What to do?

[Robert]

Well I have no idea. 1 or 2 passes of memtest = fine. I've done windows memtest at like 450fsb to 300% coverage...fine. Now when I run Orthos with blend I get errors after an hour...three times in a row. Now when I ran Orthos with Large FFTs I get no errors after 6 hours (I've ran it a few times over the last week) each time it's gone over 6-7 hours. I have tried it with 44412 and with auto, I have used 2.1/2.2/2.3 volts. I have no idea what on earth the problem is. I have had "ram performance" on normal and turbo.

At the moment I'm running 400*7 for 2.8ghz on my E6300. At the moment I'm really tempted to RMA it and get some Corsair 8500 stuff as I'm REALLY annoyed.

Any help would be appreciated. Oh and surely if I sent it back then they probably wouldn't even find a fault as memtest comes up clean, or do I need to leave it for lots of passes (surely 1 or 2 is enough?).

Also check Geil's sham of a support forum:

http://www.geilusa.com/forum2/

[Max Tractor]

Have you tried setting all the memory timings manually?

[Robert]

Memory timings have been at 44412 and auto. Memtest loop 5 & 8 ran 4 passes and 2 passes respectively. Could I have just been lucky on my last two orthos runs? It's weird. I can't see how last night it ran fine for around 8 hours, then today it errors three times in a row?

[Max Tractor]

Try setting all memory timings manually

edit

Another shot, is try 7x401

[Robert]

Yeah to what? I've had it at its recommended settings and at auto, I'm not exactly sure what the best settings would be to try.

[Max Tractor]

Unfortunately, I don't know what your memory can do. If you have a value next to auto, then select it, then trouble shoot accordingly, loosen them off etc


edit

I would also be running memtest for min overnight and memtest in windows for min 1000% coverage

[Clunk]

Try subtimings of 3, 10, 5, 13

also try 1.9v and 2.0v.

Its worth a go

[Robert]

I'm on 2v now. Anyway - my system seems stable. It's really odd.

[notsobig]

Sorry if I'm saying something wierd but why are you only bothering about memory? Have you try uping the vcore for example? Orthos error could be something related to cpu especially when you say your memtest went fine. That's probably the first thing I would try.

[Robert]

My point is Orthos ran Large FFTs for 8 hours last night and another 8 hours the other night, yet today if failed 3*blend tests all with in 2 and a half hours.

[notsobig]

Hmm, blend test is much harder to pass compared with Large FFT so I see nothing wrong with that result? I still think you need bit more voltage to your cpu. And maybe little more to your ram.

[Robert]

Ehhhhh? Ram was at default 400mhz and my CPU was at 1.425

How is blend harder to pass? Large FFTs stress the cpu more, blend simply evens it to test more ram than Large FFTs (so I believe?).

[notsobig]

I don't know how I'm just talking from my experience But I had some occasions when I had to raise my vcore in order to pass blend test when same setting passed large FFT no problem.

[Robert]

Ah ok, cheers. I'll try it later.

[Robert]

Well last night I put my RAM to 44412 @ 2v. After an hour of WoW my PC hard locked. Today, I turned "mbm" off in the sensor setting of Orthos as I haven't got it, also I upped my vcore to 1.438 and changed my RAM to back to "auto" SPD settings @ 2v. Today it has done 9 hours in the orthos blend test and it's still going. Yesterday I did 3 blend tests all failed under 2.5 hours. I don't know whether to RMA my ram or to just leave SPD on auto. If I RMA'd my ram I'd buy some 8500 dominator stuff and sell the Geil when it gets back (if they find a fault?????).

[Clunk]

Did you try the subtimings? I had the same problem as you and that nailed it for me, with 2 different types of ram.