2.9x - Crash while baking particle and/or smoke simulations #86053

Closed
opened 2021-02-27 16:38:36 +01:00 by Marc · 36 comments

System Information
Operating system: Windows 10
Graphics card: Nvidia 1060

Blender Version
Broken: 2.92 (but I've been getting this since 2.91)
Worked: 2.90

I seem to get random crashes while baking particle or fluid simulations. in previous versions of 2.91 I would only get the crash with particles. Since installing 2.92 it started happening with smoke as well.

This happens with Optix and Cuda with either CPU or GPU. There is no crash log. The app just vanishes.

Exact steps for others to reproduce the error
It seems to work fine for a while, but then something in my modifications triggers this crash and then it happens pretty frequently. In my current scene it seems to happen when I set my domain resolution to 128 or higher.Smoke_Whispy_002.blend

**System Information** Operating system: Windows 10 Graphics card: Nvidia 1060 **Blender Version** Broken: 2.92 (but I've been getting this since 2.91) Worked: 2.90 I seem to get random crashes while baking particle or fluid simulations. in previous versions of 2.91 I would only get the crash with particles. Since installing 2.92 it started happening with smoke as well. This happens with Optix and Cuda with either CPU or GPU. There is no crash log. The app just vanishes. **Exact steps for others to reproduce the error** It seems to work fine for a while, but then something in my modifications triggers this crash and then it happens pretty frequently. In my current scene it seems to happen when I set my domain resolution to 128 or higher.[Smoke_Whispy_002.blend](https://archive.blender.org/developer/F9847387/Smoke_Whispy_002.blend)
Author

Added subscriber: @Polyflogger

Added subscriber: @Polyflogger
Author

This comment was removed by @Polyflogger

*This comment was removed by @Polyflogger*
Author

im not sure, but I think error message might help somewhat -

find_node_operation: Failed for (RIGIDBODY_REBUILD, '')
add_relation(Animation -> Rigid Body) - Failed, but op_from (ComponentKey(OBForce, ANIMATION)) was ok
add_relation(Animation -> Rigid Body) - Could not find op_to (OperationKey(type: TRANSFORM, component name: '', operation code: RIGIDBODY_REBUILD))

I don't know where this error comes from, as I never added a rigid body to this scene, nor do I currently have one (that I can find). I've tried with other scenes, that don't have any cmd line errors, and was able to bake simulations the whole way through.

im not sure, but I think error message might help somewhat - > find_node_operation: Failed for (RIGIDBODY_REBUILD, '') > add_relation(Animation -> Rigid Body) - Failed, but op_from (ComponentKey(OBForce, ANIMATION)) was ok > add_relation(Animation -> Rigid Body) - Could not find op_to (OperationKey(type: TRANSFORM, component name: '', operation code: RIGIDBODY_REBUILD)) I don't know where this error comes from, as I never added a rigid body to this scene, nor do I currently have one (that I can find). I've tried with other scenes, that don't have any cmd line errors, and was able to bake simulations the whole way through.
Author

ok, so a little more progress. This error seems to come specifically from my Force type "Force", as soon as I add an animation key to the strength.

ok, so a little more progress. This error seems to come specifically from my Force type "Force", as soon as I add an animation key to the strength.
Author

i removed any animation the "force" force and im not getting anymore crashes. so baking + animating the strength on a force type "force" is bugged

i removed any animation the "force" force and im not getting anymore crashes. so baking + animating the strength on a force type "force" is bugged
Author

false alarm. it's till crashing. I've attached an updated version of the file that at least not longer contains the previous error.

Smoke_Ink_002.blend

false alarm. it's till crashing. I've attached an updated version of the file that at least not longer contains the previous error. [Smoke_Ink_002.blend](https://archive.blender.org/developer/F9853248/Smoke_Ink_002.blend)

Added subscriber: @kubagrodzki

Added subscriber: @kubagrodzki
Author

hey gang. So I've kept plugging at this and now I'm getting crashes in other scenes as well. I can't tell what's causing it, but it's definitely baking data. if i make a new scene, it starts out stable. the more I work in it, the more frequent the crashes become. what's worse is that the crashes are not typical. Blender just simply quits.

I'm attaching another blender file the same blender crash.Fire_003.blend

Some things to note, that maybe could help - my pc is an AMD pc. My blender is not installed on the C drive. It's on another drive in my pc.

hey gang. So I've kept plugging at this and now I'm getting crashes in other scenes as well. I can't tell what's causing it, but it's definitely baking data. if i make a new scene, it starts out stable. the more I work in it, the more frequent the crashes become. what's worse is that the crashes are not typical. Blender just simply quits. I'm attaching another blender file the same blender crash.[Fire_003.blend](https://archive.blender.org/developer/F9862773/Fire_003.blend) Some things to note, that maybe could help - my pc is an AMD pc. My blender is not installed on the C drive. It's on another drive in my pc.
Author

and another piece of info - i was finally able to complete a simulation by sitting there and interrupting it every ~20 frames. it seems that if you leave it go on it's own for a while it's unhappy. maybe i can write a python script to do this until this blender bug is solved.

ps - I tried the same bake on another much more powerful pc and had the same issue.

and another piece of info - i was finally able to complete a simulation by sitting there and interrupting it every ~20 frames. it seems that if you leave it go on it's own for a while it's unhappy. maybe i can write a python script to do this until this blender bug is solved. ps - I tried the same bake on another much more powerful pc and had the same issue.

Added subscriber: @god1ged

Added subscriber: @god1ged

In #86053#1123696, @Polyflogger wrote:
and another piece of info - i was finally able to complete a simulation by sitting there and interrupting it every ~20 frames. it seems that if you leave it go on it's own for a while it's unhappy. maybe i can write a python script to do this until this blender bug is solved.

ps - I tried the same bake on another much more powerful pc and had the same issue.

I'm in the same case since 2.92, just so you know you are not alone

> In #86053#1123696, @Polyflogger wrote: > and another piece of info - i was finally able to complete a simulation by sitting there and interrupting it every ~20 frames. it seems that if you leave it go on it's own for a while it's unhappy. maybe i can write a python script to do this until this blender bug is solved. > > ps - I tried the same bake on another much more powerful pc and had the same issue. I'm in the same case since 2.92, just so you know you are not alone

I think I'm experiencing the same issue. Blender 2.92 crashes during noise bake.

I think I'm experiencing the same issue. Blender 2.92 crashes during noise bake.

Added subscriber: @sebbas

Added subscriber: @sebbas

@Polyflogger I think you should assign this task to @sebbas

@Polyflogger I think you should assign this task to @sebbas

In #86053#1123899, @kubagrodzki wrote:
@Polyflogger I think you should assign this task to @sebbas

Couldn't you assign it by yourself? I think the author is inactive :)

> In #86053#1123899, @kubagrodzki wrote: > @Polyflogger I think you should assign this task to @sebbas Couldn't you assign it by yourself? I think the author is inactive :)
Sebastián Barschkis was assigned by Marc 2021-03-04 15:11:27 +01:00
Author

In #86053#1123955, @god1ged wrote:

In #86053#1123899, @kubagrodzki wrote:
@Polyflogger I think you should assign this task to @sebbas

Couldn't you assign it by yourself? I think the author is inactive :)

i assume that's sarcasm because of the quantity of posts. sorry, but im trying to be as thorough as possible to help any way i can. More so because I need this for a project, and not being able to bake my sims is a pretty major problem.

Also - i was not aware that we're the ones who are supposed to assign our tasks directly to a dev. Last time I reported a bug it was triaged and assigned. Considering I have no clue who the actual devs are, that didn't come across as how this system works. :) But I will do this. thank you.

@sebbas As suggested, I've assigned this to you. Thank you.

> In #86053#1123955, @god1ged wrote: >> In #86053#1123899, @kubagrodzki wrote: >> @Polyflogger I think you should assign this task to @sebbas > > Couldn't you assign it by yourself? I think the author is inactive :) i assume that's sarcasm because of the quantity of posts. sorry, but im trying to be as thorough as possible to help any way i can. More so because I need this for a project, and not being able to bake my sims is a pretty major problem. Also - i was not aware that we're the ones who are supposed to assign our tasks directly to a dev. Last time I reported a bug it was triaged and assigned. Considering I have no clue who the actual devs are, that didn't come across as how this system works. :) But I will do this. thank you. @sebbas As suggested, I've assigned this to you. Thank you.

@Polyflogger @god1ged No it's not require to assign task to developer. I just follow him on Twitter and I know that he is responsible for Mantaflow in Bender.

Yesterday, I had very rough ride with 2.92. I tried to bake my simulation and it was crashing almost everytime, but finally it worked. I really don't know why. It looks like it has some problems with larger domains and longer bakes. It crashes mostly at the end (70-80%).

@Polyflogger @god1ged No it's not require to assign task to developer. I just follow him on Twitter and I know that he is responsible for Mantaflow in Bender. Yesterday, I had very rough ride with 2.92. I tried to bake my simulation and it was crashing almost everytime, but finally it worked. I really don't know why. It looks like it has some problems with larger domains and longer bakes. It crashes mostly at the end (70-80%).

@sebbas @Polyflogger
Attached is my project. To reproduce bug all you have to do is bake simulation first, and then noise. It should crash on this second step, but sometimes it happens earlier. I tested it with 2.92 and 2.91.2, the result is the same and it's not associated with the domain resolution (which was my first guess). I tested it with 128 res divisions and up to 2048 - it always crashes during noise bake. My workstation is equipped with Ryzen 3900XT, 64 GB of RAM and RTX 2080 Ti.

Smoke Bake Crash.blend

@sebbas @Polyflogger Attached is my project. To reproduce bug all you have to do is bake simulation first, and then noise. It should crash on this second step, but sometimes it happens earlier. I tested it with 2.92 and 2.91.2, the result is the same and it's not associated with the domain resolution (which was my first guess). I tested it with 128 res divisions and up to 2048 - it always crashes during noise bake. My workstation is equipped with Ryzen 3900XT, 64 GB of RAM and RTX 2080 Ti. [Smoke Bake Crash.blend](https://archive.blender.org/developer/F9881642/Smoke_Bake_Crash.blend)

Added subscriber: @Xomic

Added subscriber: @Xomic

I wanted to add my own crash report to this. I am able to reproduce it with minimalistic requirements:

  1. open blender
  2. select default cube and create 'quick smoke simulation'
  3. change domain divisions to 96
  4. check noise (default settings)
  5. change cache settings from 'replay' to 'final' and bake.
    It typically crashes at some point during the bake, most frequently around frame 100 (of a 1 to 250 frame animation bake)

At 32 resolution, Blender will not crash on noise bake, nor at 64. However, I can induce a crash at 64 resolutions if I follow the above steps and turn dissolve (default settings) on as well.

I can also induce a crash in blender while baking if I bake a smoke simulation at 96 resolution with dissolve turned on (noise disabled). I'm fairly certain I've also experienced crashes at 64 resolution with dissolve + Vorticity (usually a low value like 0.1), however, I can't get the crash to occur as consistently as the above.

One other note: if you try to import the baked volume data via a volume->import OpenVDB, the simulation will play back normally until it gets to the last frame baked, which causes blender to crash as well (which maybe expected all things considered but I thought I'd mention it even so).

System info:
CPU: Ryzen 9 3900X
RAM: 32 gb
GPU: Nvidia GeForce GTX 1050 ti

I wanted to add my own crash report to this. I am able to reproduce it with minimalistic requirements: 1. open blender 2. select default cube and create 'quick smoke simulation' 3. change domain divisions to 96 4. check noise (default settings) 5. change cache settings from 'replay' to 'final' and bake. It typically crashes at some point during the bake, most frequently around frame 100 (of a 1 to 250 frame animation bake) At 32 resolution, Blender will not crash on noise bake, nor at 64. However, I can induce a crash at 64 resolutions if I follow the above steps and turn dissolve (default settings) on as well. I can also induce a crash in blender while baking if I bake a smoke simulation at 96 resolution with dissolve turned on (noise disabled). I'm fairly certain I've also experienced crashes at 64 resolution with dissolve + Vorticity (usually a low value like 0.1), however, I can't get the crash to occur as consistently as the above. One other note: if you try to import the baked volume data via a volume->import OpenVDB, the simulation will play back normally until it gets to the last frame baked, which causes blender to crash as well (which maybe expected all things considered but I thought I'd mention it even so). System info: CPU: Ryzen 9 3900X RAM: 32 gb GPU: Nvidia GeForce GTX 1050 ti

Changed status from 'Needs Triage' to: 'Confirmed'

Changed status from 'Needs Triage' to: 'Confirmed'

Hi together! It looks like this problem comes from the OpenVDB compression.
Until this is fixed, can you try using a compression method that is not "Blosc" (Cache -> Advanced -> Compression Volumes)?

Hi together! It looks like this problem comes from the OpenVDB compression. Until this is fixed, can you try using a compression method that is not "Blosc" (Cache -> Advanced -> Compression Volumes)?

In #86053#1132006, @sebbas wrote:
Hi together! It looks like this problem comes from the OpenVDB compression.
Until this is fixed, can you try using a compression method that is not "Blosc" (Cache -> Advanced -> Compression Volumes)?

Thank you @sebbas! It seems to work with "ZIP" compression.

> In #86053#1132006, @sebbas wrote: > Hi together! It looks like this problem comes from the OpenVDB compression. > Until this is fixed, can you try using a compression method that is not "Blosc" (Cache -> Advanced -> Compression Volumes)? Thank you @sebbas! It seems to work with "ZIP" compression.
Author

In #86053#1132006, @sebbas wrote:
Hi together! It looks like this problem comes from the OpenVDB compression.
Until this is fixed, can you try using a compression method that is not "Blosc" (Cache -> Advanced -> Compression Volumes)?

Hi @sebbas, thank you for that tip. this is working on my end as well!

I'm just curious. When you say "when it's fixed", is this a blender dev bug, or do we need to wait until OpenVDB is updated globally? Also - does the affect anything at all for our workflows?

> In #86053#1132006, @sebbas wrote: > Hi together! It looks like this problem comes from the OpenVDB compression. > Until this is fixed, can you try using a compression method that is not "Blosc" (Cache -> Advanced -> Compression Volumes)? Hi @sebbas, thank you for that tip. this is working on my end as well! I'm just curious. When you say "when it's fixed", is this a blender dev bug, or do we need to wait until OpenVDB is updated globally? Also - does the affect anything at all for our workflows?

@Polyflogger It seems this issue comes from OpenVDB and blosc (v1.5.0) directly: A build with the latest blosc (v1.21.0) has no problems whatsoever.

Now, OpenVDB explicitly recommends using blosc v1.5.0 because "there are known issues with using later versions of blosc" (https://www.openvdb.org/documentation/doxygen/dependencies.html). So in this case, I wouldn't just bump the version as we would when there's a bug with a lib.

I will try to find a workaround on the Blender side first. And then investigate the OpenVDB side further.

@Polyflogger It seems this issue comes from OpenVDB and blosc (v1.5.0) directly: A build with the latest blosc (v1.21.0) has no problems whatsoever. Now, OpenVDB explicitly recommends using blosc v1.5.0 because "there are known issues with using later versions of blosc" (https://www.openvdb.org/documentation/doxygen/dependencies.html). So in this case, I wouldn't just bump the version as we would when there's a bug with a lib. I will try to find a workaround on the Blender side first. And then investigate the OpenVDB side further.

Added subscriber: @dfelinto

Added subscriber: @dfelinto

@sebbas remember when setting a task to confirmed to make sure there is a module tagged. Otherwise it won't show in any of the queries.

@sebbas remember when setting a task to confirmed to make sure there is a module tagged. Otherwise it won't show in any of the queries.

Added subscriber: @Blackx

Added subscriber: @Blackx

I encountered the same issue/crash, which was then resolved by using the ZIP compresion instead of Blosc. I can create a test .blend file if it will be useful to anyone. (Crashing in both Blender 2.92 and 2.93 beta, Win10, nvidia 1070)

Also, it seems to be happening to other people in their simulations as well, see this StackExchange question .

I encountered the same issue/crash, which was then resolved by using the ZIP compresion instead of Blosc. I can create a test .blend file if it will be useful to anyone. (Crashing in both Blender 2.92 and 2.93 beta, Win10, nvidia 1070) Also, it seems to be happening to other people in their simulations as well, see this [StackExchange question ](https://blender.stackexchange.com/questions/214921/mantaflow-bake-always-crashes-cannot-be-resumed).
Member

Added subscriber: @lichtwerk

Added subscriber: @lichtwerk
Member

Changed status from 'Confirmed' to: 'Needs User Info'

Changed status from 'Confirmed' to: 'Needs User Info'
Member

Would like to gather info on whether this is resolved in 2.93? (because the suspicion is that 8dd43ac23e already fixed this.

Would like to gather info on whether this is resolved in 2.93? (because the suspicion is that 8dd43ac23e already fixed this.
Member

Changed status from 'Needs User Info' to: 'Archived'

Changed status from 'Needs User Info' to: 'Archived'
Member

No activity for more than a week. As per the tracker policy we assume the issue is gone and can be closed.

Thanks again for the report. If the problem persists please open a new report with the required information.

No activity for more than a week. As per the tracker policy we assume the issue is gone and can be closed. Thanks again for the report. If the problem persists please open a new report with the required information.

Added subscriber: @LarryB

Added subscriber: @LarryB

Having this problem in 3.0 with fluid as well, no matter which compression or cache format I select. Replay works flawlessly though but I really would like to bake it.

Follow up edit: If the emitter object / inflow is hidden in the viewport, the crash happens. Otherwise it seems to be fine.

Scratch that. I worked 2 times then back to the usual crash. Pls fix.

Having this problem in 3.0 with fluid as well, no matter which compression or cache format I select. Replay works flawlessly though but I really would like to bake it. Follow up edit: If the emitter object / inflow is hidden in the viewport, the crash happens. Otherwise it seems to be fine. Scratch that. I worked 2 times then back to the usual crash. Pls fix.
Sign in to join this conversation.
No Label
Interest
Alembic
Interest
Animation & Rigging
Interest
Asset Browser
Interest
Asset Browser Project Overview
Interest
Audio
Interest
Automated Testing
Interest
Blender Asset Bundle
Interest
BlendFile
Interest
Collada
Interest
Compatibility
Interest
Compositing
Interest
Core
Interest
Cycles
Interest
Dependency Graph
Interest
Development Management
Interest
EEVEE
Interest
EEVEE & Viewport
Interest
Freestyle
Interest
Geometry Nodes
Interest
Grease Pencil
Interest
ID Management
Interest
Images & Movies
Interest
Import Export
Interest
Line Art
Interest
Masking
Interest
Metal
Interest
Modeling
Interest
Modifiers
Interest
Motion Tracking
Interest
Nodes & Physics
Interest
OpenGL
Interest
Overlay
Interest
Overrides
Interest
Performance
Interest
Physics
Interest
Pipeline, Assets & IO
Interest
Platforms, Builds & Tests
Interest
Python API
Interest
Render & Cycles
Interest
Render Pipeline
Interest
Sculpt, Paint & Texture
Interest
Text Editor
Interest
Translations
Interest
Triaging
Interest
Undo
Interest
USD
Interest
User Interface
Interest
UV Editing
Interest
VFX & Video
Interest
Video Sequencer
Interest
Virtual Reality
Interest
Vulkan
Interest
Wayland
Interest
Workbench
Interest: X11
Legacy
Blender 2.8 Project
Legacy
Milestone 1: Basic, Local Asset Browser
Legacy
OpenGL Error
Meta
Good First Issue
Meta
Papercut
Meta
Retrospective
Meta
Security
Module
Animation & Rigging
Module
Core
Module
Development Management
Module
EEVEE & Viewport
Module
Grease Pencil
Module
Modeling
Module
Nodes & Physics
Module
Pipeline, Assets & IO
Module
Platforms, Builds & Tests
Module
Python API
Module
Render & Cycles
Module
Sculpt, Paint & Texture
Module
Triaging
Module
User Interface
Module
VFX & Video
Platform
FreeBSD
Platform
Linux
Platform
macOS
Platform
Windows
Priority
High
Priority
Low
Priority
Normal
Priority
Unbreak Now!
Status
Archived
Status
Confirmed
Status
Duplicate
Status
Needs Info from Developers
Status
Needs Information from User
Status
Needs Triage
Status
Resolved
Type
Bug
Type
Design
Type
Known Issue
Type
Patch
Type
Report
Type
To Do
No Milestone
No project
9 Participants
Notifications
Due Date
The due date is invalid or out of range. Please use the format 'yyyy-mm-dd'.

No due date set.

Dependencies

No dependencies set.

Reference: blender/blender#86053
No description provided.