Weird memtest problems

rednar

Distinguished
Feb 2, 2009
2
0
18,510
Hello guys,

As the topic says, I have weird problems with memtest. Long story short - I can't make the new wersions (2.11 for 86+ and 3.5 for 86) to work. My rig is: E8400 3.0 (didn't OC yet), 2x2 GeiL Ultra 1066 5-5-5-15 & EP-45 UD3R mobo.

It goes like that - firstly, I have created bootable CD with memtest86+. Booted with it and it hangs on the "Loading........." screen. Nothing helps, only hard reset. I tried to play with both CPU and memory voltage, mhz's and timings - with no success. Okay, I thought, lets try memtest86 - they finally updated it to version 3.5. Booted, it started fine (yippie!) and crashed like 5 seconds later on Test1. Same story as above - voltage, timings etc. - no help.

And than I saw there's the option with memtest 3.5 to run it in the old 3.4 version. So I did - it didn't identyify my CPU properly (says it's Pentium III with unknown L2 cache size), but it worked pretty darn well. After 1,5h of testing it didn't show single error. Tested the RAMs once again - and again no error.

Now, I can't really say if the rig is stable - I'm waiting for my HDD to come in few days, so I can install Vista Home Premium x64. I did try to start the instalation tho and it went perfect to the moment I had to choose HDD :D But should I be worried about stability? CPU seems to be the issue - in memtest86 3.5 it has been identified properly (Core2D 3mhz). But maybe - maybe ;) - there's a known issue with newest version of memtest and there's a way to make in run smoothly on my rig?

Thanks a lot in advance for help!
 

Mondoman

Splendid
Try booting the Ubuntu 8.10 CD -- it's a "Live" CD, so you can run Ubuntu even without a hard disk! At bootup, it also has options to run memtest86+ and to verify that the CD is OK (which exercises your CD drive).
 

rednar

Distinguished
Feb 2, 2009
2
0
18,510


Thanks for the tip! If they won't send me that HDD tommorow, I'm going to try it ;)
Meanwhile, update of the problem: I've initiated memtest86 3.5 with the "Multiple CPU" option. It seems that it test the memory with only 1 core active (quoting: "Act_CPUs: 2", "Using CPU: 0" and in the next cycle: "Using CPU: 1". So far so good, 2 cycles complete and no errors. Starting the "normal" version still results with crashes.
 

skrekk

Distinguished
Apr 3, 2009
1
0
18,510
I had exactly the same problem with an E8500 running on a DFI P965-S, and 4GB OCZ DDR2-8500. Test #1 would force a reset, no matter what bios settings I used. But XP would run fine with default settings, and with my OC settings of a 450MHz FSB @9x, as would Orthos, and OCCT...so I was sure the problem was in MemTest86 (I tried v3.5 and v3.3). I noted that MemTest86 was reporting the processor speed incorrectly (apparently assuming the max 9.5 multiplier, rather than the 9x being used), so I tried the open source MemTest86+ v2.11 instead. Voila! - no problems. Supposedly MemTest86 v3.5 is based on MemTest86+ v2.11, but the developer must have introduced a bug.