Take the 2-minute tour ×
Super User is a question and answer site for computer enthusiasts and power users. It's 100% free, no registration required.

Although I'm a long term user of other StackExchange sites this is my first question on SuperUser, so please let me know if I've neglected any community customs

I am experiencing regular BSOD crashes on a new build which seems to be linked to the number of memory modules I have installed. So far the crashes have only occurred when all 4 DIMM slots are occupied, I haven't yet seen any crashes when only 2 are occupied. The crashes seem to mostly occur when running full screen graphics.

Environment:

  • Gigabyte GA-Z87-HD3 Motherboard
  • Intel i5 4670k CPU
  • Corsair CX600 PSU
  • 16GB (4*4) Crucial Ballistix Sport RAM
  • XFX AMD 7870 DD GPU
  • Windows 8 (64bit)
  • Antec 900 case (plenty of cooling)

I originally bought the Motherboard \ CPU \ 8GB RAM as a bundle from Novatech and everything seemed to work just fine. A week later I bought an extra 8GB of RAM (identical to what shipped with the bundle) and this is when I began to see the BSOD crashes.

I've analyzed the memory dumps with WinDBG and there doesn't seem to be a pattern to what's causing the crashes - WinDBG has pointed the finger at graphics, audio, and even USB drivers as the culprits.

I took the additional RAM back out and the system stabilised again. At this point I thought the new RAM might be faulty, but I tried swapping the original RAM for the new RAM and the system was again stable. The crashes only appear to occur when both sets of RAM are installed (all 4 slots occupied).

I have run Memtest86+ against all 4 DIMMS (only 1 pass) and it did not find any problems.

I read somewhere that some motherboards have trouble when fully loaded with RAM running at maximum speed (1600MHz in this case). Although the motherboard manual makes no mention of this, I went into the BIOS and underclocked the RAM to 1333Mhz. After experiencing another BSOD, I underclocked again to 800Mhz. I am not yet sure if that's fixed the issue.

I'm running out of ideas on what to look at next, does anyone have any recommendations on where to go from here or what might be at fault?

I have a horrible feeling in my stomach that it might be the PSU, are these crashes consistent with what you'd expect to see with a faulty PSU? Is 600W enough to power this system?

It could equally be the video card. I might take it out and try running the full 16GB RAM in conjunction with the onboard GPU and see if that's stable.

EDIT

After living with the BSOD crashes for a few more days, some patterns have emerged. The crashes mostly occur during full screen graphics, and most of the time, analyzing the memory dumps WinDBG is identifying the graphics drivers as the culprits.

Although I had the latest graphics drivers (13.4), I decided to give the latest beta driver a try (13.10). I wouldn't usually install a beta driver but I had nothing to lose.

That does seem to have made the system a lot more stable. The BSOD crashes still occur but far less often, maybe once every 4-6 hours. It looks like this whole issue is nothing to do with the RAM, just unstable ATI graphics drivers on Windows 8.

I've been using nvidia cards for years but I thought I'd give AMD a try... maybe that wasn't a great call :)

share|improve this question
1  
Many Intel boards (or rather boards with an Intel CPU with on-board memory controller) allow for tighter timings when only one set of DIMMs is installed. Did you adjust the timings to the new maximum values when you use all slots? (The BIOS should do this automatically on the default setting. However if you manually set things and add DIMMs/ranks than things may get unstable). –  Hennes Sep 4 '13 at 13:51
1  
A single pass isn't enough to test the stability of your memory. –  Ramhound Sep 4 '13 at 13:56
1  
I'd recommend trying it with only the onboard GPU with full RAM, I assume that should probably also rule out the PSU sufficiency (the onboard GPU is likely to require less power, unless it's about the power requirements of the motherboard as a whole) if the result changes to either direction. Also you could run some sort of a high load on the box so that the GPU is not involved (I do not know if enabling HW screen saver helps reduce the GPU power drain but it's worth trying anyway if that makes any difference). –  zagrimsan Sep 5 '13 at 10:10
1  
Or even with 2 DIMMS, can you try them in different slots? So if the system is stable with 2 DIMMS in 1+2, then try putting this same DIMMS in 3+4 (if your mobo allows it). It could be that one of the slots is fuggered. –  icabod Sep 5 '13 at 13:00
1  
@MattDavey - More then a single pass. Move on to the next possible cause. –  Ramhound Sep 10 '13 at 12:42

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Browse other questions tagged or ask your own question.