• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Please help me understand my memory test results....

K

KutterMax

Member
I've been having some problems on my machine with some apps crashing, especially World of Warcraft. Ultimately Blizzard Tech support recommended running Microsofts Memory Test app: Link

I have 2 x 512MB sticks of Kingston Hyper-X. When I ran the test initially with the two sticks in place, the results showed a lot of errors but it seemed they were all from one stick. Here is what the results looked like after 2 passes:

Windows Memory Diagnostic found 596412 errors. The error distribution by test is:

MATS+ (cache enabled) 22065
INVC 23534
lLRAND 49104
Stride6 (cache enabled) 7600
WMATS+ 196416
WINVC 297693
MATS+ (cache disabled) 0
STride38 0
WStride-6 0
ERAND 0
Stride6 (cache disabled) 0


One note - the results with the 0 are the extended tests which I did not run.

Here are the results by memory module - again this is with both modules installed:


Bank 'BANK0': Slot 'DIMM0': 596412
Bank 'BANK1': Slot 'DIMM1': Not populated
Bank 'BANK2': Slot 'DIMM2': 0
Bank 'BANK3': Slot 'DIMM3': Not populated



I interpreted the results as the stick 'DIMM0' was bad, so I pulled it and reran the tests.

Here are the results with DIMM0 pulled out and DIMM2 left in place; I only ran one pass:


Windows Memory Diagnostic found 301940 errors. The error distribution by test is:

MATS+ (cache enabled) 15200
INVC 12868
lLRAND 24552
Stride6 (cache enabled) 3800
WMATS+ 98208
WINVC 147312
MATS+ (cache disabled) 0
STride38 0
WStride-6 0
ERAND 0
Stride6 (cache disabled) 0



Bank 'BANK0': Slot 'DIMM0': Not poplulated
Bank 'BANK1': Slot 'DIMM1': Not populated
Bank 'BANK2': Slot 'DIMM2': 301940
Bank 'BANK3': Slot 'DIMM3': Not populated


So pulling out the DIMM I though was bad resulted in errors in the "good DIMM".

So I then decided to swap out DIMM2 and put back DIMMO and rerun the test.

Here are the results with DIMM0 in place and DIMM2 taken out. Again I only ran one pass:


Windows Memory Diagnostic found 305006 errors. The error distribution by test is:

MATS+ (cache enabled) 15200
INVC 15934
lLRAND 24552
Stride6 (cache enabled) 3800
WMATS+ 98208
WINVC 147312
MATS+ (cache disabled) 0
STride38 0
WStride-6 0
ERAND 0
Stride6 (cache disabled) 0



Bank 'BANK0': Slot 'DIMM0': 305006
Bank 'BANK1': Slot 'DIMM1': Not populated
Bank 'BANK2': Slot 'DIMM2': Not populated
Bank 'BANK3': Slot 'DIMM3': Not populated




So I'm not really sure what is going on. Could both sticks be bad? Could something else be wrong - like a problem with the motherboard? I was going to order some new RAM but until I understand the results better I am going to hold off.

Thanks in advance.







 
Have you tried bumping up your DRAM voltage by .1?

I was having trouble getting my Hyper-X to pass prime95 at 2-2-2-5. 3DMark would fail also.

Turned out my board was only supplying ~2.5V, but the memory label said 2.6V. Bumped my voltage by .1 in the BIOS and all my problems disappeared. I am now long term P95 stable at 2-2-2-5 settings, and 3DMark completes without errors.
 
I haven't played around with the DRAM voltage before, so here is what I did.

I went into my BIOS and set the "AI Overclock Timer" to manual. Previously it had been on "Standard".

Doing this opened up a bunch of options. There is now a setting "DDR Reference Voltage" - I assume this is the one to adjust. It gives the following options: 2.85v, 2.75v, 2.65v, 2.55v, Auto.

I set it to 2.65v and reran the memory test with both sticks. I still got a ton of errors with the vast majority on DIMM0 but actually a handful on DIMM2 as well.

I'm then reran at 2.75v to see if there is any difference but there wasn't - still tons of errors.


Does anyone think that the problem could be anything other than bad RAM?
 
The program that has the most problems is World of Warcraft - it is very unstable on my desktop with both #131 and #132 errors.

That being said, I've had a few other unexplained crashes - but it always hard to pinpoint whether this is software or hardware. I've had Pinnacle Studio 9 crash during compiling of a DVD (but Pinnacle software has a rep of being buggy). Most other apps are stable but I recently had a lot of problems after installing the desktop search tool X1. It would crash while indexing Outlook. Again - hard to know if this is software or hardware related.

I went and downloaded Memtest86 and created a boot CD. It's running on my desktop right now and I can see it has already generated a bunch of errors, so it is finding them too.

My guess is that I'm going to have to buy a couple of new sticks of RAM and try them out. If I still get errors, then I suspect the problem is either my CPU or my motherboard. Agreed?

 
Just to update - here is the error summary I'm getting in Memtest86.

Test: 0 Errors: 0
Test: 1 Errors: 0
Test: 2 Errors: 1
Test: 3 Errors: 8
Test: 4 Errors : 34
Test: 5 Errors: 363
Test: 6 Errors: 0
Test: 7 Errors: 1
Test: 8 Errors: 0
Test: 9 Errors: 0


Again, not sure what the individual test results mean but this is what is happening after running for about 30 minutes or so.

 
The only other thing left to check is your memory timings. Make sure they are set the same as the memory. You can also try loosening them up to see if that helps. Try setting the timings to SPD if they aren't.

Have you tried turning off PAT, if it's on?

I would be suspecting the memory at this point.
 
I need to update my sig - the first thing I did after I started having the problems with World of Warcraft crashing was to go back to stock clock speeds, so all the problems I've reported have been at stock 3.2GHz.

My memory timing are at 2.5-3-3-8 with SPD enabled.

I turned SPD off and manually set to 2.5-4-4-8 and got dramatically fewer errors running MemTest86. I still had a couple but now nearly as many. I then booted from the Microsoft Memory Test CD I made and ran the Standard Tests at the same 2.5-4-4-8 timings and I got the same 300000+ errors that I always got.

I agree that this is really looking like a RAM problem. I think I am going to bite the bullet and order a couple sticks of Mushkin RAM and give it a try and see if it solves the problems. Not sure what else to do.






 
Yeah, try 2T command rate if it's on Auto or on 1T in the bios. To say the least, if that actually did improve stability I would invest in better RAM if you have the budget.2T command rates on a Intel Northwood is detrimental on performance. At least you know you have good odds it isn't the motherboard.
 
Check the SPD manually to make sure you didn't get PC2700 or something like that. Hopefully your motherboard is supported by this program.

Download and install Everest. Run it, and go to its Motherboard -> SPD section. Hopefully, your memory modules will be visible after a short (less than 5 second) delay. In the bottom section, are two columns - Field and Value. Look for the section called Memory Timings. My system gives this:
@200MHz - 2.5-4-4-8 (CL-RCD-RP-RAS)

Sometimes it'll give other speeds too, like:
@166MHz - 2.0-3-3-7, indicating that you can tighten the timings at lower MHz ratings. Other RAM likely can do this of course; this particular stick just has it documented.

So post back what Everest says about your RAM.
 
OK - I installed Everest and ran it. Here is what it said:

Memory Timings:

@217 MHz 2.5-4-4-9
@200 MHz 2.0-3-3-8


Other info it says:

Memory Speed - PC3500 (217 MHz)

 
You need to get rid of that overclock first. You have to rule that out as the cause of the errors. Put everything back to stock and see if the errors remain.
 
Originally posted by: KutterMax
I turned SPD off and manually set to 2.5-4-4-8 and got dramatically fewer errors running MemTest86. I still had a couple but now nearly as many. I then booted from the Microsoft Memory Test CD I made and ran the Standard Tests at the same 2.5-4-4-8 timings and I got the same 300000+ errors that I always got.
Click to expand...

Originally posted by: KutterMax
OK - I installed Everest and ran it. Here is what it said:

Memory Timings:

@217 MHz 2.5-4-4-9
@200 MHz 2.0-3-3-8

Other info it says:
Memory Speed - PC3500 (217 MHz)
Click to expand...

Have you tried adjusting the timings to match what Everest reported? Also have you updated your BIOS? I have the ASUS P4P800 Deluxe & I had a few problems with the Mushkin memory I installed untill I updated the BIOS.
 
An update:

Since loosening my timings to 2.5-4-4-8 at stock clock speeds (3.2GHz, 200FSB), things seems to be much more stable. World of Warcraft has not crashed once and usually it crashed multiple times an hour.

2.5-4-4-8 is significantly slower than what the RAM is supposed to do at 200MHz (2.0-3-3-8), so something is definitely wrong with it. Not to mention I still get some errors with MemTest86 at this speed (although far fewer than when I tested at 2.0-3-3-8).

I decided to get a couple of new sticks of Mushkin 2-2-2 RAM. I'm interested to see if this will give me any errors - hopefully it won't and then I will know for sure the Hyper-X RAM was the problem. I will then send the Hyper-X back to Kingston for replacement.

I really appreciate everyones help in helping me solve my problem. This, along with the Blizzard tech support person who told me to check my RAM in the first place, were invaluable in getting to the bottom of things.

I'll post as soon as I get the Mushkin RAM installed and tested.

Thanx again.
 
Update:

I got the 2 x 512MB sticks of Mushkin Dual Channel 2-2-2 PC3200 RAM.

I pulled out the Hyper-X and inserted the Mushkin. I then rebooted and went back into the BIOS and re-enabled SPD.

I then booted using my MemTest86 boot CD. I did a quick run through of the test. The RAM was set at 2.5-2-2-6 according to MemTest86. I had no errors after doing one pass of each test - good sign. I always had a least a couple of errors with the Hyper-X, even with the looser timings.

I booted into the OS - everything seemed find. Everest listed the RAM at 2-2-2-6.

I then rebooted with the Microsoft Memory Diagnostic CDROM. I got a ton of errors again here - just like with the Hyper-X.


So bottomline is that MemTest86 says the RAM is fine but Microsoft Memory Diagnostic doesn't. The MemTest86 results are more in keeping with what was happening on my computer - when I had settings that generated errors in MemTest86 - I got crashes, especially in World of Warcraft.

It still leaves the unanswered question of why Microsoft Memory Diagnostic is generating errors? Perhaps it has an incompatibility with my system?
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top