whats a good memory tester?

[MantaBase]

You may both have a point.

XP (and I suspect Vista) is very sensitive to RAM. I got that info from a guy who goes by "Logic" and know MS systems rather well. I didn't believe him (I never believe him - it's a thing between us).

So, one day I put some ValueRam in an XP system (no OC) and suddenly have a harddrive failure (so Windows says). Long story short, it was the RAM - which passed memtest86 just fine. Memtest86+ might have given me a different result, but it was a lesson I carry.

[One4yu2c]

Originally Posted by polobunny

If trying to pinpoint faulty memory

Between this and his other thread, that appears to be the case - looks to be a newly built (and troubleshooted) system with frequent BSODs.

I'll also add (or reiterate, really, as Manta makes this point above) that while Memtest86+ does a very good job, it is possible to have faulty sticks pass without errors. Most often, however, if your RAM is bad, expect to see lots of errors running the diagnostic.

[RangerXLT8]

This sounds like a problem I just had. Sticks would pass memtest and then fail to boot windows, everything stock.. Re-installing vista never helped. I concluded it was the MB chipset and sent that sucker back.

[Capper]

Orthos is a great program for testing overall system stability. Memtest is great for telling if there are errors with the memory.

If Memtest passes, then the memory is fine, booting into Windows is something completely different, and not representative of bad memory. Most often its a voltage issue, either the memory voltage or NB Voltage....or it can be a latency issue.

Memory testing apps work by throwing series after series of number sequences at the sticks in different patterns......if there are errors on the modules, they will fail......when you have problems like this, its almost always the motherboards chipset, or a memory latency out of whack. When board makers talk about validated memory, its not actually that the memory won't work with that particular board, because most ICs and PCBs come from only a couple different places.....its the SPD settings programmed into the module, which controls how the module interacts with the motherboard.

[MantaBase]

Originally Posted by One4yu2c

I'll also add (or reiterate, really, as Manta makes this point above) that while Memtest86+ does a very good job, it is possible to have faulty sticks pass without errors. ...

I will add, that this is rare - if it passers, it should be OK. I only mention it to stop anyone from eliminating RAM issues as a possibility and going thru and exchanging tons of parts during a track down. Capper is right, if the RAM passes the test, it's not likely the problem. You should move on to other issues.

However, don't dismiss it (RAM) totally.

If you have eliminated all else - consider a RAM issue - especially if you are using consumer grade (not enthusiast) grade stuff.


NT systems are sensitive and inflexible with system memory addressing. That want the space they want in other words. Memory that passes a test should be fine 99%+ of the time - but if there is the slightest problem with a particular area in RAM, NT systems can act funky. They don't dynamically change the addresses they want to use like *nix systems will.

[polobunny]

Originally Posted by MantaBase

I will add, that this is rare - if it passers, it should be OK. I only mention it to stop anyone from eliminating RAM issues as a possibility and going thru and exchanging tons of parts during a track down. Capper is right, if the RAM passes the test, it's not likely the problem. You should move on to other issues.

However, don't dismiss it (RAM) totally.

If you have eliminated all else - consider a RAM issue - especially if you are using consumer grade (not enthusiast) grade stuff.


NT systems are sensitive and inflexible with system memory addressing. That want the space they want in other words. Memory that passes a test should be fine 99%+ of the time - but if there is the slightest problem with a particular area in RAM, NT systems can act funky. They don't dynamically change the addresses they want to use like *nix systems will.

Yep that's true. Sometimes it will take more than one pass of memtest to find errors, but in the end it usually finds them. :)

[Capper]

I'd consider RAM fine if it passes 10-15 loops of Memtest....If you are squeezed for time, reduce it to tests #5 and #7, which are the two that really press the RAM.

[theburan]

I had a kit of ram pass memtest, even 3 full 24 hour test... only to be unstable in windows.

It was kinda lame, although I'm still not sure if it just was the ram not agreeing with my mobo.

That was a one time thing though... memtest has proved reliable for me all the other times I've used it.

[Jesta]

well my ram passed a single pass. guess i will let it run overnight.

Thanks for the help though :)