Memtest86 has immediate errors but RAM is fine

[VERTIGGO]

I have had my system overclocked at the specs in sig, and Prime was fine, but I decided to give Memtest86 a go and I have errors galore. I swapped out the memory one stick at a time in my old rig with my Corsair Twinx2048s and all 4 sticks passed in the old rig. I'm convinced it's the proc or mobo now. What's the most likely problem?

Possible causes that I know of:

when lapping the CPU I had some bent pins, but in benching and Priming I had no problems even with a 900MHz overclock

I've had vCore as high as 1.4875V I think, but only for maybe 15 minutes.

HTT has been up to 310MHz (and wasn't stable) but I never changed the voltage.

The weird thing is this rig has NEVER crashed randomly yet.
Anyone think its the CPU memory controller, or hopefully the mobo since I'm replacing it anyway?

 

[Madwand1]

I have a similar problem a different board, CPU, and RAM. Whenever I don't overclock the CPU the RAM tests fine over a very long time. When I overclock the CPU, even when using a RAM divider so that it's not overclocked, even underclocking it, even loosening the timings, I get RAM errors -- can take a couple of cycles of MemTest, but I get them before too long.. When I substitute some other RAM, it tests and works fine when overclocked even more.

I'm not sure what the problem is, but clearly there's an overclocking + specific RAM issue in my case.

 

[VERTIGGO]

Even that would be a simple problem, though, but I get errors with CPU at 1700MHz, DDR500 ram at 400MHz, and stock voltages.

Do I need to get a new opty on the way? Or could it be from the mobo.

 

[VERTIGGO]

bump

 

[snor]

I know some ram sticks CANNOT run lower than their rated specs. For instance, my cousin's g.skillz DDR500 will not allow him to boot under DDR 460 (230). You might try to use default cpu speeds and use a 233:200 or 250:200 divider.

 

[VERTIGGO]

Please, is it my CPU, because I'd like to order one soon if it is.

 

[snor]

Originally posted by: VERTIGGO
Please, is it my CPU, because I'd like to order one soon if it is.

Its probably NOT the the cpu, most likely the memory or the mobo or incompatibilities between the 3! I havnt heard of CPU's produce memory read problems, let allow a server grade cpu.

 

[TrevorRC]

Originally posted by: snor

Originally posted by: VERTIGGO
Please, is it my CPU, because I'd like to order one soon if it is.

Its probably NOT the the cpu, most likely the memory or the mobo or incompatibilities between the 3! I havnt heard of CPU's produce memory read problems, let allow a server grade cpu.

Memory controllers can act up, not agree with certain dividers... etc.

Just a note.

Your CPU is probably fine.

--Trevor

 

[F1shF4t]

Ok i dont know if applies to all situations, but from testing i did when i was overclocking on my boards, some dividers when set the system does not read the timings of the ram properly, and so it makes them lower than what the ram can handle. Try seting the mem timing manually.

I'm using a 133 mem divider, and before it was unstable, until i realised that the timing were set incorrect and cause the mem to be unstable, now i set them manually and have no problems at all.

 

[VERTIGGO]

errors at 100 mem divider, 3-4-4-8 timings and cpu at 200 x 6 (1200MHz)

 

[palouse]

I understand that I do not have the same motherboard that you do.

However, I saw similar behavior in my system after many BIOS changes during OC testing. The cure, for me, is to clear CMOS frequently during a long series of tests. Then, when you find the final, "perfect" OC, clear CMOS one last time, and reset BIOS settings back to the ideal OC. I am not going to guess a technical explanation, other than to say it works. I got the idea for the "cure" from other posts on the web, none of which are handy at the moment.

It actually does not take many BIOS changes to cause a memory problem in my system.

I hope this helps. (Ah feel yo' pain!)

 

[xhalation]

search google for memtest and a8n32, you will see that ALL of us can't run memtest. the board isn't properly supported by memtest or there is problems that need to be updated with the BIOS.

I hear a temporary fix is going into the BIOS and turning "USB Legacy Support" to "Disabled". I'm about to try it myself. I have the exact same setup as you almost and my oppy was running 300HTT stable but I was still getting lke 700 errors in memtest. I couldn't figure it out but really it turns out it IS stable and memtest is the problem lol.

btw, this mobo is great even though there is this problem. so far it's OCing better with my 165 and 144 than my DFI did.

 

[LTC8K6]

Have you tried bumping the voltage to the ram?

 

[dunno99]

Or it can just be your motherboard. My Shuttle XPC has a problem with the motherboard overheating and then causing RAM errors (I swapped the two and the memory error is still at the same address). This has nothing to do with the on-die memory controller, since I'm running an Athlon XP, which doesn't have one.

Either that, or it's your power supply that isn't stable enough or can't handle the load? Might want to check those out.

 

[VERTIGGO]

Originally posted by: LTC8K6
Have you tried bumping the voltage to the ram?

no, my ram is at 2.8V currently, which might even be overkill, but I'm not going to 2.9V unless I get serious about vapor chilling.

 

[VERTIGGO]

and yeah, i'm still running only one GPU on my Turbo Cool SLI psu; it's not that.