DynDOLOD 3.00 Alpha 182

[sheson]

2 hours ago, m1sf1t711 said:

Thank you for the suggestion. I'm not overclocking my PC. I'm not really that technically savvy but I am very happy to follow instructions and suggestions and learn whatever I can as I go along.

If you use the PC as delivered, it wouldn't be the first time that the factory or the default BIOS settings are the cause of problems.

I suggest to google the make of the PC or mainboard and learn about its BIOS settings, especially how to change/select different memory XMP/EXPO settings.
I would also check if some kind of auto overclock is enabled, CPU temps, etc. This obviously requires some experience, so best to find a related hardware forum that disucsses your specific hardware - typically based on the mainboard.

[sheson]

49 minutes ago, z929669 said:

Here's my logs ... no bugreport, no debug, etc.

Situation:

Running DynDOLOD Alpha 182 with the following settings on the Step build (standard settings for me, run hundreds of times on this and similar builds):

DynDOLOD_SSE_Default.ini 105.76 kB · 1 download

The problem occurs during occlusion processing. I forget which worldspace (might not matter). My PC simply shuts down, indicating that the CPU becomes too hot or the system is otherwise critical.  use water cooling and no overclocking. My system specs are in my sig. Due to the automatic shutdown, logging is cut off. Events are not able to print until reboot, and then only a general kernel shutdown error.

This is my second such crash, so that's why I'm reporting it.

Starting now, I have enabled system minidumps and set the following in DynDOLOD_SSE.ini:

Debug=1
realtimelog=1
GLListDebug=1
GLDebug=1

I will repeat the process as the previous 2x, and report back with additional info/logs.

For what it's worth, occlusion is always taxing on my CPU, but never to the point of a reproducible protective system shutdown. I could lower my occlusion quality setting, but I would rather troubleshoot the problem if possible, since my current config is historically reliable on my PC and mod list.

The PC shutting down, BSOD is a hardware, BIOS or OS problem.
I suggest to stress test the PC with designated tools like prime95/OCCT and query an appropriate hardware forum.

Occlusion generation uses the CPU and not the GPU/OpenGL.
If you want to "troubleshoot" by reducing the CPU workload, use a lower quality setting or change process priority/core affinity with task manager or 3rd party tool.
If using lots of memory is the problem see https://dyndolod.info/Help/Occlusion-Data#Out-of-Memory-while-Generating-Occlusion-Data

[m1sf1t711]

27 minutes ago, sheson said:

If you use the PC as delivered, it wouldn't be the first time that the factory or the default BIOS settings are the cause of problems.

I suggest to google the make of the PC or mainboard and learn about its BIOS settings, especially how to change/select different memory XMP/EXPO settings.
I would also check if some kind of auto overclock is enabled, CPU temps, etc. This obviously requires some experience, so best to find a related hardware forum that disucsses your specific hardware - typically based on the mainboard.

Ok thank you - that's all a little daunting but I'll see what I can do.  I appreciate all the time you've been putting into helping me, thank you!

[z929669]

9 minutes ago, sheson said:

The PC shutting down, BSOD is a hardware, BIOS or OS problem.
I suggest to stress test the PC with designated tools like prime95/OCCT and query an appropriate hardware forum.

Occlusion generation uses the CPU and not the GPU/OpenGL.
If you want to "troubleshoot" by reducing the CPU workload, use a lower quality setting or change process priority/core affinity with task manager or 3rd party tool.
If using lots of memory is the problem see https://dyndolod.info/Help/Occlusion-Data#Out-of-Memory-while-Generating-Occlusion-Data

My hardware is unlikely the issue, and stress testing would be redundant, as I have done all of that in the recent past with no hardware changes since. As mentioned, I've generated hundreds of times with the settings I provided. Memory is also not the issue. I rarely see more than 32 Gb of my 64 Gb committed when running the process.

My system did not BSOD. It simply shut down, so the problem was most likely with the CPU or system bus overheating past the threshold I have set. I will make it a point to monitor my temps though to verify. I may need to relax the threshold I have set once I know what's happening. It's fairly conservative at 85 C.

I will have more evidence soon with the realtime log and crashdump. Perhaps it will not crash, since the RT logging slows things down quite a bit.

[m1sf1t711]

50 minutes ago, sheson said:

If you use the PC as delivered, it wouldn't be the first time that the factory or the default BIOS settings are the cause of problems.

I suggest to google the make of the PC or mainboard and learn about its BIOS settings, especially how to change/select different memory XMP/EXPO settings.
I would also check if some kind of auto overclock is enabled, CPU temps, etc. This obviously requires some experience, so best to find a related hardware forum that disucsses your specific hardware - typically based on the mainboard.

Hi 

I'm not sure that my PC is the problem as I am not getting any crashes just a failure to generate some of the lod, usually Tamriel Aut and Tamriel Sum but sometimes the other seasons when trying the test versions you've shared.  I also haven't changed anything with my hardware since I was able to successfully run DynDOLOD Alpha 182 when it was first publised. However I will still try to look into this as I appreciate something might have been changed by something in the background that I'm unaware of, thank you.

[sheson]

11 minutes ago, m1sf1t711 said:

Hi 

I'm not sure that my PC is the problem as I am not getting any crashes just a failure to generate some of the lod, usually Tamriel Aut and Tamriel Sum but sometimes the other seasons when trying the test versions you've shared.  I also haven't changed anything with my hardware since I was able to successfully run DynDOLOD Alpha 182 when it was first publised. However I will still try to look into this as I appreciate something might have been changed by something in the background that I'm unaware of, thank you.

There might be changes to the load order that increased CPU, memory demand to higher levels than before and you were just lucky so far.

Other reasons might be OS update, drivers, loose cables, BIOS settings resetting to default, hardware going bad, cooling being affected because of dust (dead mouse not unheard of) etc. It is possible the issue is storage/disk related.

[m1sf1t711]

Just now, sheson said:

There might be changes to the load order that increased CPU, memory demand to higher levels than before and were just lucky.

Other reasons might be OS update, drivers, loose cables, BIOS settings resetting to default, hardware going bad etc. It is possible the issue is storage/disk related.

I understand thank you

I'll try to find out what I can.

[z929669]

57 minutes ago, sheson said:

The PC shutting down, BSOD is a hardware, BIOS or OS problem.
I suggest to stress test the PC with designated tools like prime95/OCCT and query an appropriate hardware forum.

Occlusion generation uses the CPU and not the GPU/OpenGL.
If you want to "troubleshoot" by reducing the CPU workload, use a lower quality setting or change process priority/core affinity with task manager or 3rd party tool.
If using lots of memory is the problem see https://dyndolod.info/Help/Occlusion-Data#Out-of-Memory-while-Generating-Occlusion-Data

As I suspected, my PC is shutting down due to CPU temp spike exceeding the now-higher threshold and fan speeds I've set. As mentioned previously, I use water cooling, and my PC is pretty optimized. I seriously doubt that my hardware and dust buildup have changed significantly in the past 30-60 days, but I will perform my routine PC fan/dust maintenance and do some performance evaluations with Prime95.

Rather, I suspect the occlusion routine is more aggressive or demanding than in previous Alphas for whatever reason. Based on some of the other recent posts for Alpha 182, there seems to be increased hardware demand with Alpha 182.

No crashdump, but here's my logs, including realtime debug for the record.

At this time, I will test with setting occlusion to '1' and disable the realtime/debug logging, since we have a plausible explanation for my recent crashes (overheating of CPU, specifically during occlusion gen of Sovngarde using quality level '3')

[sheson]

33 minutes ago, z929669 said:

As I suspected, my PC is shutting down due to CPU temp spike exceeding the now-higher threshold and fan speeds I've set. As mentioned previously, I use water cooling, and my PC is pretty optimized. I seriously doubt that my hardware and dust buildup have changed significantly in the past 30-60 days, but I will perform my routine PC fan/dust maintenance and do some performance evaluations with Prime95.

Rather, I suspect the occlusion routine is more aggressive or demanding than in previous Alphas for whatever reason. Based on some of the other recent posts for Alpha 182, there seems to be increased hardware demand with Alpha 182.

No crashdump, but here's my logs, including realtime debug for the record.

At this time, I will test with setting occlusion to '1' and disable the realtime/debug logging, since we have a plausible explanation for my recent crashes (overheating of CPU, specifically during occlusion gen of Sovngarde using quality level '3')

The actual occlusion calculation code did not change in a long time.
The last notable change was with Alpha-129 in regards to the preparation part importing the heightmap ..\meshes\Terrain\[worldspace]\Objects\[worldspace].objects.heightmap.dat created by LODGen when generating object LOD instead of having to read all BTO. I suggest to restore the default OcclusionMaxThreadsObjectLOD=4 for whatever BTO it might still have to load.

If the CPU actually overheats, then there is a problem with cooling.

[m1sf1t711]

3 hours ago, sheson said:

There might be changes to the load order that increased CPU, memory demand to higher levels than before and you were just lucky so far.

Other reasons might be OS update, drivers, loose cables, BIOS settings resetting to default, hardware going bad, cooling being affected because of dust (dead mouse not unheard of) etc. It is possible the issue is storage/disk related.

Hi

Just in case this is of any value to know, I have just tried running the original Alpha 182 again having tried reducing my mod list a little incase it made any difference (I will be looking at the CPU testing tomorrow when I am less tired and can apply a fresher brain). I opened up task manager while it was running and at no point did CPU use go above 59% or memory above 66%. I was also checking to see if I had anything else running in the background at the same time which might have been interfering but there was nothing making active use of the CPU or Memory.

 

If I get any progress out of checking all the possible reasons on your list tomorrow I'll provide an update, just in case anything I find out proves helpful in any way.

 

Thanks again

[MaverickMartyn]

For what it's worth, I've had issues generating LODs too, although that was using xLODGen. Also Access Violation errors.
Stress testing using OCCT revealed instabilities in my system, and after resetting my BIOS to factory defaults I was able to generate terrain LODs.
I've now just encountered yet another Access Violation error, this time with DynDOLOD. I doubt it's a stability issue this time, since the OCCT test passed after the BIOS reset.

It's late, so I won't be investigating any further tonight, but I'll return with any updates. 

Bugreport: https://pastebin.com/NZgaTETW
Relevant bits of the pop-up dialogue:

[Main Instruction]
Access violation at address 0000000001ED0CAD in module 'DynDOLODx64.exe' (offset EF0CAD).

[Content]
Read of address 0000000000000000.

Edited October 27 by MaverickMartyn
Slight formatting fix.

[z929669]

8 hours ago, sheson said:

The actual occlusion calculation code did not change in a long time.
The last notable change was with Alpha-129 in regards to the preparation part importing the heightmap ..\meshes\Terrain\[worldspace]\Objects\[worldspace].objects.heightmap.dat created by LODGen when generating object LOD instead of having to read all BTO. I suggest to restore the default OcclusionMaxThreadsObjectLOD=4 for whatever BTO it might still have to load.

If the CPU actually overheats, then there is a problem with cooling.

Just following up: It turns out it actually was a hardware issue. Most likely CPU voltage related to AMD TPU setting I had neglected to disable over a year ago when I thought I had disabled all of the built-in overclocking. The only explanation as to why it became a bottleneck in my latest run is degraded performance over time as a result of stress on my hardware, due to the pervasive overclocking features of my motherboard.

Anyway, restoring optimized defaults corrected the problem.

[sheson]

3 hours ago, MaverickMartyn said:

For what it's worth, I've had issues generating LODs too, although that was using xLODGen. Also Access Violation errors.
Stress testing using OCCT revealed instabilities in my system, and after resetting my BIOS to factory defaults I was able to generate terrain LODs.
I've now just encountered yet another Access Violation error, this time with DynDOLOD. I doubt it's a stability issue this time, since the OCCT test passed after the BIOS reset.

It's late, so I won't be investigating any further tonight, but I'll return with any updates. 

Bugreport: https://pastebin.com/NZgaTETW
Relevant bits of the pop-up dialogue:

[Main Instruction]
Access violation at address 0000000001ED0CAD in module 'DynDOLODx64.exe' (offset EF0CAD).

[Content]
Read of address 0000000000000000.

Read the first post and/or https://dyndolod.info/Official-DynDOLOD-Support-Forum#Post-Logs which DynDOLOD log and debug log to also upload when making posts.

[MaverickMartyn]

5 hours ago, sheson said:

Read the first post and/or https://dyndolod.info/Official-DynDOLOD-Support-Forum#Post-Logs which DynDOLOD log and debug log to also upload when making posts.

Just finished running it again, with the same result. Same memory address and offset as before.
Here are all the logs. https://ufile.io/alah81mb

 

Slightly off-topic: How the fudge do you attach a file to a post here?
I see others have done so, but I can't see any option for it.

[sheson]

1 hour ago, MaverickMartyn said:

Just finished running it again, with the same result. Same memory address and offset as before.
Here are all the logs. https://ufile.io/alah81mb

Slightly off-topic: How the fudge do you attach a file to a post here?
I see others have done so, but I can't see any option for it.

Restore CreateMod=0. It is for mod authors and not for users to generate LOD for load orders full of mods (that also have lots of errors and warnings) and seasons etc.

Upload new bugreport.txt and logs if issue persists.