Windows Memory Diag Program Fails...

[ItsRenew]

I just put together a new computer after my old Alienware bit the dust. Since the first night I have been having issues with it while I am playing games (hasn't had any issues while just surfing the net, messing around in Photoshop etc). Sometimes it will be stable all night, but them sometimes it just starts hard locking, blue screening etc.

So alot of the Blue Screens have brought up issues concerning possible bad memory. I have Vista (was also BSOD'ing in XP) and ran the memory diagnostic tool. When it gets to around 15% or so, the computer just reboots without finishing the memory test.

Does this mean my memory is bad, or something else? Any replies would be appreciated.

 

[Lemon law]

If you are going to be testing memory, something like memtest86 is a better basic tool. It comes on before the OS so its OS independent and can be put on a floppy or a cd.
Just google memtest86.----but one hooker here---if memtest86 shows memory errors, it can't isolate them to a single memory module in a multi-module set up. So you may need to test one stick at a time if you get errors.

The other thing to suspect is bios settings if you get memory errors---very often the ram is good and it just won't run error free at the bios setting you have---but a slight bios tweak regarding timings or voltages will often get you memory running error free.

But I am putting horses before the cart----step one is to establish if your memory is running error free or not. And if your memory is running error free---you are barking up the wrong tree.

But given your symptoms---I would add some temperature monitoring software---and see if something is running real hot---even something like everest 2.2 would do for a freebie.
Too hot always equals errors until some thing totally breaks.

 

[Smilin]

Originally posted by: Lemon law
If you are going to be testing memory, something like memtest86 is a better basic tool. It comes on before the OS so its OS independent and can be put on a floppy or a cd.
Just google memtest86.----but one hooker here---if memtest86 shows memory errors, it can't isolate them to a single memory module in a multi-module set up. So you may need to test one stick at a time if you get errors.

The other thing to suspect is bios settings if you get memory errors---very often the ram is good and it just won't run error free at the bios setting you have---but a slight bios tweak regarding timings or voltages will often get you memory running error free.

But I am putting horses before the cart----step one is to establish if your memory is running error free or not. And if your memory is running error free---you are barking up the wrong tree.

But given your symptoms---I would add some temperature monitoring software---and see if something is running real hot---even something like everest 2.2 would do for a freebie.
Too hot always equals errors until some thing totally breaks.

I can tell you've never even used Windows Memory Diagnostics based on your 2nd sentence. Windows Memory Diagnostics are run from CD or Floppy and work in real mode as required during memory testing. It's also *gasp* better than memtest..hard to believe I know. Memtest rocks.

Windows diags come with all the same test patterns as memtest plus an extra test pattern that I've seen spot things on first pass when the other patterns take multiple or don't find anything at all. It can make the difference between a 10 minute run and a 10hr run. Plus I think they started charging for memtest86 didn't they?

Also you can take an educated guess as to which stick is failing with both Windows memory diagnostics and memtest by looking at the address where the failure occurs. You're in realmode during the test so the lower half of addresses will generally correspond to the stick closest to the CPU.

I would suggest taking out one stick of memory at a time and rerunning the test. In this case try taking out the first stick first. The code being executed is designed to run with damaged memory so unless the bad bits are happening where the code actually resides this may very well be a memory controller or CPU issue.

Lemon's suggestion about heat is a good one.


For you, ItsRenew: http://memtest86.com/

For you, LemonLaw: http://oca.microsoft.com/en/windiag.asp

As a memtest user I think if you try it you'll be sold.

 

[Lemon law]

Gotta confess you are right Smilin---I will try the microsoft tool. But Memory errors can also because by wrong bios settings---making either tool ideal for testing that. But still its more likely heat.

 

[ItsRenew]

I got an email from Newegg asking me to rate the items I purchased. So I took the link and from there clicked on the memory link to see if any new comments had been made about the product...

Well, I guess Lemon hit it right on the head. I had to up the voltages and also the clock speed. After I did that, the memory test ran and showed I had no errors. I am assuming the hard locks and blue screens were caused by the memory not running at it's wanted clock speed / voltage.

Hopefully it stays stable and I don't need to head back here to bug you all Thanks again.

 

[Smilin]

Cool.

Good call, Lemon.

 

[Lemon law]

Well even a blind sow finds an acorn once in awhile---I'll still take cool, under advisement. But you might try a stress test called prime95 to be doubly sure you are stable. And its also good to use to see what your cpu temp goes up to under load.