A month of hell

[strullis]

Hey guys, so after trying essentially everything in my arsenal I've decided I really, really need some help here. Here's the deal:

About a month ago, after my last desktop finally gave in after 4 faithful years (the harddrives), I decided to get myself a new rig instead of just buying a new HDD.

OS: Microsoft Windows 7 64bit
Processor: Intel Core i7-2600K
Processor cooler: Retail
Motherboard: ASUS P8z68-V Pro/Gen3
Memory: 4x4MB Corsairs XMS3 DDR-3
GFX: ASUS GTX570
HDD: Intel SSD 160MB 320 + 2TB Seagate 5600RPM
PSU: Corsair PowerSupply (PSU) 600W Gaming series 80 Plus
Additional cooling: 2 chassi fans
Additional componets: Old ass DVD-burner
Temperatures: About 42 celsius idle(55 while playing skyrim, while testing OCCT stress test and 100% stress it does reach high 70's though so I've never dared doing that for long, just tried it today as a matter of fact), 25-30 in the chassi and 50-60 on GPU

Now, initially I had really big problems when installing windows, bluescreened and zipfiles would end up corrupt etc; this turned out I had stupidly used the SSD on the marvell SATA-controller; that didn't work well. Anyway after fixing that stuff went much smoother, except like on average 4-5 times a day the screen either just completely freezes and the led-light on the outside chassi goes dark (i.e no activity) while the fans etc are still rolling, the only way to get stuff working again is by rebooting the system, the screen just completely freezes and nothing is responsive at all.

Maybe 1 out of 15 times out of these 4-5 times a day the computer randomly reboots itself instead. As for when the freezing (or as mentioned, way less often rebooting) occurs, it's completely random, it could happen three times in 30 minutes or it could happen once in 12 hours, it doesn't occur more or less often depending on load whatsoever, happens just as often while just idling in windows.

So I went to work trying to troubleshoot, started by memtest as with some of my previous rigs corrupt memory sticks has often been the problem, went 10 passes (10 hours or so) without a problem. Next step, eliminating the harddrives so I made fresh installs first on the SSD, to no help, and then removed the SSD and just usedt he slow harddrives, and it froze there as well.

After that I tried my old GFX (a radeon 6950), froze as well. This was a couple of weeks ago and I've just basically been too busy to get working on it so I've just had to live with the random freezes which of course tilted the shit out of me.

Anyway, at this point I figured it was basically motherboard or PSU left, so yesterday I actually went and bought a new PSU (a corsair 850w one) and a new motherboard (asus maximus IV gen3) figured this should finally solve my problem and installed them both (I know, technically I should just have tried the PSU change at first to not have to uninstall everything should that have been the problem but I was frankly too tired and just decided I'd pay for the freaking motherboard as well to get this overwith). Installed these (reapplied some arctic silver after cleaning the processor with non-oil based alcohol etc as always, still using stock coooler tho :S) and everything went smooth and I figured I'd finally be able to put this behind me.

That was light night, and after having the computer running when asleep and passing with no problems I figured it was all over. Until tonight, when the screen froze just in the same way as it had before. Yaaaaaaaawn.

So at this point I'm down to I guess two things, CPU or that the memories while having passed 10 hours of memtest really are corrupt anyway. The chassi is of course entirely not to blame, I guess I will however just put my computer in an isolated power outlet or something to not share with other peripherals like monitor etc but that seems like such a remote possibility given the older computer never once froze. I have no idea just how a corrupt CPU would manifest itself, and it seems strange it wouuld happen so (relatively) rarely as 4-5 times a day.

So here I am with no freaking idea of what to do except just randomly gamble-buy a new CPU which I really don't feel like, I don't have any other 1155-socket CPU to try on at the moment. I just do not know what to do, any suggestions? Oh and btw, I assume this is likely standard but the processor seems to throttle up and down (I see this while using the OCCT-stress test), clocks in at 1605mhz while idle and goes up to 3800 when playing skyrim for example, I assumet this is basically standard with I7's (or perhaps it's a feature of the asus motherboard in conjunction with the CPU). As for voltage to the cpu it sits rather steadily at 1,24V even during ultraload.

Should I try the OCCT-stress test for a longer while even though it takes the CPU up to basically 80 degrees celsius in only 3 minutes (from the idle temp of 43 or so), does this put the CPU in danger? FWIW, playing skyrim (and as I said this takes the CPU to 3.8ghz) it never goes above 55 degrees, it's only the real stresstest of OCCT that takes it this high.

Anyway, sorry for the long post but I'm at a loss at what to do currently, when the screen froze a couple of hours ago I just couldn't believe it.

(And yes, this likely means I bought a new mobo and PSU while having two fully functioning ones...doh)

Thanks a bunch in advance, and thanks for reading!

Best regards,

Strullis

 

[janas19]

Wow man, sorry to hear about your issues. Good thing you caught the SSD on marvell controller thing.

I would say the thing to do is return the motherboard and PSU, and try replacing the RAM. Get some regular, value RAM such as Kingston or Samsung, nothing performance. Try that for a period of time.

If you still have problems, the memory controller on the chip may be at fault.

 

[strullis]

Hi janas19,

Thanks. Yeah perhaps you missed it (in my long cluster**** of a post it's understandable) but I actually went and bought a new motherboard (asus maximus IV) and a new PSU (a corsair 850W one) yesterday so I've basically eliminated stuff like memory controllers being the issue, it's basically down to CPU, RAM that somehow is defect yet passing 10 hours of memtest or some other thing that I just have no idea about

 

[janas19]

Hi janas19,

Thanks. Yeah perhaps you missed it (in my long cluster**** of a post it's understandable) but I actually went and bought a new motherboard (asus maximus IV) and a new PSU (a corsair 850W one) yesterday so I've basically eliminated stuff like memory controllers being the issue, it's basically down to CPU, RAM that somehow is defect yet passing 10 hours of memtest or some other thing that I just have no idea about

I meant return the new mobo and PSU you purchased and use the old ones. You've eliminated them as being at fault, correct?

Oh, further clarification: the memory controller on the CPU is what may be faulty. But like I said, try swapping some standard RAM on there first. That may fix it.

 

[strullis]

Oh sorry, me being stupid again Yeah, problem is since I've opened/used them I don't think theyll have them returned

Oh memory controller on the cpu, of course, me being back in..95 or something when the controller was on the motherboard Anyway, I'll try some standard RAM as soon as I get the chance, thanks for the suggestion!

 

[janas19]

Oh sorry, me being stupid again Yeah, problem is since I've opened/used them I don't think theyll have them returned

Oh memory controller on the cpu, of course, me being back in..95 or something when the controller was on the motherboard Anyway, I'll try some standard RAM as soon as I get the chance, thanks for the suggestion!

This depends on the store. Micro Center and Fry's accept open box returns up to 30 days, I believe.

 

[krose]

It can require some tweaking to get 4 sticks of RAM stable on LGA 1155, and you have 8 GB sticks which is even harder on the IMC. Do you actually need 32GB of RAM?

 

[Ketchup]

It can require some tweaking to get 4 sticks of RAM stable on LGA 1155, and you have 8 GB sticks which is even harder on the IMC. Do you actually need 32GB of RAM?

Run one stick at a time and see how it goes. It can be hard to get memory to show errors when you want it to. Make sure the type and amount are listed on Asus' website.

 

[strullis]

janas19: I'm from Sweden and the stores unfortunately never have that policy here

krose: for some reason I must've written that by mistake, it's actually 4x4 sticks (and for almost all the time with the old motherboard I only really ran two of them (yeah I made sure the channels were correct) just to try fewer), but I guess checking into the timing of the RAM could be an idea, I've only used autotimings for the motherboard(s), maybe I should really look into that!

ketchup: yeah while I've tried running two of the ram sticks for most of the past month, I only really tried one of the pairs and let the other one collect dust (well, not literally ), I'll actually follow your example and run them one by one for something like 48 hours each and see whether that works.

Would the screenfreeze (and occasional reboot) be in line with memory errors? I remember an old rig where I had corrupt ram and then it often bluescreened for me, which hasn't been the case (I think, at least I can't remember a blue screen, only the freezes and reboots) since the very beginning when I had the SSD connected to the marvell controller.

 

[krose]

I see, you are having the problem with only two sticks. I have found that Memtest does not detect errors that well, I would try HCI memtest for windows and see what kind of errors you get when actually in the OS. For 2x4GB ram you should run 8 instances with about 850MB each, for a total of 6800MB. This finds memory instability better than anything else IMO. It could be your motherboard just doesn't like that particular memory.

 

[strullis]

Hi guys,

So last night I tried out the HCI memtest, but with all 4 by 4 sticks in, starting 7 instances at 1700mb each, only after a few % the following occured (in sequence although it all went fast) firefox as I had in the background doing nothing crashed simulatenously with memtest saying there was an error, and only 3-4 seconds later the (now infamous) screen freeze arrived and I had to hard reboot. So while the test worked in that sense, it didn't work as planned (i.e in finding errors it usually just stops the test and you can go on, here as soon as it found an error the computer just froze in the characteristic way of my problem). Anyway, finally I had some hope that perhaps the RAM really is the culprit!

However, after that reboot I again started the same nr of instances of HCI memtest at 1700mb each, and lo and behold, after waking up today some 7 hours later all the instances had passed 530% with no errors...I have no idea whats going on here, I guess I should do the plan of using just one stick of ram for multiple days and see how that goes?

 

[Ketchup]

It may just be a voltage issue. Giving the memory an extra .2 volts may do the trick. It won't hurt the memory.

 

[strullis]

Thanks for the swift response once again ketchup (and all of you), I'll try the voltage increase.

Oh and btw; given that rather odd turnout of the hci memtest, would you guys say it's likely that the RAM (or potentially the memory controller on the CPU) is the actual problem, I'm only asking because of the computer actually freezing seconds after the memtest said there was an error with the memory rather than the normal procedure of finding an error and just stopping the test but without the actual computer crashing.

I.e would the memtest basically give a memory error regardless of what causes the freeze (when the freezing arrives as it did) or is that quite indicative of likely RAM error/corruption?

 

[Ketchup]

Thanks for the swift response once again ketchup (and all of you), I'll try the voltage increase.

Oh and btw; given that rather odd turnout of the hci memtest, would you guys say it's likely that the RAM (or potentially the memory controller on the CPU) is the actual problem, I'm only asking because of the computer actually freezing seconds after the memtest said there was an error with the memory rather than the normal procedure of finding an error and just stopping the test but without the actual computer crashing.

I.e would the memtest basically give a memory error regardless of what causes the freeze (when the freezing arrives as it did) or is that quite indicative of likely RAM error/corruption?

That one is a tough call. But one thing I will say is that a bad CPU is extremely rare. A bad stick of memory, not so much.

 

[janas19]

That one is a tough call. But one thing I will say is that a bad CPU is extremely rare. A bad stick of memory, not so much.

It is rare, but not unheard of... I don't know how exactly one would diagnose the on-chip mem controller as bad. One AT member, SaltBoy, had an issue where his system would not POST. We basically tried everything to fix it. He replaced every single part except the CPU, and it still wouldn't POST. Finally, he said when he replaced the CPU, it worked. He said the mem controller was bad.

So perhaps, if the mem controller is bad, it won't even POST. But then again, your mem controller may work but be glitching. I'm not sure. Again, it is rare, but it can happen.

 

[Cr0nJ0b]

I'm so sorry to hear you are having issues. I wouldn't normally post a response to this type of topic, but as I read your it struck me that I had the exact same issues.

It's funny, I did almost the same thing when i got my new motherboard (ASUS P8z68-V Pro/Gen3).

I went so far as to get an RMA on the board and my Intel SSD.

at first it ran fine, when i moved my disks from the old motherboard to the new one, but when i reloaded the OS, all hell broke loose. Half of my attempts to install the OS failed with memory issues and then I would get that freezing...and for me it was amplified by the fact that I corrupted my RAID 0 data set (all my data was on that) when i did the reload and I couldn't get it back because of the freezing.

So what happened to me? I don't know for sure....because I tried so damn many things.

I tried an OS reload, new SATA drivers for both Intel and Marvell, swapping sata cables, ports etc.

In the end I think it was a combination of the right ports (all HDDs are on Intel) and selecting a bios option for CPU speed step of something. I can't recall right now but I'm sure the folks on the board will know. There is a setting in the bios that allows the various cores to go to sleep and wake up, and I read a couple of places where this can cause lockups. I changed that and the ports and I've been sailing 100% error free for the past few months. I did some hardcore benching and I game a lot and have only had one or two lockups (probably HDD related). If I can dig up my manual I'll try to post the setting.

 

[raf051888]

I actually just finished an RMA for a bad CPU. I thought it was rare also, but the motherboard booted just fine with another CPU.

Have you checked the event viewer after rebooting from a freeze?

 

[polnord]

I have the same os, processor, mobo, and memory as strullis in post #1 with a Rosewill 1000w psu, Seagate 1TB 7200 rpm hdd, Hitachi optical drive, Hyper 212 Plus cpu cooler in a HAF 932 case. I do not have a graphics card because I am not a gamer. I have been experiencing what are probably the same problems as strullis described that have included BSODs, freeze-ups, sudden automatic shutdowns & restarts, and what has been called "video distortion" by tech experts. The video distortion has included "snow" and moving diagonal lines like something from the days of B&W TV. I had an Intel mobo originally and was told that it was most likely the problem. Changing out to the ASUS mobo did not make the problems go away. There must be some design flaw that is behind the rash of enigmatic problems, and my best guess at this point is the i7-2600K processor. I'm paying a knowledgeable expert with years of professional experience to get my system to work, and it has stumped him so far--more than two months into this with lots of diagnostics, etc. For me, my new system has been more than two months of hell.

 

[strullis]

Earlier today I flashed my Maximus motherboard just to make sure I have the latest bios (I did flash the earlier Asus board as well so that was never the problem), and am currently running on a single 4mb stick of RAM. I'll be running each of my sticks for at least 48 hours (since there's never been that long of a period uninterrupted of freezes/random reboots), and if its working, I'll add a second at that point. If it then freezes I'll try the added stick on its own and repeat.

Also, just a curious question: for some reason in the device manager it says my computer is an ACPI-system (but the hard drive controllers says ahci), and yes I do have AHCI set in bios. It seems this is purely cosmetical, i.e the hard drives are running with an AHCI controller whereas the "computer" says ACPI x86 PC. Anyone have a clue why this might be? FWIW, on the previous asus motherboard it did say AHCI if I recall correctly.

raf051888: All it ever says is unexpected shutdown/didn't shut down correctly (critical error), and never anything else.

 

[krose]

ACPI is a power management feature, Advanced Configuration and Power Interface. It has nothing to do with AHCI. All modern computers will show ACPIx86 or ACPIx64 under computer in device manager. AHCI is Advanced Host Controller Interface and is the controller for your SATA disk drives. Totally unrelated.

 

[strullis]

Oops, that's what I get for being stupid Thanks for the clarification krose.

As an update, running only one memory stick and leaving the computer idle over night it's now not crashed for about 15 hours. I'm just so so so hoping it stays stable for another 24 hours or so to be able to finally isolate the problem.

 

[strullis]

Hi,

So yeah, by this point the culprit is clear: it's one of the RAM. After 72 hours of stability on one stick I can finally draw a sigh of relief to have the problem located. When I find the faulty RAM it's a cointoss between RMA and going at it with a hammer for all the problems it's caused me lol.

Anyway, thank you all for the support and suggestions!

 

[polnord]

At long last my neighbor who really knows computers came to my rescue and solved my problems. The "expert" who had cost me much had made a number of mistakes in installing things. The metal standoff posts were in the wrong place, so the board was shorted out against the case. Some cables were connected wrong. My hardware was all good. The system is now up and running in good working order.

 

[Ketchup]

Thank goodness for nice neighbors!

 

[Meaker10]

Installing the motherboard correctly is 101, I would ask for my money back and sue if he refuses to be quite honest (if a "professional")