Memory Test Utility available from MS - Free!

[Zepper]

Linkage .
.bh.
:sun:

 

[GTaudiophile]

NM.

 

[bjc112]

I trust Memtest 86 over M$...

At least there are more options now ...

 

[thorin]

Ya this came up on BetaNews about a month or month and a half ago. Everyone that tested it and added comments to the DL basically said "don't bother stick to memtest86"

Thorin

 

[Mark R]

I've just tested this program on my Abit NF7s (which are all buggy/faulty).

Memtest86 will typically find about 0.5 errors per pass - needing between 2-10 hours to find an error. This is consistent with all my machines.

The MS utility found over 5000 errors in the time it took me to put the kettle on (again, consistent with all my machines). It appears to be due to the differing algorithms used:

With memtest86, the only test that ever fails is the 'modulo-21' test which fails only occasionally- no other tests have failed. Even after a 72 hour run. With MS, the test that fails is the 'Stride-6' test which produces thousands of errors within each pass.

I will experiment with my other RAM sticks when I have time, but I suspect I won't get any different results. For the fault that my systems have, it appears that the MS test is much more sensitive.

P.S. Please can someone recommend a replacement for an Abit NF7.

 

[bjc112]

8rda+

However, the Abit board is great... May-b it all just faulty ram...

 

[Zepper]

If you don't require the nforce chipset, look at the Tyan S2495AN/ANRS.
.bh.

 

[Mark R]

I don't think the RAM itself is faulty - I've got 4 different sticks, from 2 different manufacturers. All sticks do the same thing.

Anyway, I contacted crucial and they wouldn't accept results from Memtest86 or the microsoft utility - they recommended DocMemory from www.simmtester.com

DocMem is shocking. It found 0 errors even after 24 hours running the extended tests - it is MUCH slower than either memtest86 or MS, and the display would get garbled during running.

I'm very surprised that different programs can give such different results.

DocMem is slow, inflexible and doesn't find errors - Errors which I know are there because Prime95 fails and I've had some odd results running some other software.

Memtest86 is faster, finds a few errors, but is limited in that it doesn't appear to test all memory. It is flexible.

The there is the MS utility - it's faster than memtest, it finds more errors, does appear to test all memory, and has a more flexible and more frequently updated interface.

 

[OverVolt]

Does this prog cost any bling bling? i didn't see link !

 

[RalfHutter]

Originally posted by: Mark R
Anyway, I contacted crucial and they wouldn't accept results from Memtest86 or the microsoft utility - they recommended DocMemory from www.simmtester.com

Somebody at Crucial is jerking your chain or they just changed their policy. I've RMA'd plenty of sticks of RAM to Crucial on the strength of bad Memtest86 results and they've never given me any hassle about it.

DocMemory is not a particularly good memory tester as you have found out. I've tried Docmem on a suspect box and got no errors and then run Memtest86 and found tens of thousands of errors, replaced the RAM and been good to go. This has happened several times and is all the proof I need to use Memtest86 over Docmem.

 

[corkyg]

Go here and get the free download of the best bootable CD I have found. It has MemTest along with HDD fitness tests and features from all main makers plus a few more. It is a zipped file - and just unzip it and then use Explorer and say "Burn." It makes a bootable ISO CDR that is really useful.

UBCD

 

[Zepper]

Thanks for the link, corky!
.bh.

 

[CZroe]

Is this SOFTWARE in the HARDWARE forum?!

j/k, it's an alternative to those expensive hardware module testers they have here at work

 

[corkyg]

It's a borderline crossover - we're talking testing hardware.

 

[Sunner]

Originally posted by: Mark R
I've just tested this program on my Abit NF7s (which are all buggy/faulty).

Memtest86 will typically find about 0.5 errors per pass - needing between 2-10 hours to find an error. This is consistent with all my machines.

The MS utility found over 5000 errors in the time it took me to put the kettle on (again, consistent with all my machines). It appears to be due to the differing algorithms used:

With memtest86, the only test that ever fails is the 'modulo-21' test which fails only occasionally- no other tests have failed. Even after a 72 hour run. With MS, the test that fails is the 'Stride-6' test which produces thousands of errors within each pass.

I will experiment with my other RAM sticks when I have time, but I suspect I won't get any different results. For the fault that my systems have, it appears that the MS test is much more sensitive.

P.S. Please can someone recommend a replacement for an Abit NF7.

Are you using the "All tests" option?
I had a bad stick of Crucial PC2100, when running the standard test, it found an error once in a while, but running "All tests" found thousands upon thousands of errors.

 

[redbeard1]

Docmem is kind of light duty, if it does find an error, it must be pretty bad. I've had it pass ram that was bad.

Memtest86 works well, but I've had trouble getting it to test intel motherboards with the integrated intel video. If the system has a real video card, it works fine.

I've tried the MS tester, a couple of times, once I read about it in these posts. It seems like it would be light duty also since it can test a lot of ram in a short time. Time will tell.

With all of these, what can you expect for free.

 

[zephyrprime]

With all of these, what can you expect for free.

There really isn't that much to testing ram that's on a PC. You just read a write a bunch of patterns, etc., etc. So I don't think the fact that these programs are free detracts too much from their ability because their ability isn't that difficult to attain.