• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Help understanding MT86+ results

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Tugboat

Registered
Joined
Feb 14, 2005
Location
California
Okay, I'm at a loss here. I cannot understand how I'm able to do what I do with the results I get from running Memtest86+ v1.55.

I originally started with 512MB of RDRAM from Kingston and was sucessfully running at 1200 with a FSB of 150. When I purchased a second pair of Kingston I couldn't run it at anything past a FSB of 133. I sent them back to Kingston and got a new pair, but it was the same. The only thing I could tell that was different was that my original mem had black heat spreaders and was from Japan. The others were with silver HSs and from the US. I could not find any Kingston mem anywhere that had the black HSs and made in japan, so I gave up. These allowed the running of 1200 w/no errors in MT86. I got tired of only running 512, so I decided to try some Samsung memory. I got a pair of 256s on ebay and installed them.

I ran the new 1024MB without any testing as I figured XP would crap out like before if it didn't work. It worked like a charm and for the past 4 months I've been running with no restarts or mem errors. I've been doing video editing that has used up to 800+MB and all was well. I've been quite happy. That is, until, I ran Memtest86 earlier this evening. The results weren't good, but also leaves me wondering why.....

To start, I'm assuming that my first two modules (Kingston) should still be the first 512MB and the new modules should be the second 512MB. This being true, I would have expected any errors to be above 512MB in testing as the Kingstons had never failed in testing before. Also, I would expect that any failure in any point in the 1024MB would mean problems for me running XP and my apps.

So here's a short list of my errors running Memtest86+ v1.55.... Test 5 have me 6 errors in the 194 to 218.5MB range. Test 6 have me 4 errors in the 201.4MB range. On the second pass (that's where I stopped the test) on test 3 I got 4 errors in the 205MB range (for which I didn't on the first pass of this test).

Now if I'm getting memory errors like this, how can I be running XP and all my memory intensive apps (including dual instances of Prime95) with no problems at all? Wouldn't / shouldn't I be having all sorts of problems?

I'm at a loss.

Thanks
 
So no one can help me with this? I've read many posts here from those that seemed to have a grasp on Mem86. A number of them stressed the importance of test 5 & 8. I'm not meaning to be a pain, but just trying to understand the results of my testing and why it doesn't seem to be affecting my use of XP and apps.

Thanks!
 
Well, there is a point at which these memory erros start becoming invisible.

For a real test, run a 32M superpi test. This will stress the memory more than memtest. If it passes you are fine. If not, then you aren't as stable as you thought you were.

Its just that sometimes, the errors aren't always effecting things in easily identifiable way.. for instance, you are having trouble with a 10-20 meg block of memory, and thats it. If a non essential, rarely used program is loaded into that section of memory, things will look fine.

Yet, one day you will install something, and the program load order will change, and a vital system function is loaded into this memory block. Then you could be crashing all over the place.. or maybe just crashing infrequently.

The point is, memory doesn't work in an all or none fashion. It can "sorta" work, or even "mostly" work.. It can even "apparently" work. Memtest rarely lies though...
 
I'll give the Superpi a try. I've been running for months now with no reboots, lockups, data corruption, etc. I have been able to run two instances of Prime95 for 12 hours w/no problem as well. Not to mention Adobe Premiere for video editing and been up to over 900MB used. Hmmm... Maybe just been lucky.

One thing I find odd is that the errors in MT are below 512MB. I'm assuming that since I have the Kingston modules in RIMMB1 and alone they pass MT with no problems, adding the Samsung to RIMMB2 would make those modules above 512MB and leave the Kingston as 512MB and below. So if I'm not completely wrong on how the memory is used, why am I now getting errors on the Kingston modules (assuming they are still 512MB and lower)?
 
There is a sticky called Memtest86+ Guide/How To's. It gives you very good information about where your memory errors if it happens to error.

What comes to memory errors, even a one single error is considered as "too many". You may not see this in short time usage, but you will probably face problems after long time usage.
 
Back