Help needed with machine lockups, thinking memory at fault

ukman408

Honorable
Oct 8, 2012
29
0
10,530
Sorry for the long write but I wanted to provide as much detail as I could.

To start here's my system: (all factory, no overclocking)
Win7 Pro
Intel i5-2500k CPU @ 3.3GHZ
ASRock: Z75 Pro3 Motherboard
Nvidia Geforce GTX 650TI Vid Card
OCZ Agility 4 128GB SSD
500GB Seagate HD (slaved)
8GB G.SKILL Sniper low voltage series DDR3 1600 SD Ram (2X4GB)
Cooler Master Hyper 212 Evo Direct Touch 4 Heatpipe Heatsink

CPU-Z is reporting that JEDC#1 #2 & #3 are at 1.28V and the XMP-1600 is at 1.25

Issues started when using chrome to play flash media, mostly facebook games and vidoes on sites like yahoo/hbogo/hulu. PC would do a hard lock. Couldnt move mouse or do anything, only hard reboot.

Swapped to using Waterfox (64bit firefox). Was able to go a week or so before the issues started the same there. With waterfox I mostly noticed it when I would make a video or stream full screen. A few days later any flash use over a few minutes would cause the hard lock

Uninstalled each browser and flash player associated with them. Cleaned the registry and PC of any trace of the software and reinstalled fresh. Ran multiple maleware and virus scans as well. No threats found. Issue remained however.

Swapped to IE (GAWD HELP ME!) It worked for a day or so but started doing the same thing as well. Finally had to swap to using a virtual XP machine to pull any flash related media or games.

Took this oportunity and updated every single driver on my system. Most were only a version or two out of date but did them all anyway. Even upgraded the bios.

Issue remained. Sadly for me, a week or so after giving up on flash media (outside the virutal PC i run for older games) I started seeing the same lockups when playing regular games like Assassin's Creed IV and Arkham Origins. I could play for anywhere from 5-15 minutes and get the lockup. The sound would go bonkers screatching and squacking in loops. If I let the system sit for 5-10 minutes it would USUALLY shutoff and startup again reporting an unexpected crash. Once or twice I let it sit for 30 minutes and it never shutdown or restarted on its own. No blue screen errors. Nothing.

Checked the system event logs, nothing in there. Nothing prior to the crash or after giving detail as to what might be happening.

Ran memtest86 for 12 hours and it reported no errors.

Ran a PC burn in test (i forget the name of the program, already uninstalled it) but it reported no errors.

I tried running OCCT but I couldnt really decifer anything out of it. I ran the GPU test first and the only thing that stuck out to me was a spike in heat up to around 90celcius during the peak of the test and that only lasted a minute max.

I ran the OCCT CPU test and thats where I started seeing results...sorta. I chose the test mode as a large data set first not knowing what it did. It ran for maybe 5 minutes and locked. Again I could still see my screen but nothing was moving. Couldnt move mouse or use any keystrokes. No errors were showing on the screen the the voltages and temps seemed ok. I read online to find that the large data set test focuses more on memory testing. The medium test does a balance of CPU, heat and memory while the small dataset checks for CPU and heat.

I reran the test again at medium and it ran for maybe 20 minutes before the same lockup happened. After reboot I ran the same test again witht he small data set and it ran for hours without a crash and no reported errors.

Read some more and saw suggestions to use Prime95. Seemed similar to OCCT in the types of tests but it spelled it out better. Prime ran fine for an hour on the CPU test so I swapped to the balanced test, it ran fine for an hour before I stopped it. I then ran the "blend" test which is ram heavy. It made it through the first 4 tests and locked up while doing the 5th. The first 4 reported no errors but the 5th test left me nothing as it locked before it could report a pass or fail. I checked for a results log after a reboot from the crash and it didnt leave one. Guess it locked hard before it could write anything. Ran the test again and again it locked on the 5th test.

All this leaves me to think that the issue is not with the GPU nor the CPU but somewhere with the memory. Its getting late and I wanted to get this posted before bed but I plan tomorrow to try swapping out one stick of memory at a time to see if I can narrow it down to 1 stick.

Also as a note, I do work IT at a helpdesk and have a lot of experience troubleshooting software but not so much with hardware. i did take the PC apart carefully (grounded myself against shocks) and cleaned the system with a can of compressed air that we use often at work. There was some dust buildup but nothing major.

Thoughts? I'm at a loss where to go if swapping out the memroy sticks doesnt lead anywhere.

One suggestion a co-worker had was to play with the memory timing. I know its in the bios but I've been to afraid to do any overclocking. I'm just dangerous enough to think I can build a PC myself (which I did) but havent been brave enough to try out overclocking yet.

If there's anymore info someone needs, I'll be glad to get it and post it back here tomorrow. Not sure what else might be useful.

Thanks in advance.
 

ukman408

Honorable
Oct 8, 2012
29
0
10,530


Ok. I checked the settings when I got home tonight. The timing was 9-9-9-28 2T The voltage was set to 1.28V

My co-worker suggested lowering the timings to something like 10-10-10 or lower. I wanted to try it the other way first just to see what would happen. I changed it to 8-8-8-24 2T and set the voltage to 1.5.

I booted into the PC fine and right now I've been running that same stress test that kept locking up last night and its been running for 30 minutes without a crash so far. Odd.

One question though, the DRAM voltage is showing as 1.5v in the asrock motherboad utility but when I look in cpu-z its showing the memory voltage is at 1.28 still but does reflect the new timings. There were a LOT of options in the bios but i left everything at auto except for the timings and voltages.

One other note to add, this system had run stable without any issues for over 9 months. These lockups are more recent issues over the past 2 or 3 weeks. The bios is up to date though to answer your other question.
 

ukman408

Honorable
Oct 8, 2012
29
0
10,530


Left for a few hours and came back to find the PC still running. One of the 4 workers in prime95 had stopped around 110 tests with a fatal hardware error but didnt give any further detail as to what the error was. Something about expecting a number below 0.4 and the actual number came out 0.483738747 or something like that. I swapped the sticks to different memory slots to see if that makes a difference and will test again overnight. At least it seems modifying the rates at least lets me run now without a hard lock but I do know there's still issue. If I see more errors in the morning I'm gonna take 1 of the sticks out and test again while I'm at work, and then the other stick tomorrow night. Thanks for the help. if I can at least get it down to one stick causing the problem I can replace easily enough. Based on the board I mentioned above, do you have any suggestions of a replacement that would run well on it?