Sporadic Bluescreens

[harbin91]

Currently typing this up on my dad's PC whilst my PC memtests, and apologies if this is the wrong forum.

Basically this is a problem that I have been having for a while now. I am getting sporadic blue screens during gaming. This has occurred most commonly in Civilisation V but also in GTA IV and Just Cause 2, although there are one or two other games that I haven't had this problem in (Skyrim, although that just CTDs instead, PES13 and FM13 have been fine, although FM13 isn't a graphically intensive game).

These BSODs have so far centred around two things, 'Page fault in Non-Paged Area', the odd 'IRQL_not_less_or_equal' and most recently a 0xBE error (An Attempt was made to write to read-only memory). This issue has directed me to believing the RAM may be at fault hence the current running of Memtest86+. It has been on overnight and has just completed it's 4th pass after 12 hours with no errors as of yet. I'm not quite sure how long I should leave it running before I can be confident that is not at fault. I have also ran FAH sessions on the CPU and GPU (which is full load on the cores) and in the past done a Furmark burn in test on the GPU, which put it at full load. Both ran fine and not one blue screen. Alongside this I have done some instances of RAM benchmarking with no issue, although I haven't done a Prime95 test.

These blue screens have only ever occurred in games, with no errors opening in Windows which generally indicates to me that the problem could be with faulty hardware that can't handle being on full load (which if that was the cause should've failed on the stress tests), but at the moment have ran out of ideas on where to look. DXDiag has reported no problems, and all my drivers are uptodate as far as I can tell.

My system specs:

MOBO: Asrock 970 Extreme 3
CPU: AMD Phenom II X4 965 BE @ stock 3.4Ghz
RAM: G.Skill RipjawsX 1600 CAS9 16Gb (4x4Gb sticks)
GPU: Sapphire 2Gb Radeon HD6950
PSU: Thermaltake Toughpower 575w (I believe)
SSD: 64Gb Crucial M4 (Windows Installation)
HD1: 1.5Tb Seagate Barracuda (data drive)
HD2: 1Tb Samsung F3 (partitioned into three drives for games, programs and page file)

I have got a RAMDisk operating on the PC, at 4Gb which is where cache files for Spotify, browsers and Zip files are extracted to if I only need them once or twice go. I did change the software that ran it but these errors occurred before and after that change.

I also have DaemonTools installed but don't run it when I don't need an .ISO mounted. I only start it up and create a virtual DVD drive as and when needed and close it when I've finished. I have read that people have had issues with that in the past with blue screens but like I said it's inactive during gaming sessions.

The System is not currently overclocked. I did have an overclock running which I tested to be stable but for some odd reason Civ 5 decided it wanted to blue screen every ten minutes or so in the game. That was running with the CPU at 4Ghz and the RAM having a rating of 8-7-8 (or 8-8-8) with no touch to the voltage. At the moment the RAM is running at it's stock 1600 speed since the board runs it at 1333 overwise, although I've no idea why this could cause issues since it is running at the spec stated. The board has four RAM slots with all populated. I do have a custom air cooler installed (a Noctua D-14) so overheating is not an issue, although the size of it has made the RAM very difficult to access, but there is a clearance and nothing is making anything short.

I have a 16Gb Pagefile on a partition on HD2, and running a small page file on the SSD for Windows to write dumps to, since the SSD is only 64Gb and space fills up damn quickly if I'm not careful and conservative with it.

Memtest86+ is currently working on it's fifth pass with no errors (halfway through Test 4 as I'm writing this), should I stop it soon after and assume the RAM is free? This is the standard test in Memtest86+ btw.

Basically I'm running out of options as to why it is giving out these sporadic blue screens in games, since it is only in games that the issues occur (or have done so far). I've ran several VMs with no errors of the sort.

[Tradesman]

I would guess it might be a couple of things, the primary being that with 4x4GB and tight timings it may be stressing the MC (memory controller) of your CPU (AMD has weak MCs that are actually aimed more towards 2 stick setups, 1 DIMM per channel)....might want to raise your CPU/NB voltage by + 0.06 and possibly the actual DRAM voltage + 0.04 and see if that helps. Another possible is to ensure your video drivers are the latest available as they are constantly updating them for the various games out there. Check those and let us know

[harbin91]

Hi Tradesman, thanks for the reply.

It did do the fifth pass without any errors popping up.

But you're basically saying that it could be overtaxing the memory controller? The tight timings are what is on the spec for the RAM though but basically that question now is tight timing vs slacking off the speed. If that is the case however, would RAM benchmarks be able to indicate if the memory controller is what's causing the bluescreens by being pushed too far by the RAM?

I would've preferred the 2x8Gb sticks though but at the time I didn't know I was gonna get an additional 8Gb. The only 2x8Gb sticks I can find are the 1866Mhz ones (cheapest found being £78) or a single 8Gb stick of the 1333mhz one at £38. I'm not really willing to spend the money on that if it isn't going to fix the problem.

I have increased the NBFreq speed to 2400, although without checking the BIOS settings I don't know if or what the voltages are changed to. AFAIK they are still on auto though, along with the RAM which should be at 1.5v.

The video card drivers are up to date though, running Catalyst 13.1. I did end up having to do a clean install of it because the installer messed up all chances of a straightforward upgrade.

I did read somewhere that the AMD CPUs don't fully support speeds above 1333mhz on the RAM so could running it at 1600Mhz contribute to the problem?

Not sure if it matters much but it is the C3 stepping version of the CPU.

[Tradesman]

Looks like you've done your research, yes, I mention the MC as the C3 965 is the lowest AMD CPU I recommend for running 1600 sticks (and many have problems with more than 8GB due to MC limitations, which is why normally suggest a little extra voltage to compensate for the 4 sticks (and/or over 8GB), AMD does their testing w/ 4GB sticks generally as a max so when you climb up to 8GB sticks you can also run into problems. I myself have long advocated using large amounts of DRAM as there are a number of things you can do with it (like RAM disk) more multi tasking, etc and it was the weak MC that led me to wuit building on the AMD platform....I still keep hoping though, I picked up an 8150 just before they were released and the same issues continued as they have with the 8350, it's better today with the FX line (can actually run up to around 1866 and 16GB or 32GB at 1600 w/ a good 8350, but the Intel CPUs just continue to go up and pull away, an older 2500K can easily run 32GB of 2133 and OCs better, plus even though it's only 4 core it will outshine an 8 core 8350 whose performance per core declines as load increases (plus it can't handle the additional memory Intel can to help compensate. Below is a hidden away freq guide for AMD CPUs :

http://support.amd.com/us/kbarticles...encyguide.aspx

[harbin91]

TBH I am hoping to upgrade this PC in the near future when budget allows, to an Intel/Nvidia purely because generally the performance is ahead of AMD, which at the time I went with due to cheaper costs, alongside the Intel having lower power consumption, and temperatures.

Would I be right in assuming the 'Memory Rank' is just AMD jargon for channel mode?

If AMD had been more impressive with their FX range, I would've probably just upped to an FX-6xxx or FX-8xxx series CPU but they didn't really pull off any major improvements in performance. Hell from some benchmarks I've see, the Phenom X4s outperformed the FXs. With the Bulldozers and the Piledrivers it sounds like AMDs marketing and development departments went completely into thinking more cores = more power = better performance and then got trumped when it turned it was false.

Makes me wonder sometimes though, what is the point of having 2400Mhz RAM if the controller ain't gonna be good enough to take full advantage of it at the rapidly increasing capacities... especially since looking at the table, AMD recommend me putting the RAM down to 1066Mhz speeds, not even sure how much of a read/write hit I'll see with it.

I'll start tweaking and messing about with it either tonight or tomorrow though, and hopefully give you an update, but not sure how long it'll take to find out how reliable the changes would be. Like I said, it's sporadic.

[harbin91]

Right so I've increased the DRAM voltage from 1.5v to 1.54v (+0.04 as stated) and the CPU-NB from something like 1.10 to 1.165v (as near as possible to the +0.06v).

I've not touched the frequency or timings of the RAM yet, but I'll see how that goes.

[Tradesman]

Basically yes on the rank/channel thing (though it depends on who you talk to. And I agree on the FX line - the performance itself is pretty good but since they skimp on the R&D in improving the MC (memory controller) your stuck with smaller total amounts of DRAM than Intel and lower freqs which put's a hit on performance as the 8 'cores' could be much more effective if everything was in DRAM when multitasking rather than the cores all having to run through the HD Controller and swapping out info from the virtual DRAM on the HD.

Anyway, yes, keep us updated on how the voltage change do

[Britton30]

You might try timings of 10-10-10-28 too and see if that helps.

[GSKILL TECH]

Did you use the guide to configure RAM settings properly? Special settings are required for that hardware combo.

Thank you
GSKILL TECH

[harbin91]

Sorry about the time difference between posts. I did fix the issue I was having, but unfortunately yesterday another problem has started occurring with this PC, though I'm not sure what is now causing this.

I'm just getting a random STOP: 0XF4 bluescreen... or was previously, now it seems Windows doesn't want to chuck a bluescreen in my face, it just crashes instead though the mouse was still moving the previous time it did that.

[GSKILL TECH]

Test each RAM to see if one is defective.

Thank you
GSKILL TECH

[harbin91]

I discovered the problem was with my SSDs firmware causing it to stop every hour of uptime it had. Fixed by updating it now.

[GSKILL TECH]

Fantastic! Great to hear you found the culprit.

Thank you
GSKILL TECH