Solved Random BSODs, MEMORY_MANAGEMENT, ntoskrnl.exe

[Dave76]

Okay, I think I've made some progress:

1. System specs are updated - please let me know if anything else would be useful to you.

Can you get information on:
PSU, you can check you computer documentation or open the case and get it off of the PSU.
Case, brand of computer case, this may be helpful for cooling issues.
Cooling, CPU cooler Stock? or ? How many case fans?

2. Memtest86+: After an earlier BS (approx 1 month ago), I ran this for 49 hours straight with no errors (although as mentioned earlier in the thread, I had all RAM sticks in at once).

This should be a good indication that your RAM is working, with 24GB RAM or any RAM it can pass memtest86+ several times and then fail, it can be an intermittent issue.
We'll try a couple setting changes, if you still get crashes you may need to remove three RAM cards for testing purposes.
Let's try setting adjustments first, will put them below.

3. CPU-z: CPU, Mainboard, Memory, and SPD snips are attached

Your motherboard site lists FF as the latest BIOS version, yours is FA. No information on the BIOS improvements were listed.
Updating BIOS can be risky, so we will leave it for now but, may have to come back to it later.

Your RAM timings are set to JEDEC#2, the frequency is lower than the RAMs max frequency, which is good because with that much RAM you will usually need to lower the frequency and/or timings to get it to run stable.

4. RealTemp: I've downloaded the program but from a quick look at the docs it seems that there is a bit of a learning curve to this one. I will read through the documentation and figure out how to create idle, under load, and testing environments and then report the numbers.

Real Temp just monitors the CPU core temps, idle is when you have no apps open and you computer is not doing anything.
You will have to induce the load by running your highest resource usage program, video encoding or game, there are specialized apps to stress your system.
We may need to run one of these to test the CPU and CPU cooling abilities.
Post your idle and load (as mentioned above) temps first.

5. Voltages from BIOS:

Main Screen:
-VCore: 1.248V
-DRAM Voltage: 1.520V

Advanced Voltage Settings (all set to Standard or Auto):
-CPU VCore: 1.28125V
-QPI/Vtt Voltage: 1.175V
-CPU PLL: 1.800V
-PCIE: 1.500V
-QPI PLL: 1.100V
-IOH Core: 1.100V
-ICH I/O: 1.500V
-ICH Core: 1.100V
-DRAM Voltage: 1.500V
-DRAM Termination: 0.750V
-Ch-A Data VRef: 0.750V
-Ch-B Data VRef: 0.750V
-Ch-C Data VRef: 0.750V
-Ch-A Address VRef: 0.750V
-Ch-B Address VRef: 0.750V
-Ch-C Address VRef: 0.750V

These voltages look fairly stock except for the slight bump in QPI/Vtt voltage, this is the CPU integrated memory controller voltage.

6. Out of date drivers: These are all related to the LogMeIn service. I spoke to their tech rep and was told that I could delete lmimirr.sys (which I did). He also told me that I have the most up to date versions of the other two as they come along with the LogMeIn installation/updates and my LogMeIn version is only 3 weeks old. Any more I should do here?

You may need to disable or un-install for testing purposes, let's adjust RAM settings first.

7. All MS and MSE updates should be installed.

If a MS update fails to install it will notify you, have you seen any notifications regarding a failed update?
You should get 4-5 MSE updates a week.

Most important is to post your CPU temps, post a screen shot of Real Temp, at an idle and under a load, your most CPU/RAM intensive program.
The max recommended CPU Core temp from Intel is 73°C.
If your temps go above this, shut down any programs you have running and check if the CPU is cooling down.

Bump your DRAM voltage to 1.550v and your QPI/Vtt voltage to 1.20v.
Try to re-create your crash scenario, or just use it normally if it has been random.

Let us know the results.

 

[jonnyb]

Had another BSOD overnight - attached are the files. Ironically, before I started troubleshooting them it would be weeks or months in between incidents and now that I've tried to eliminate them it's become more like once a day...who knows, that may be a good thing.

Dave- going to work through your post now...

 

[jonnyb]

1. System specs: requested updates have been posted. Not sure how to answer the # of case fans question, hopefully the info on the cooler brand/model will cover that.

2. RAM: Will be updating the voltages as suggested immediately after posting this.

3. BIOS: nothing to do for now.

4. RealTemp: Have attached screenshots for underload and idle. Not sure if my underload scenario is sufficient - essentially it is a program that runs 8 simultaneous threads across the 4 cores 125 times, then pauses for one second, and then starts over. The RT snip for UnderLoad is taken while running that program for 2h40m. The idle shot is after closing that program and letting the computer run nothing for 20mins. Let me know if I should take a different course of action. Also, it's worth noting that I didn't calibrate anything on RealTemp - I simply doubleclicked on RealTemp.exe and let it run.

5. LogMeIn: nothing to do for now.

6. MS Update: I have looked through my Windows Update history and there are about 5 updates marked failed since late 2010. However, in each of those cases, a subsequent attempt at the failed update was successful. This may not be relevant, but I have Windows Update set to automatically download but not install updates - this is to avoid an unwanted restart when away from my machine. However, I normally will manually install them within 24 hours. Let me know if that could be an issue.

7. MSE: Appears to be up to date. I have it set to run a full scan every night at 2am. This did catch and remove about 7 or 8 potential viruses the night after my hotmail was hacked. I'm not sure why it didn't prevent them in the first place or if they were only caught because that was the first time I switched from Quick Scan to Full Scan. Now that I look more closely, it seems most of the container files that MSE identified are on my second hard drive inside Windows Backup folders; perhaps, they were old virus files. (One other note here...when I first went to post here about my BSOD problem, I switched from Avast to MSE. I removed Avast, restarted my machine, then downloaded MSE and installed it. Perhaps, those few minutes with no anti-virus installed were enough to leave me vulnerable...although I did no web surfing during that time other than to download MSE. Regardless, the right thing to do was probably to have the different antivirus programs overlap for a few minutes instead.)

 

[jonnyb]

I was not able to make the voltage changes exactly as you specified. For QPI/Vtt, the closest I could get was 1.195. For DRAM, the closest I could get was either 1.540 or 1.560 (I went with 1.540). Is this a problem?

 

[Dave76]

1. System specs: requested updates have been posted. Not sure how to answer the # of case fans question, hopefully the info on the cooler brand/model will cover that.

Looks better, that is a good CPU cooler, your idle temps seem a bit high.
What is the ambient (room) temp?

Your max temp of ~65°C is also a little high for your CPU cooler.

2. RAM: Will be updating the voltages as suggested immediately after posting this.

4. RealTemp: Have attached screenshots for underload and idle. Not sure if my underload scenario is sufficient - essentially it is a program that runs 8 simultaneous threads across the 4 cores 125 times, then pauses for one second, and then starts over. The RT snip for UnderLoad is taken while running that program for 2h40m. The idle shot is after closing that program and letting the computer run nothing for 20mins. Let me know if I should take a different course of action. Also, it's worth noting that I didn't calibrate anything on RealTemp - I simply doubleclicked on RealTemp.exe and let it run.

Just run Real Temp as it is.

D/L Prime95, this is a CPU and RAM stress test.
Be sure to open and monitor RealTemp while running this test.
Your CPUs max temp is 72.9°C, stop the test if it reaches this temp.
Intel® Core™ i7-950 Processor
It will run your CPU at 100%, depending on the test you choose will also test the RAM. The torture test is the one that will also test the RAM.

CPU - Stress Test with Prime95

Run the torture test for 1-2 hours, this will give you a good indication of the CPU temps.
For a full CPU/RAM stability test, 4-12 hours are suggested, I usually suggest the minimum depending on your usage.

This program will keep running until you stop it.

6. MS Update: I have looked through my Windows Update history and there are about 5 updates marked failed since late 2010. However, in each of those cases, a subsequent attempt at the failed update was successful. This may not be relevant, but I have Windows Update set to automatically download but not install updates - this is to avoid an unwanted restart when away from my machine. However, I normally will manually install them within 24 hours. Let me know if that could be an issue.

This is not unusual, good that they installed.
D/L but not install is how I have mine set.

7. MSE: Appears to be up to date. I have it set to run a full scan every night at 2am. This did catch and remove about 7 or 8 potential viruses the night after my hotmail was hacked. I'm not sure why it didn't prevent them in the first place or if they were only caught because that was the first time I switched from Quick Scan to Full Scan. Now that I look more closely, it seems most of the container files that MSE identified are on my second hard drive inside Windows Backup folders; perhaps, they were old virus files. (One other note here...when I first went to post here about my BSOD problem, I switched from Avast to MSE. I removed Avast, restarted my machine, then downloaded MSE and installed it. Perhaps, those few minutes with no anti-virus installed were enough to leave me vulnerable...although I did no web surfing during that time other than to download MSE. Regardless, the right thing to do was probably to have the different antivirus programs overlap for a few minutes instead.)

Changing AV programs is fine, better if you don't go on the web with no AV protection.

D/L and install the free version of Malwarebytes, be sure to update the virus definitions before running.
Malwarebytes will run along with any other AV you have installed and is a supplemental anti-malware program.
Run it at least once a week, depending on you internet activity.

I was not able to make the voltage changes exactly as you specified. For QPI/Vtt, the closest I could get was 1.195. For DRAM, the closest I could get was either 1.540 or 1.560 (I went with 1.540). Is this a problem?

These are fine for now.

 

[jonnyb]

Dave- Thanks for the continued help here. I won't be able to address these items until tomorrow afternoon, but will report back once I have.

 

[Dave76]

No problem, I'll be out until then.

 

[jonnyb]

1. Room temp is 21.1 C.

2. I ran the small FFT test with Prime95 and as you can see in the attached snip, the temp on the one core hit 73C in just over a minute. Should I go ahead with the "Blend" test to involve the RAM as well, or does this temp problem need to be addressed first?

3. I just kicked off a full scan with Malwarebytes.

 

[Dave76]

You need to re-apply the thermal paste on your CPU cooler, those temps are too high.
They should be below 60°C.

Make sure it is firmly attached.

Have a look at this: Application on HDT Coolers

 

[jonnyb]

Okay - I have been thinking of replacing my current cooler because of the noise, so I think I'll use this as an opportunity to replace it with something quieter.

The Tuniq Tower 120 Extreme seems to get decent reviews here:

FrostyTech - Best Heat Sinks & PC Cooling Reviews

However, it seems there are definitely some quieter options:

Thermaltake Acoustic Sample and Heatsink Sound Levels - Frio FrostyTech Review

I just don't know if any of those are any good. Any recommendation for a solid but quiet cooler?

 

[Dave76]

Check the clearance for the RAM slot closest to your CPU.
You might consider getting new fans for your Frio. Look for low noise, high volume and high pressure, the pressure is important for a CPU cooler.

I was looking at freeing up my first RAM slot (blocked by the Coolermaster Hyper 212+ CPU cooler, good cooler but blocked any RAM with a heatsink) so I got a Corsair H80. Great water cooler, very noisy fans when on high. Got Noctua NF P12 (quiet, high volume, high pressure - made for CPU coolers) and they are very quiet.
There are other quiet CPU cooler fans on the market now.

Any of these should work fine.
Top 5 Heatsinks

 

[jonnyb]

Interesting development... I was reading up on my Frio and determined that if the installation / thermal paste was done properly, running the fans on low (instead of the current high setting) should be able to handle the cooling while reducing the noise level from 62 dba to 37. So I decided to keep the Frio and just re-apply the thermal paste. I've never opened up a case before, so I grabbed the old Frio box to see if I could handle the instructions in the manual. When I opened the box, one of the two fans was still in there...the place (micro center) that built my machine never installed it. I'm guessing that could be a major reason why this is happening. I'm going to head there after work and have it installed (hopefully along with a thermal paste re-app).

 

[Dave76]

Two fans, in a push/pull configuration, will give you lower temps.

Try your 2-fans setup on low/med/high speed settings to check the cooling difference.
Your Gigabyte board should have a auto CPU Fan setting, this will run the fan at lower speed until more cooling is needed, then it will increase up to max.
The stock Corsair fans were actually a couple degrees cooler on medium then when on high.

For comparison.
The Noctua NF P12 are 19.8 dba on high speed.
Airflow: 92.3 m³/h
Static Pressure: 1.68 mm H2O

There are other fans with similar specs.

 

[jonnyb]

Okay, I've learned a lot about coolers in the last day or so. Turns out there was a very good reason the second fan wasn't installed on my Frio -- it didn't fit b/c of the first RAM slot (as you mentioned Dave). Microcenter still should have told me and I would have went with a different cooler that fit, but that's water under the bridge at this point.

I've reapplied the thermal paste for the Frio and ran Prime95 again (but only for an hour...need sleep); the results were much improved over the last time when temps hit 73* in 1 minute. However, I'm not sure if they're good enough? (pic attached) Dave, I know you said they should be under 60* earlier but this is with only one fan...not sure if that makes these temps acceptable or not.

If the cooling is still insufficient, I'm going to go with the liquid cooling and those quiet fans. It appears that there are very few air cooled options out there that address everything I'm looking for: 1) fit my G.Skill Ripsaw RAM (w/ heatspreaders are ~41mm tall) 2) solid temp performance 3) are quiet.

If the cooling is okay, I will proceed with the next steps...which I believe were running torture tests that involved the RAM.

 

[Dave76]

Those temps are fine.
If you want to upgrade later, they will drop a bit further but, I wouldn't rush out and do it now.

Run the torture test and let us know the results.

 

[jonnyb]

Continuing with the one thing after another theme here, I failed the torture test after 4h5m (ran the 'Blend' test this time instead of the 'Small FFT'). Attached are snips from the thread that failed and the max temps during the torture test. I'm guessing the temps are okay, does this mean a RAM issue?

 

[Dave76]

The temps are ok, further testing is needed, could be the RAM or CPU.

With your RAM, I would start there.

Take out half your RAM and run the same Prime95 Blend test.

This should let us know if the 24GB of RAM is putting stress on the CPU IMC (Integrated Memory Controller).
If it passes we can put the 12GB back in (24GB total) and adjust some of the settings.

 

[jonnyb]

Thanks Dave, I will do that tonight...two quick questions on it:

Should I change the BIOS voltages that we bumped up back to 'Auto' for this test?

Also, when I pull the RAM sticks out is there anything else that I need to do or is it as simple as pulling them out and restarting (and the change will be automatically detected)?

 

[Dave76]

Leave the BIOS RAM voltage settings where they are, this will help the RAM stability and tell us if it needs adjusting.

Just take out the RAM, Leave the remaining RAM in slots 1,3 and 5.
(Counting slots from your CPU it will be slot 2,4, 6)
See you motherboard manual, page 16.
GA-X58A-UD3R (rev. 2.0)

 

[jonnyb]

Gotcha, thanks. Will report back once testing is complete.