Bake diffuse causes exits the program and core dump #83600

Open
opened 2020-12-09 18:25:02 +01:00 by Karol Szczerba · 26 comments

System Information
Operating system: Ubuntu 20.04.1
Graphics card: Sapphire RX 5700 XT with latest drivers 20.45

Blender Version
Broken: 2.83.10, 2.83.9, 2.83,8, 2.90.1
Worked:

Short description of error
I have cube with simple material. When I hit bake (diffuse), Blender exits immediately.
Info from terminal:
"Memory access fault by GPU node-1 (Agent handle: 0x7f9faac09b00) on address 0x7f9e1782c000. Reason: Page not present or supervisor privilege.
./apps.sh: line 34: 84735 Aborted (core dumped) /home/karol/port_apps/blender-2.83.10-linux64/blender"

**System Information** Operating system: Ubuntu 20.04.1 Graphics card: Sapphire RX 5700 XT with latest drivers 20.45 **Blender Version** Broken: 2.83.10, 2.83.9, 2.83,8, 2.90.1 Worked: **Short description of error** I have cube with simple material. When I hit bake (diffuse), Blender exits immediately. Info from terminal: "Memory access fault by GPU node-1 (Agent handle: 0x7f9faac09b00) on address 0x7f9e1782c000. Reason: Page not present or supervisor privilege. ./apps.sh: line 34: 84735 Aborted (core dumped) /home/karol/port_apps/blender-2.83.10-linux64/blender"
Author

Added subscriber: @Baszczer

Added subscriber: @Baszczer
Author

I've removed amdgpu-pro drivers and installed rocm-dkms and everything seams to work fine.

I've removed amdgpu-pro drivers and installed rocm-dkms and everything seams to work fine.
Author

I've removed old config files and baking works fine with amdgpu-pro.

I've removed old config files and baking works fine with amdgpu-pro.
Member

Added subscriber: @lichtwerk

Added subscriber: @lichtwerk
Member

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

Changed status from 'Needs Triage' to: 'Resolved'
Philipp Oeser self-assigned this 2020-12-11 17:55:34 +01:00
Member

This sound like issue is resolved, right?

Will close, but feel free to comment again if issues persist...

This sound like issue is resolved, right? Will close, but feel free to comment again if issues persist...
Author

No, still have problems.
New config and blender exits with this:

"Split kernel error: failed to load kernel_lamp_emission
Memory access fault by GPU node-1 (Agent handle: 0x7fb0674c2d00) on address 0x7fae2e0e6000. Reason: Page not present or supervisor privilege.
./apps.sh: line 34: 6726 Aborted (core dumped) /home/karol/port_apps/blender-2.83.10-linux64/blender"

I will reinstall Ubuntu, and then see if there are problems.

No, still have problems. New config and blender exits with this: "Split kernel error: failed to load kernel_lamp_emission Memory access fault by GPU node-1 (Agent handle: 0x7fb0674c2d00) on address 0x7fae2e0e6000. Reason: Page not present or supervisor privilege. ./apps.sh: line 34: 6726 Aborted (core dumped) /home/karol/port_apps/blender-2.83.10-linux64/blender" I will reinstall Ubuntu, and then see if there are problems.
Member

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

Changed status from 'Resolved' to: 'Needs Triage'
Philipp Oeser removed their assignment 2020-12-16 12:17:23 +01:00

Added subscriber: @mano-wii

Added subscriber: @mano-wii

Can you provide the .blend file that demonstrates the problem?
It is quite possible that this is a problem in the AMD graphics driver of your hardware and not really in Blender.
Driver updates sometimes resolve these issues.

Can you provide the .blend file that demonstrates the problem? It is quite possible that this is a problem in the AMD graphics driver of your hardware and not really in Blender. Driver updates sometimes resolve these issues.
Author

In #83600#1077673, @mano-wii wrote:
Can you provide the .blend file that demonstrates the problem?
It is quite possible that this is a problem in the AMD graphics driver of your hardware and not really in Blender.
Driver updates sometimes resolve these issues.

I have latest drivers installed (20.45) on a clean Ubuntu and Blender crashes on a simple scene with default cube only.

Today's attempt on a fresh system:

Cycles: compiling OpenCL program base...
Cycles: compiling OpenCL program bake...
Kernel compilation of base finished in 0.69s.
Kernel compilation of bake finished in 57.62s.
Cycles: compiling OpenCL program background...
Kernel compilation of background finished in 12.32s.
Split kernel error: failed to load kernel_path_init
Memory access fault by GPU node-1 (Agent handle: 0x7f16421d7a00) on address 0x7f15682df000. Reason: Page not present or supervisor privilege.
Aborted (core dumped)

> In #83600#1077673, @mano-wii wrote: > Can you provide the .blend file that demonstrates the problem? > It is quite possible that this is a problem in the AMD graphics driver of your hardware and not really in Blender. > Driver updates sometimes resolve these issues. I have latest drivers installed (20.45) on a clean Ubuntu and Blender crashes on a simple scene with default cube only. Today's attempt on a fresh system: Cycles: compiling OpenCL program base... Cycles: compiling OpenCL program bake... Kernel compilation of base finished in 0.69s. Kernel compilation of bake finished in 57.62s. Cycles: compiling OpenCL program background... Kernel compilation of background finished in 12.32s. Split kernel error: failed to load kernel_path_init Memory access fault by GPU node-1 (Agent handle: 0x7f16421d7a00) on address 0x7f15682df000. Reason: Page not present or supervisor privilege. Aborted (core dumped)
Author

I found this in Release Notes:
"Blender 2.90.1 users may experience page faults (details in dmesg log)."

I found this in Release Notes: "Blender 2.90.1 users may experience page faults (details in dmesg log)."
Member

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

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

So reading this again, I am still confused if we had a working driver version here?
If not: do you have a chance to downgrade driver versions to see if a former version works?

So reading this again, I am still confused if we had a working driver version here? If not: do you have a chance to downgrade driver versions to see if a former version works?
Author

In #83600#1086234, @lichtwerk wrote:
So reading this again, I am still confused if we had a working driver version here?
If not: do you have a chance to downgrade driver versions to see if a former version works?

Previous version will not install on the latest kernel and I don't want to mess with kernels.
I've contacted AMD and they are investigating this issue.
I remember that I also have had problems with the earlier drivers (20.40). Baking with 0% progress and frequent freezes of the system.

> In #83600#1086234, @lichtwerk wrote: > So reading this again, I am still confused if we had a working driver version here? > If not: do you have a chance to downgrade driver versions to see if a former version works? Previous version will not install on the latest kernel and I don't want to mess with kernels. I've contacted AMD and they are investigating this issue. I remember that I also have had problems with the earlier drivers (20.40). Baking with 0% progress and frequent freezes of the system.
Author

I recently changed the motherboard and ram.
Today I've checked baking again.
On clean, not modified config Blender baked with no problems.
I've swapped default config with my prefered one and Blender exits with this:
Split kernel error: failed to load kernel_path_init
Memory access fault by GPU node-1 (Agent handle: 0x7efea0b6c600) on address 0x7efe11e7a000. Reason: Page not present or supervisor privilege.
./apps.sh: line 38: 10542 Aborted (core dumped) /home/karol/port_apps/blender-2.83.10-linux64/blender
So I've removed config, Blender created new config and exited with this:
Split kernel error: failed to load kernel_path_init
Memory access fault by GPU node-1 (Agent handle: 0x7fdf2248f400) on address 0x7fdeac65c000. Reason: Page not present or supervisor privilege.
./apps.sh: line 38: 10627 Aborted (core dumped) /home/karol/port_apps/blender-2.83.10-linux64/blender
So I'm confused too.

I recently changed the motherboard and ram. Today I've checked baking again. On clean, not modified config Blender baked with no problems. I've swapped default config with my prefered one and Blender exits with this: Split kernel error: failed to load kernel_path_init Memory access fault by GPU node-1 (Agent handle: 0x7efea0b6c600) on address 0x7efe11e7a000. Reason: Page not present or supervisor privilege. ./apps.sh: line 38: 10542 Aborted (core dumped) /home/karol/port_apps/blender-2.83.10-linux64/blender So I've removed config, Blender created new config and exited with this: Split kernel error: failed to load kernel_path_init Memory access fault by GPU node-1 (Agent handle: 0x7fdf2248f400) on address 0x7fdeac65c000. Reason: Page not present or supervisor privilege. ./apps.sh: line 38: 10627 Aborted (core dumped) /home/karol/port_apps/blender-2.83.10-linux64/blender So I'm confused too.
Author

New observations:

  1. I can bake on CPU
  2. Ryzen 2700 renders faster (viewport rendering) than RX 5700 XT, but only scene with default cube. More complex materials are rendered faster by RX 5700 XT.
New observations: 1. I can bake on CPU 2. Ryzen 2700 renders faster (viewport rendering) than RX 5700 XT, but only scene with default cube. More complex materials are rendered faster by RX 5700 XT.
Author

After yesterday's Ubuntu update, I get this prompt:
"Split kernel error: failed to load kernel_path_init"

After yesterday's Ubuntu update, I get this prompt: "Split kernel error: failed to load kernel_path_init"

Closed as duplicate of #84461

Closed as duplicate of #84461

Changed status from 'Duplicate' to: 'Needs Developer To Reproduce'

Changed status from 'Duplicate' to: 'Needs Developer To Reproduce'

Oops, I closed it as a duplicate but I didn't see that the problem here is in Ubuntu (not Windows).

Oops, I closed it as a duplicate but I didn't see that the problem here is in Ubuntu (not Windows).
Author

I've received the info from AMD about my problem:

"The issue has been reproduced and a fix has been identified and will be in implemented in 21.Q2 driver that will be released soon.
Thank you for your patience"

I've received the info from AMD about my problem: "The issue has been reproduced and a fix has been identified and will be in implemented in 21.Q2 driver that will be released soon. Thank you for your patience"
Author

After updating the drivers to version 20.50, Blender bakes textures using the gpu.
The following message appears in the terminal:
Split kernel error: failed to load kernel_path_init

After updating the drivers to version 20.50, Blender bakes textures using the gpu. The following message appears in the terminal: Split kernel error: failed to load kernel_path_init
Author

...but now when I render using GPU Compute, Blender crashes with this information in the terminal output:

Memory access fault by GPU node-1 (Agent handle: 0x7f8442658b00) on address 0x7f810073a000. Reason: Page not present or supervisor privilege.
./apps.sh: line 42: 2817 Aborted (core dumped) /home/karol/port_apps/blender-2.83.13-linux64/blender

https://www.youtube.com/watch?v=Od6hY_50Dh0

...but now when I render using GPU Compute, Blender crashes with this information in the terminal output: Memory access fault by GPU node-1 (Agent handle: 0x7f8442658b00) on address 0x7f810073a000. Reason: Page not present or supervisor privilege. ./apps.sh: line 42: 2817 Aborted (core dumped) /home/karol/port_apps/blender-2.83.13-linux64/blender https://www.youtube.com/watch?v=Od6hY_50Dh0
Author

Blender 2.83.13, amdgpu-pro 20.50

I did some research.
I've created scene with Susane and sun lamp. Rendered without problems (GPU Compute).
Then I've added Environment Texture. Three times in a row, Blender quit with the output in the terminal:

  1. Memory access fault by GPU node-1 (Agent handle: 0x7fe147d87b00) on address 0x7fdfe09d6000. Reason: Page not present or supervisor privilege.
    ./apps.sh: line 42: 2983 Aborted (core dumped) /home/karol/port_apps/blender-2.83.13-linux64/blender
  2. Memory access fault by GPU node-1 (Agent handle: 0x7fc38017ab00) on address 0x7fc1019a6000. Reason: Page not present or supervisor privilege.
    ./apps.sh: line 42: 5262 Aborted (core dumped) /home/karol/port_apps/blender-2.83.13-linux64/blender
  3. Memory access fault by GPU node-1 (Agent handle: 0x7f1645d78b00) on address 0x7f14019a6000. Reason: Page not present or supervisor privilege.
    ./apps.sh: line 42: 5412 Aborted (core dumped) /home/karol/port_apps/blender-2.83.13-linux64/blender

monkey_gpu_node_-1_test.blend

Blender 2.83.13, amdgpu-pro 20.50 I did some research. I've created scene with Susane and sun lamp. Rendered without problems (GPU Compute). Then I've added Environment Texture. Three times in a row, Blender quit with the output in the terminal: 1. Memory access fault by GPU node-1 (Agent handle: 0x7fe147d87b00) on address 0x7fdfe09d6000. Reason: Page not present or supervisor privilege. ./apps.sh: line 42: 2983 Aborted (core dumped) /home/karol/port_apps/blender-2.83.13-linux64/blender 2. Memory access fault by GPU node-1 (Agent handle: 0x7fc38017ab00) on address 0x7fc1019a6000. Reason: Page not present or supervisor privilege. ./apps.sh: line 42: 5262 Aborted (core dumped) /home/karol/port_apps/blender-2.83.13-linux64/blender 3. Memory access fault by GPU node-1 (Agent handle: 0x7f1645d78b00) on address 0x7f14019a6000. Reason: Page not present or supervisor privilege. ./apps.sh: line 42: 5412 Aborted (core dumped) /home/karol/port_apps/blender-2.83.13-linux64/blender [monkey_gpu_node_-1_test.blend](https://archive.blender.org/developer/F9902153/monkey_gpu_node_-1_test.blend)
Author

I have disposed of this card and will no longer report the current situation.

I have disposed of this card and will no longer report the current situation.
Philipp Oeser removed the
Interest
Render & Cycles
label 2023-02-09 14:01:58 +01:00
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
No Assignees
3 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#83600
No description provided.