Memtest86+ found problem. Now what?

Please see attached pics. Can I fix this, or get a second opinion at least? What's next?

System: HP Compaq 6710b, 1.8GHz, 3GB of RAM (2x2GB SDRAM cards), 120GB hard disk, Windows XP/SP3. More details in the attached report.txt file.

Background: I've had some memory related problems recently (virtual memory running low despite increasing pagefile.sys to ~4.5GB, system out of memory error messages/crashes, frequent hangups on shutdowns. Often these were caused while I was streaming radio over the internet, and/or surfing. I suspected it was caused by some software conflict because I recently installed Intel Proset/Wireless and some other wireless stuff, but ran the memtest86+ anyway.

 

Well, Memtest is usually pretty accurate, but if you doubt the validity of the test results you can always run a different RAM diagnostic. I like the Microsoft Memory Diagnostic; it's a free download here at Major Geeks, and it's bootable, and you'd press 'T' for the extended tests. But there is no way to fix faulty memory other than removing that offending RAM stick, and install a new stick, or a used one that is known to be good.....

 

I have two RAM sticks installed; how do I know which one to pull if it comes to that?

 

This one; Microsoft Memory Diagnostic beta: http://www.majorgeeks.com/Memory_Diagnostic_d3955.html

You sure it's not too old?

 

Test them individually. Them RMA the one that fails the test

 

One more question (for now...): I started MS Memory Diagnostic, and after 6 clean passes in standard mode, I switched to extended tests, which take forever... How many clean passes do I need before I can conclude that memtest86+ was wrong? (My laptop is on pass #2 now, but each pass takes a long time, and I need the computer for work... I'm posting off a different (slow) computer now).

 

If you run 2-3 passes on each stick (press T for the extended tests) that should be good. One thing, I have seen RAM sticks that pass diagnostics individually, but fail when run together. I have also seen the opposite, where a single stick will fail, but when run with a 2nd stick it passes... Strange, but true. If the MS Diagnostic says the sticks are OK, but memtest said there was an error, personally, I'd track down another diagnostic and run it on the pair, and each stick individually. I think that Everest and/or FreshDiag and/or Sandra have RAM tests...
Everest http://www.majorgeeks.com/download.php?det=4181
FreshDiag http://www.majorgeeks.com/Fresh_Diagnose_d440.html
Sandra http://www.majorgeeks.com/SiSoftware_Sandra_Lite_d4664.html
You can find some other memory tests in the Major Geeks Memory downloads section and many of them are free. Good luck, and keep us posted!

 

Pass #2 on MS Memory Diagnostic was also clean (both sticks together). I'll let it do a third pass, then try something else.

The problem with these tests is that it takes 2-4 hours for each pass, so if I repeat them for each stick individually, that's another 24 hours (with 3 passes per stick with extra time included for the extra things like reading up on how to do the RAM changes again, etc.). And then another ~24 hours for the tests with third software (assuming my laptop continues passing cleanly...). Since I need the computer during the day, that'll mean roughly a week of testing. Can running the laptop continuously for ~7 days in a row compromise any of these tests? (It's been on for about 48 hours non-stop already...)

[2 things just in case: First, MS Diagnostic did give some kind of disclaimer message before I started the scan - something about too much ram, but I only have 4GB which should be fine (I'll look it up after 3rd pass is done). Second, my OS treats the 4GB of ram as if it were only 3GB (see report.text in OP) - but apparently that's common for these laptops. The good news is if I have to pull one of the 2GB sticks out, and put in one of the old 512MB sticks, I only end up losing half a gig of useable RAM...]

 

UPDATE on MS Memory Diagnostic: I complete 3.5 clean passes with both stick in (1st attachment). I also included a pic of the non-sense warning it gave before the scan (attachment 2). It warned that it couldn't process the system memory map because it ranged above 4GB, and therefore wouldn't test all of the memory. However, the instructions say it tests first 4GB, so regardless of the perceived range, since I only have 4GB of RAM I figured the program would test it all even if it thought it wasn't testing it all.

UPDATE on memtest86+: I pulled out the bottom RAM stick, but decided to test the remaining one with memtest86+ first. The idea was that once I found which stick memtest86+ had a problem with, then I could just concentrate on testing that with MS Memory Diagnostic and the third program if my laptop makes it that far. However, memtest86+ found the same problem as before right off the pat (attachment 3), so I guess I'll put Microsoft's program to work on that now...

Few questions: Is it safe to assume now that the stick I removed is fine, and therefore I don't need to test it further?

 

I would test it just to play it safe

 

Another UPDATE: I completed 3 clean passes with MS Memory Diagnostic (attachment 1) of the top RAM stick only; the one memtest86+ failed. No warning messages from MS Memory Diagnostic this time with only 2GB of total RAM.

However, I then decided I needed use the computer, so I took out the top RAM stick and replaced it with the other. I decided to do a quick memtest86+ per advice above, and the result is in attachment 3; 3 problems without even a full pass.

I'm not sure what to do next. I need to work on the computer now, so for the moment I'm ignoring the errors and using both sticks.

I'm sure MS Memory Diagnostic wouldn't complain about the other one either if I test it independently, while memtest86+ would probably find more and more errors, faster and faster each time I run it...

I'll try something again in a few days, but need to focus on other things now...

 

LAST UPDATE: I ran another ram tester: MemTest 3.7 (freeware from here). The freeware version can be run while doing other things to test part or all of your free memory. The first tests I ran before even posting this thread. The ones that found errors are from today.

I called manufacturer, and the tech support guy said they use memtest86+ v2.01, which is the tester I referenced in the title. He said the cards have most likely gone bad since memtest86+ is reporting errors. He said they would send me new ones.

Conclusion: memtest86+ v2.01 wins. MS Memory Diagnostic (beta) gets thumbs down, at least in this case (in expectation...). However, the latter is dated 2003-07-31 on MG while I bought those RAM cards less than 4 months ago, so maybe that has something to do with it...

RAM maintenance tip (fwiw): I asked the tech guy if there's anyway to prevent or slow down the deterioration of these cards. He said it's possible to sometimes specify in BIOS at what voltage to run them, and some people use 2.2v instead of the 2.0v that's recommended. Apparently, this can contribute to faster deterioration. Personally, I just plugged them into my laptop; I don't know what voltage they've been running at, and didn't see that mentioned anywhere in the instructions.

 

One more update (in case anyone finds this during a search years from now, maybe with a similar problem, and wonders how things worked out):

I'm waiting for the new RAM sticks, and using my old ones in the meantime (which I checked with memtest86+ also - they passed 5 rounds before I stopped). However, my memory problems continued. I increased Virtual Memory to 20MG and installed FreeRAM, memclean and some other memory cleaners. Apparently, my page file had been filling up, RAM maxed out and that caused the computer to freeze and hand. UPHClean may have helped a bit too - but it wasn't until I uninstalled all the HP ProtectTools that came with the computer, that my problems went away. Some of those tools could have been pretty sweet, but they just weren't getting along with my other programs.

Also, AnVir Task Manager Free, DriverView, and some other similar utilities were helpful in sorting things out. AnVir also lets you see how much of the pagefile (Virtual Memory) each process is taking up. Also, my NDIS Usermode I/O Protocol driver (ndisuio.sys) had been disabled, so that may have added to the problems...