of 32 minutes... while the cpu (xpm2500) and mobo (nf7s v2) are second hand, my single stick of 512mb pc3200 kingston value ram (ch6) is brand new... the question is, what do i do now?

please help


PS: i was slightly sloppy applying as5 to my xpm2500... could that be causing any problems?

PSS: odd thing... i thought everything was set at default but while my cpu is running at 166fsb, memtest reports my ram is running at 199fsb... wth?

PSSS: my pc is acting disturbingly slow and weak -like i wrote this post in wordpad and couldn't select the whole thing at once weak (!!!)... it's acting like there's no free resources... i don't know what's up

I wouldn't worry about memtest that much since not many people consider it as a stability tester.

The best way to test your stability is with SuperPI and running 32MB test or with Prime95’s Blend test.

If you still get errors you might think about relaxing your timings or feeding it more Vdimm cause those are the most usual suspects when it comes to this. The fastest way to resolve it would be to raise RAS to CAS by 1 point and see how it goes.

RedDragonXXX said:

I wouldn't worry about memtest that much since not many people consider it as a stability tester.

The best way to test your stability is with SuperPI and running 32MB test or with Prime95’s Blend test.

If you still get errors you might think about relaxing your timings or feeding it more Vdimm cause those are the most usual suspects when it comes to this. The fastest way to resolve it would be to raise RAS to CAS by 1 point and see how it goes.

Click to expand...

hmm... if i just want to make sure my pc is moderately stable, how long do i have to run each of these to obtain a credible result (ie: unfortunately, i really need to use this pc)?

EDIT: this sounds like a stupid question but do i run both at once?

EDIT: i forgot to post a link describing my predicament but oddly enough, the timings for my ram are way looser than they should be and are running 1mhz under default speed -thus i was truly taken aback when two errors suddenly showed up in memtest :/

zer0nix said:

hmm... if i just want to make sure my pc is moderately stable, how long do i have to run each of these to obtain a credible result (ie: unfortunately, i really need to use this pc)?

EDIT: do i run both at once? prime95 doesn't even seem to be doing anything :/

Click to expand...

Just do SuperPI 32MB, if it passes then you can consider you comp stable.

RedDragonXXX said:

Just do SuperPI 32MB, if it passes then you can consider you comp stable.

Click to expand...

ouch... i press start and immediately an error window pops up... not sure what to do now :/

You shouldn't be running the memory and FSB asynchronously on an AXP system. Go into setup and change FSB and memory settings such that they are both running at 200 MHz. Also, put your CPU multi lower than default so that overall CPU speed is close to or lower than stock. Memtest will show errors if the CPU is overclocked too much for the voltage you are providing.

Also, have you installed your video card drivers yet? If not, this is the reason that simple tasks in Windows, like selecting text in Notepad, seem to be taking a long time. Once you install your video card driver it will return to normal.

ouch... i press start and immediately an error window pops up... not sure what to do now :/

Click to expand...

Your system is way too unstable. Generally, if you get errors in memtest, there is no way that Windows will be stable. Shoot for stability in memtest first.

KillrBuckeye said:

Shoot for stability in memtest first.

Click to expand...

I get 100 passes on test#5&8 each no errors with my RAM at 290MHz and when I get into Windows it won't even pass 1MB SuperPI on occasion and not even one pass on Prime95 and I seen this with other people to.

SuperPI and Prime95 are the best stability testers, no qustion about it!

RedDragonXXX said:

I get 100 passes on test#5&8 each no errors with my RAM at 290MHz and when I get into Windows it won't even pass 1MB SuperPI on occasion and not even one pass on Prime95 and I seen this with other people to.

SuperPI and Prime95 are the best stability testers, no qustion about it!

Click to expand...

Whoa, man. I'm not arguing with you. All I'm saying is that if he is getting *ANY* errors in memtest, then he doesn't have a prayer of being Windows stable. It's the same thing you just stated. Therefore, there is nothing wrong with trying to become stable in memtest before Windows... baby steps.

KillrBuckeye said:

Whoa, man. I'm not arguing with you. All I'm saying is that if he is getting *ANY* errors in memtest, then he doesn't have a prayer of being Windows stable. It's the same thing you just stated. Therefore, there is nothing wrong with trying to become stable in memtest before Windows... baby steps.

Click to expand...

Not arguing

Just saying that I don't rely on memtest as a big stability tester, but you are right it should give you some indication if you are doing things right

P.S. If you do memtest then do tests# 5&8 those are the most commonly used ones.

If you have errors in memtest never go into windows. Errors that are memory related will corrupt your files so why risk it?

Have you tried adjusting your memory voltage and or flashing to a bios that will let you run with CPC disabled? Not all NF2 boards can run 200FSB with 512MB sticks of ram with CPC enabled...oh joy I know

SteveOCZ said:

If you have errors in memtest never go into windows. Errors that are memory related will corrupt your files so why risk it?

Click to expand...

Worst thing that happened to me with unstable RAM was random restart.

RedDragonXXX said:

Worst thing that happened to me with unstable RAM was random restart.

Click to expand...

You're lucky. I've had everything from drivers failing to load to corrupted Windows installations--all of this within the past month since putting together my A64 system and trying to OC. As I found out, all it takes is one hard crash due to memory instability to demolish a Windows installation. Thank God for Acronis True Image .

KillrBuckeye said:

You're lucky. I've had everything from drivers failing to load to corrupted Windows installations--all of this within the past month since putting together my A64 system and trying to OC. As I found out, all it takes is one hard crash due to memory instability to demolish a Windows installation. Thank God for Acronis True Image .

Click to expand...

Wow, didn't know it can get that bad but thanx for sharing

I do lots of RAM tweaking and never had any serious problems like that beside burning one of my sticks cause I always take things step further

RedDragonXXX said:

I wouldn't worry about memtest that much since not many people consider it as a stability tester.

Click to expand...

I'm not sure which one of the programs you're directing that "memtest"? There's still something I'd like to share with you guys so keep reading.

I would consider Memtest86 as a stability tester, still it's dependent about the platform you're using. Athlon64 systems are the ones that may not find any errors in Memtest86 but Windows based HCI Memtest will do the job and will find them.

So, for all the other platforms than A64 based, it should be considered as a stability tester, and only for A64 if it happens to find any errors. If you have A64 system and you use Memtest86 to test out your memory, I would suggest to verify it with HCI Memtest since it does the job better, alot better.

I've seen few cases where Memtest86 shows no errors after ~4 hours and HCI memtest finds them in less than 30 minutes.

Zebbo said:

I've seen few cases where Memtest86 shows no errors after ~4 hours and HCI memtest finds them in less than 30 minutes.

Click to expand...

My point exactly, that's why I don't use it that much beside testing out new timings and RAM burn-in.

RedDragonXXX said:

My point exactly, that's why I don't use it that much beside testing out new timings and RAM burn-in.

Click to expand...

I think you are forgetting the main use of memtest which is to test for faulty ram

Kendan said:

I think you are forgetting the main use of memtest which is to test for faulty ram

Click to expand...

True it is used from RAM testing and my argument was that it can't be used as a complete system stability tester.

You will be way better of with Prime95 and SuperPI if you are looking for complete system stability.

I just stumbled upon This program from the Microsoft web site, thanks to a thread I read somewhere the past day or two. Like Memtest86, it loads at boot from either a floppy or CD. It caught memory errors tha snuck past Memtest86 and once I found settings that passed it, they have handled everything I could throw at them from the Windows environment, such as Super Pi 32M, Prime95 Blend, the entire 3d Mark host of tests, Everest, etc. With my 2x1gb it takes about 5 or 6 minutes to run. I like it, but then I like it because these Ballistix I got today are doing so well for me also. I can't comment upon how well it tests CPU and Video overclock stability though.

Hoot