F5F Stay Refreshed Power Users Overclocking BSODs connected to computations - not due to overclocking

BSODs connected to computations - not due to overclocking

BSODs connected to computations - not due to overclocking

L
LadyErinicorn
Member
150
06-17-2016, 08:35 AM
#1
Hello everyone,
I'm dealing with a BSOD issue that appears occasionally, likely related to Trimble SketchUp. Since it's rare, I initially thought it might be a software problem. But when I tried using 3DFlow's Zephyr, the BSOD kept happening every time. The developer explained that the software processes 3D objects from still images and requires intense calculations over long periods, which could stress the CPU, GPU, or RAM.

Prime95 showed no issues with the small FFT setting, but other tests—especially those using FFTs at 1024—triggered an instant BSOD. Memtest86+ detected two errors on each run, both pointing to the same memory addresses (Test 7). No other problems were found.

I’m not very experienced with these diagnostics, so I can’t fully interpret the results. On one hand, if RAM were faulty, why would I get BSODs only when running calculation-heavy programs? On the other hand, if it’s the calculations that cause the problem, why aren’t CPU and GPU tests showing issues, just RAM?

I suspect Memtest might not pinpoint the exact cause, but since it found two addresses repeatedly across multiple runs, it seems unlikely the CPU caches are the issue. It’s possible the RAM is the culprit.

Should I try swapping my memory modules, rerun Memtest, and check for new error locations? Also, I’m not overclocking—just using the standard configuration from DigitalStorm (Windows 8).

[UPDATE]
Memtest on a single module shows no problems. Prime95 results are stable when testing one memory at a time, though BSODs still occur after some minutes. This suggests the memory might be fine, but running it with the full 16GB could trigger the issue.

[UPDATE II]
Even with Prime95 using the SMALL FFT setting, temperatures reach 100°C and the CPU throttles aggressively to prevent a crash. This hints that the cooling system may not be sufficient. I’m concerned about overheating and wonder how to verify coolant levels in these water-cooled systems.

Thank you for helping me understand what’s happening.
L
LadyErinicorn
06-17-2016, 08:35 AM #1

Hello everyone,
I'm dealing with a BSOD issue that appears occasionally, likely related to Trimble SketchUp. Since it's rare, I initially thought it might be a software problem. But when I tried using 3DFlow's Zephyr, the BSOD kept happening every time. The developer explained that the software processes 3D objects from still images and requires intense calculations over long periods, which could stress the CPU, GPU, or RAM.

Prime95 showed no issues with the small FFT setting, but other tests—especially those using FFTs at 1024—triggered an instant BSOD. Memtest86+ detected two errors on each run, both pointing to the same memory addresses (Test 7). No other problems were found.

I’m not very experienced with these diagnostics, so I can’t fully interpret the results. On one hand, if RAM were faulty, why would I get BSODs only when running calculation-heavy programs? On the other hand, if it’s the calculations that cause the problem, why aren’t CPU and GPU tests showing issues, just RAM?

I suspect Memtest might not pinpoint the exact cause, but since it found two addresses repeatedly across multiple runs, it seems unlikely the CPU caches are the issue. It’s possible the RAM is the culprit.

Should I try swapping my memory modules, rerun Memtest, and check for new error locations? Also, I’m not overclocking—just using the standard configuration from DigitalStorm (Windows 8).

[UPDATE]
Memtest on a single module shows no problems. Prime95 results are stable when testing one memory at a time, though BSODs still occur after some minutes. This suggests the memory might be fine, but running it with the full 16GB could trigger the issue.

[UPDATE II]
Even with Prime95 using the SMALL FFT setting, temperatures reach 100°C and the CPU throttles aggressively to prevent a crash. This hints that the cooling system may not be sufficient. I’m concerned about overheating and wonder how to verify coolant levels in these water-cooled systems.

Thank you for helping me understand what’s happening.

I
Inezze009
Senior Member
716
06-18-2016, 10:27 PM
#2
use memtest86 to test each stick individually to determine if it's a RAM stick or the memory controller. also swap the RAM and check if the other two slots work.
I
Inezze009
06-18-2016, 10:27 PM #2

use memtest86 to test each stick individually to determine if it's a RAM stick or the memory controller. also swap the RAM and check if the other two slots work.

G
GoMigs
Senior Member
614
06-20-2016, 01:00 AM
#3
use memtest86 to test each stick individually to determine if it's a RAM stick or the memory controller. also swap the RAM and observe if the other two slots work correctly.
G
GoMigs
06-20-2016, 01:00 AM #3

use memtest86 to test each stick individually to determine if it's a RAM stick or the memory controller. also swap the RAM and observe if the other two slots work correctly.

I
IAmLiam
Member
193
06-22-2016, 12:37 AM
#4
The smorizio is perfectly accurate. The BSOD will appear only when that specific address is accessed. The issue is probably with one of your RAM modules, though it would be wise to test the other stick in each slot as well. Memtest86 has consistently shown reliable results, and I usually run just one stick at a time to identify the faulty one. A single fault on each stick is an unlikely but possible scenario—just very unlucky.
I
IAmLiam
06-22-2016, 12:37 AM #4

The smorizio is perfectly accurate. The BSOD will appear only when that specific address is accessed. The issue is probably with one of your RAM modules, though it would be wise to test the other stick in each slot as well. Memtest86 has consistently shown reliable results, and I usually run just one stick at a time to identify the faulty one. A single fault on each stick is an unlikely but possible scenario—just very unlucky.

G
GamerTV1
Member
116
06-22-2016, 01:17 AM
#5
people should emil memtest people ask them during their tests when there is a failure it indicates in plan english which slots have failed. for us users who can read hex and the screen can display what ram failed as on the screen. (10)(20(30)40) 10-slot 1 20 slot 2. with today's ram that have eproms containing speed and serial info memtest should be reading that information and stating dim #ser number has error in location x.
G
GamerTV1
06-22-2016, 01:17 AM #5

people should emil memtest people ask them during their tests when there is a failure it indicates in plan english which slots have failed. for us users who can read hex and the screen can display what ram failed as on the screen. (10)(20(30)40) 10-slot 1 20 slot 2. with today's ram that have eproms containing speed and serial info memtest should be reading that information and stating dim #ser number has error in location x.

J
jettli3
Junior Member
44
06-22-2016, 05:58 AM
#6
Hey all.... SO, a bit of an update that seems to suggest that things are not so simple as they might first seem.
Memtest of one memory module at a time: No issues!
Here is where it gets more interesting... running Prime95 when only one mem module is connected (8GB)... and I do NOT get an instant BSOD! If I let it run for a few minutes... and do some browsing for example... then yes, it goes BSOD like before.
I usually get an "uncorrectable error" message, but twice now I have seen a "clock watchdog" error.
This is really odd.
I wonder if voltage is sagging a bit to the memory modules... and running two at a time makes it all the worse... 2 modules = instant BSOD... one module = eventual BSOD (with heavy loading).
Rather odd.
Again, I remind the kind readership that I am not in fact overclocking (nor have I ever attempted it). This is a stock machine.
Any new ideas out there given these results?
Thanks for your kind help!
J
jettli3
06-22-2016, 05:58 AM #6

Hey all.... SO, a bit of an update that seems to suggest that things are not so simple as they might first seem.
Memtest of one memory module at a time: No issues!
Here is where it gets more interesting... running Prime95 when only one mem module is connected (8GB)... and I do NOT get an instant BSOD! If I let it run for a few minutes... and do some browsing for example... then yes, it goes BSOD like before.
I usually get an "uncorrectable error" message, but twice now I have seen a "clock watchdog" error.
This is really odd.
I wonder if voltage is sagging a bit to the memory modules... and running two at a time makes it all the worse... 2 modules = instant BSOD... one module = eventual BSOD (with heavy loading).
Rather odd.
Again, I remind the kind readership that I am not in fact overclocking (nor have I ever attempted it). This is a stock machine.
Any new ideas out there given these results?
Thanks for your kind help!

D
DoodleDad
Member
131
06-28-2016, 01:47 PM
#7
I might start adjusting memory timings in BIOS. Depending on your BIOS, there could be many settings or none at all. For a quick test, you could try installing new RAM to see if it resolves the issue, if not, the problem might lie elsewhere.
D
DoodleDad
06-28-2016, 01:47 PM #7

I might start adjusting memory timings in BIOS. Depending on your BIOS, there could be many settings or none at all. For a quick test, you could try installing new RAM to see if it resolves the issue, if not, the problem might lie elsewhere.

T
TheKlesty_YT
Junior Member
18
06-28-2016, 05:29 PM
#8
Another update has been made...
I purchased a new Corsair liquid cooling system to upgrade from the standard DigitalStorm setup. The results are noticeable—I can now execute Prime95 for roughly 5 to 6 seconds before encountering a BSOD!
Do my expectations match what I hoped? Should I be able to run Prime95 on a typical machine without any problems? It looks like it should work, so thank you for your advice.
Appreciate your support.
T
TheKlesty_YT
06-28-2016, 05:29 PM #8

Another update has been made...
I purchased a new Corsair liquid cooling system to upgrade from the standard DigitalStorm setup. The results are noticeable—I can now execute Prime95 for roughly 5 to 6 seconds before encountering a BSOD!
Do my expectations match what I hoped? Should I be able to run Prime95 on a typical machine without any problems? It looks like it should work, so thank you for your advice.
Appreciate your support.