Solved Occasional BSODs during everyday use

[smiles86]

Custom built my desktop during June-July 2013. I have been experiencing occasional BSODs during regular use such as web browsing or word processing, though only the latest 4 logs seem to have been saved. I did not observe any pattern to the BSODs or any correlation to a specific activity.

I initially suspected faulty memory sticks to be the culprit after investigating the earliest of the 3 logs available. The memory sticks in use were Crucial Ballistix Sport 8GB (2 x 4GB) 240-Pin DDR3 SDRAM DDR3 1600 (PC3 12800) Low Profile Intel Desktop Memory, which came free in a bundle with the case I bought. I have since replaced them with Corsair Vengeance 8 GB ( 2 x 4 GB ) DDR3 1600 MHz (PC3 12800) 240-Pin DDR3 Memory Kit, but a BSOD occured on that very same night after replacing the sticks. However, note that the BSOD dump now refers to a generic hardware fault (0x124) as opposed to specifically mentioning a memory fault. Have I now replaced the initial problem with another? Or is the same inherent issue, whatever it may be, still to blame?

The BSODs do not occur frequently enough to be a major problem, but they remain an annoyance as I lose work when they occur without warning. Any help is appreciated.

*Edit: I have already run sfc /scannow and memory tests. Neither reported any issues. All drivers and BIOS are updated to latest versions as far as I am aware.
*Edit2: Nothing is overclocked.The GPU comes factory overclocked, as noted below.

 

[Arc]

Hi smiles86.

Your latest BSOD reports a hardware issue, and the before latest one reports a display failure.

So, the first suspect goes to overclocking.

Some of your hardware are factory overclocked, or overclocked otherwise? The GPU seems to be.

Name    NVIDIA GeForce GTX 650 Ti BOOST

As you are getting BSODs, you should stop overclocking and run all the hardware components like CPU, GPU and RAM to their default settings.

Is the computer hot? Report us the heat of the computer after a couple of hours of your normal usage. Upload a screenshot of the summery tab of Speccy. Alternatively, you can publish a Speccy snapshot too: http://www.sevenforums.com/tutorials/311593-speccy-publish-snapshot-your-system-specs.html .

Check if the Power Supply Unit (PSU) supplying adequate power to the computer or not.
eXtreme Power Supply Calculator
Also let us know about the wattage of the PSU.

Install the NVIDIA DRIVERS 314.22WHQL only.

• Uninstall All nvidia items in Control Panel > Programs and features
  • 3D Vision Control Driver
  • 3D Vision Driver
  • Graphics Driver
  • HD Audio Driver
  • PhysX
  • nvidia Update
  (Are you using nvidia chipset drivers? If so, dont uninstall anything other than those are listed).
• Now follow Drivers - Clean Left over Files after Uninstalling
• Download 314.22 WHQL. While installing, Select Custom (Advanced) install. In the next page, follow this settings:
  
  

Make it sure that the Samsung SSD is running with the latest firmware.

• Samsung SSD (Solid State Drive) 840 series | Samsung SSD Website
• How to Upgrade Samsung SSD Firmware | StorageReview.com - Storage Reviews

Test your RAM modules for possible errors.
How to Test and Diagnose RAM Issues with Memtest86+
Run memtest for at least 8 passes, preferably overnight.

If it start showing errors/red lines, stop testing. A single error is enough to determine that something is going bad there.

Stress test the Graphics Card using Furmark.
Video Card - Stress Test with Furmark

Stress test the CPU.
Hardware - Stress Test With Prime95

Let us know the results.
________________________________________________________________________________________

[COLOR=Red]BugCheck 124[/COLOR], {[COLOR=Red]0[/COLOR], fffffa800713f028, be000000, 100110a}

[COLOR=Red]Probably caused by : hardware[/COLOR]

Followup: MachineOwner
------------------------------------------------------------------------------
BugCheck 50, {ffffffff8c0f3dfa, 1, fffff80003787c25, 5}


Could not read faulting driver name
Probably caused by : ntkrnlmp.exe ( nt!ExpQuerySystemInformation+283 )

Followup: MachineOwner
---------

 

[smiles86]

I am in the process of going through all the above listed tasks. However, I am not sure how to "undo" the GPU overclocking since it came that way by factory setting and I have never overclocked before. Moreover, I am unsure what the "original" clock speed would be. Should I reference a GTX 650 Ti? Or a GTX 650?

 

[Arc]

Should I reference a GTX 650 Ti?

That would be a good attempt

Edit: I requested paulpicks21, our official overclocker to guide you about it.

 

[smiles86]

Thanks for the request. One more thing to note: I ran Prime95 for about 2 minutes (it did 7 tests) and my CPU reached ~90 C, at which point I stopped the test fearing overheating. Is 90 C still safe? Or was that reaching a dangerous level? I am using the stock CPU cooler that Intel provides with the CPU.

*Edit: Under regular use, the CPU sees ~30-40 C. When doing light gaming, it reaches ~50-60 C.

 

[Arc]

2 hours is a long time, so we may give the temps a benefit of doubt at this stage of troubleshooting. Still, a snapshot would be good, for documentation.

 

[smiles86]

Just in case you misread, I ran the test for only 2 minutes, NOT 2 hours. And it had reached ~90 C already.

 

[paulpicks21]

Hello smiles86 and welcome to the forums.

If your GPU came with a factory OC then that should not be the issue, it was already designed to run that way. I will give that some more thought in a minute.

But for your CPU to hit 90C in 2 MINS Indicates an issue immediately to me. Is the CPU at stock speed? I know the Haswells run hot, but if that is at stock speed that seems a little toasty to me. I would consider some better cooling for it.

 

[Arc]

That's dangerous.

Seeing Paul active here .... he will guide you the further steps.

 

[paulpicks21]

Sorry, I need to read things better, I was rushing.

So you say 50-60C During gaming use, that sounds fine.

Prime is known to push things to the limit and Haswell on stock cooling I suppose 90C is acceptable.

 

[smiles86]

As far as I am aware, the CPU did not come overclocked by factory setting (it was Intel OEM), nor have I done any sort of conscious overclocking on it. From what I understand however, the stock CPU cooler that Intel provides is "garbage", so I might invest in a better one. However, given that temperatures barely go over ~40 C during regular use such as web browsing, and only peak at ~60 C during light gaming, would CPU overheating be the cause of my BSODs?

*Edit: Ok, so should I leave Prime95 or should I keep on testing beyond 2 minutes, assuming the temperature will only keep rising?

 

[paulpicks21]

As far as I am aware, the CPU did not come overclocked by factory setting (it was Intel OEM), nor have I done any sort of conscious overclocking on it. From what I understand however, the stock CPU cooler that Intel provides is "garbage", so I might invest in a better one. However, given that temperatures barely go over ~40 C during regular use such as web browsing, and only peak at ~60 C during light gaming, would CPU overheating be the cause of my BSODs?

*Edit: Ok, so should I leave Prime95 or should I keep on testing beyond 2 minutes, assuming the temperature will only keep rising?

Based on your "regular use" temps I doubt the CPU will be causing any issues. I would not bother with Prime until if and when you get a better cooler, it is very tough on things.

 

[paulpicks21]

Have you gone into the bios and set the memory speed/timings and voltage? That can be very important. Have a look at the CPU settings in the bios too, to see the speed and voltage that is set to.

I must admit I am not very good with BSOD but in my limited knowledge if it's hardware related it's usually down to speed, timings or voltage.

 

[smiles86]

With my original memory sticks, everything in the BIOS was left at factory default. When I installed the new Corsair sticks, I did set the memory profile to X.M.P. 1600 as it looked like the memory would have run at 1333 MHz instead. I only set the profile to be enabled, and did not touch anything else.

The first three BSOD logs I posted occurred with the original memory sticks. The last BSOD log occurred with the new Corsair sticks (X.M.P. enabled).

 

[paulpicks21]

Do you have CPUz installed? It might be an idea to run that, then just run prime for a few seconds and capture a screenshot of the CPU speed and voltage under load.

 

[smiles86]

Here are the results from the steps Arc suggested.

• PSU supplies almost double the supposed required wattage.
• Nvidia drivers uninstalled and cleaned. 314.22WHQL installed.
• SSD drivers up to date.
• Memtest86 ran overnight for ~10 hours (8 complete passes) without a single error.
• Furmark ran for ~3 minutes before artifacts began appearing. Test was stopped at that point, but GPU temps were still rising very slowly.
• Prime95 ran for ~2 minutes without any errors. CPU temps reached ~90 C.

*Edit: Furmark was run after Prime95. Disregard the ~90 C max temp values for CPU, as those are from Prime95 and not from Furmark.

 

[smiles86]

The following attachments are CPUZ when running Prime95 and CPUZ during regular use, in that order.

 

[essenbe]

Smiles86, I would recommend you not run Prime95 until you get a better cooler. Your TJmax (the temp at which the computer begins to shut itself down to minimize damage) is 100C. You had 1 core at 91C after only 2 minutes. That is too close for that short of a time. Also, You should not have had artifacts after only 3 minutes of Furmark. That indicates a problem with your GPU or the GPU drivers. What version are you running and do you know how to install graphics drivers? It needs to be done right to avoid problems. Arc has a very good explanation of exactly how it should be done. Please go to Nvidia and check the driver version that is current and see if it matches what you have installed. You will want the most current WHQL driver.

 

[smiles86]

I had uninstalled the Nvidia drivers and installed the 314.22 one following the steps Arc provided. Should I instead be installing the most recent one for my card (Nvidia's site says 331.82) instead of 314.22, which Arc recommended?

I was most likely using 331.82 before since I had automatic updates enabled, but Furmark was run with 314.22. I will give the test a go again with 314.22 and let you know how it goes. Just to be sure, the artifacts I should be looking for are pixels that don't look like they fit in? I was seeing little bits of white once in a while during the test.

*Edit: Just ran Furmark again and right from the very beginning I am seeing little bits of white. I assume these are artifacts. I will hold off on updating drivers to the most recent (331.82) until otherwise advised.
*Edit2: The little bits of white flicker, but consistently appear in the same location.

 

[paulpicks21]

Here are the results from the steps Arc suggested.

• PSU supplies almost double the supposed required wattage.
• Nvidia drivers uninstalled and cleaned. 314.22WHQL installed.
• SSD drivers up to date.
• Memtest86 ran overnight for ~10 hours (8 complete passes) without a single error.
• Furmark ran for ~3 minutes before artifacts began appearing. Test was stopped at that point, but GPU temps were still rising very slowly.
• Prime95 ran for ~2 minutes without any errors. CPU temps reached ~90 C.

*Edit: Furmark was run after Prime95. Disregard the ~90 C max temp values for CPU, as those are from Prime95 and not from Furmark.

In my experience if artifacts are appearing it is down to 1 of 2 things, either not enough voltage or too much speed for the voltage.
In either case it should not be happening to you with a factory overclocked GPU. If you have not messed with it yourself then it should run just fine. You should not need to lower the clock speed or raise the voltage yourself, the whole point of a factory OC card is they take the work out of it for you and you pay extra for it.

My recommendation would be to RMA the card.