Possible memory leaks and BSODs

[PCRHitman]

Hi All

I have a rather frustrating problem with my new rig, I’m getting numerous BSODs. The prominent one is page_fault_in_nonpaged_area, I have also gotten system_service_exception and memory_management.

I’m running an i7 920, OCZ Reaper PC3-15000, Gainward GeForce GTX275, X-Fi Fatal1ty Pro on an Asus P6T, powered by a Be Quiet! 650W.

I set up this rig about 2 months ago, and everything ran beautifully. I started out using windows 7 RC build 7201 x64. this worked fine for about 2months. But about 2 weeks ago I suddenly started to get BSODs. The cheeky little buggers would sneak in when I went to the loo and I’d return to the windows log in screen.

After a good googling, I came to the conclusion it could be one of a few things, 1)memory problems, 2)corrupt drives / hard disk, 3) motherboard problems. I ran the windows memory test, which brought back errors, so then I ran Memtest 3.4, 3.5 just doesn’t work, for a number of passes and the memory came back clean. This was mighty odd, I know windows memory test is a bit shody, but didn’t think it would detect errors that memtest doesn’t. so then I ran the windows memory test again, that cam back clean. So I decided to replace my hard drive, it needed doing any way and this gave me a good excuse, so now I have 2x 320gb Samsung Spinpoints in RAID0. I reinstalled windows, now using win7 RTM x64, and again everything seemed to work fine, for about a day. Then the BSODs returned.

So, I’ve pretty much ruled out hard drive / corrupt drives, and I thought I’d ruled out memory problems. So I then updated my bios to the latest version to see if that would work. Alas no joy, and instead of blue screening straight aways sometime things just start crashing, the windows transparent themes dies, then explorer, then other programs like photoshop and visual studio, just crash then I get my blue screen. So I decided to run mem test again, and now I’m getting some errors.

Before I RMA my memory, does anyone have any experience with any similar problem, could this be my motherboard playing up as well as, on instead of my memory. As on occation during my crashes, the system would boot up showing 4gb memory instead of 6. this I assumed at the time to be a memory problem with 1 stick just dying, but I’m not sure, haven’t had that same glitch for a few days.

Thanks in advance

 

[twist3d0n3]

have you tried just having 1 stick of ram in at a time and doing memtest? have you overclocked your system?

 

[PCRHitman]

I haven't yet tired just running a test on a single stick of ram, presumably doing this would just help me identify which stick if the faulty one, and as its a tripple channel set i have to send back all three sticks so i didn't see the point. Is there any other reason for testing one stick at a time?

And no i haven't gotten round overclocking yet, my system is watercooled, but my temperatures are higher than my last set up becuase i have a completely internal system now. but thats a different issue, that i'll combat when i have the time, but my idle temps are not high, CPU is at 32-36C and System is at 40-42, and under load they don't go much above 50. So i don't think there is a problem there

 

[twist3d0n3]

not sure on the RMA process, might be the 1 stick they'd swap. emailing them would probably give the best answer unless someone on here can enlighten.

what voltages are things running at? it could be a dodgy qpi voltage? had it before with my mates' rig (i7 920 d0, 6Gb corsair ram, asus rampage 2) where we set up an overclock, restarted it said it failed, went into bios and it set ddr voltage to 2.25v :S worth a look if you haven't checked. it could be something to do with the mem controller on the cpu - if the ddr voltage/qpi has automatically gone to some random high voltage then it could've damaged the cpu instead. just a thought

 

[Moeks]

It certainly sounds like a memory issue to me, or possibly the motherboards memory controller and or ram slots.

You need to try testing each stick of memory in each ram slot really. I have been having 'similar' issues to you with my new I5 setup, here is a link to my thread, you might find some of what I have to say in there useful.

But I'll say this now, ditch Memtest 3.5, it has serious problems, infact version 3.0+ does, it's been recoded by the guy that wrote the program I believe, the general recommendation is to use Memtest86+ v2.11. Memory problems can be some of the most frustrating and difficult issues to diagnose definitively.

I haven't been able to get any closer to resolving my issues yet because OcUK didn't deliver my new memory today (despite me completing the order at 15:58 ...this is twice they have done that now, in 2 weeks ...come on guys, you used to be great at delivering on time), however it has been dispatched so I should get it tomorrow.

As for the RMA process, if the memory was a matched kit, then it needs to be sent back as a kit, i.e. all the sticks ...and you will get a new kit.

 

[PCRHitman]

Have just had a read through your issue Moeks, looks like your having an equally hard time. I do seem to remember reading a while back about memory working together in dual channel and having to get memory that will work together. like you said each stick works fine, but when put together they just don't work.

I had memtest 3.4 working ok before, but now it just doesn't want to work. I've RMA'd my ram, in doing so though i spoke to tech support, who i have to say are relitively useless. all they could give me was telling me not to run windows 7, which is total crap.

I do think i've got to the bottom of this though, as you said moeks, it really comes down to the Ram stick of the motherboard (either the memory controller or the ram slot) so to rule ok the motherboard i am now using only 1 stick of my ram. The first stick i tried blue screened within 5mins. Swapped that out with another and my machine has been stable for almost a day now... that doesn't mean that its not the motherboard, however it does seem to suggest that its certainly my memory.

Which is a bit of a relief, i really can't be doing with taking everything out of my machine, even though with my new cooling set up it is a lot easier than it was.

Considering i am not overvolting anything, and everything is at factory settings, apart from the memory speed which has to be manually set to 1866, is there anything obvious that could be corrupting the RAM? Could it be running too hot?

 

[Moeks]

Hehe well I was, I seem to have solved my problem, just updated my thread. Basically I put the new memory I ordered it a voila, all seems well.

I'd say you have 1 good stick then from what you have said, as you say, it's not 100% certain but highly likely.

Best of luck getting it all sorted out anyway, memory issues can be a real pain in the rear to get to the bottom off.

Now it remains to be seen weather my Windows installation is ok or not, I did install it with dodgy memory after all, I haven't had a crash yet, and I've been running for an hour now. I've put the machine under some stress too. Fingers crossed.