GPU Glitches, Freezing, crashes, speckles, ghosts, etc. #69208
Labels
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
2 Participants
Notifications
Due Date
No due date set.
Dependencies
No dependencies set.
Reference: blender/blender#69208
Loading…
Reference in New Issue
No description provided.
Delete Branch "%!s(<nil>)"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
System Information
Operating system: Win10 (currently up to date 2019-08-27)
Graphics card: NVidia GeForce RTX 2080 Ti Driver: 436.02 aka 2019-08-16 [26.21.14.3602]
Blender Version
v 2.80
Short description of error
The Graphics is glitching badly in various ways.
The two attached images show some of these issues.
1 has 'Ghosting' where multiple frames are overlaid.
2 has Green Speckles and pixel spot anomalies that flicker. There are also more in the display then what can be Screen grabbed. It's something like fireworks in a way.
The system will potentially Blue Screen Reboot, after Graphics Driver from glitching out also begins freezing & ultimately crashing and the entire OS becomes disabled by this.
I guess you should know this is happening, with latest Win10 OS updates and NVidia Drivers too?
Not sure how you can reproduce it. It does seem to be caused by Blender or maybe GPU usage.
Added subscriber: @MasterJames
Here is the Event Viewer during a Blender crash in this situation, where the NVidia Driver is failed. I remind you this is a high-end system and it simply can not be out of Memory.
Here is the Blender Event Viewer log...
It leads to a WER file that follows
This Event in the Viewer is of the Level: Application Error
I also ran it with --debug-gpu flag and got loads of output there.
Maybe this excerpt reveals something interesting to someone in the know?
I found a way to reproduce this following the Procedural Crystal tutorial.
https://www.youtube.com/watch?v=d-zpR1E84nQ
This file attached when switched to "Rendered" Viewport mode will typically crash Blender.
ProceduralCrystals-01.blend
Here is what one sees in the cmd window.
I thought it not only had to do with displacement but also in Eevee doing large 4096 shadow map settings, but really I guess it's anything pushing the system hard. Maybe my brand new state of the art graphics card is hooped.
Added subscriber: @lichtwerk
No problems with
ProceduralCrystals-01.blend
here ( linux, 970m, 430.26 drivers )But we seem to be dealing with multiple issues here (
ProceduralCrystals-01.blend
is about cycles rendering)?So first we need the files, then we can see if this needs to be separate reports....
I am going to have the hardware checked and I think sent back to Asus. The 2080 Ti I had to wait two months for and I guess they rushed these cards out?
I will attempt to report back later.
Still I am thinking maybe there should be a way you assess instability in the graphics card response times on average and pop up or log a warning saying graphics is performing poorly.
The Blender system could have a setting to use 99% by default and let uses tune it down if it's causing the card to fry etc. I'm not sure it seems like not only a good idea to limit issues but to also let the user know it's not Blender but Blender has detected sporadic frame rate etc.
I get green speckles dancing and other obvious indicators.
I have run a few tests on the hardware and it seems more like the problem since your test of the Cycles GPU Crash file above was not a problem. I suppose other then my thoughts above I am a little embarrassed to have brought this to your attention. Still it seemed like only Blender was causing this but then only Blender was pushing the GPU to it's limits. With things like 4096 shadow maps trying to get cycles like quality out of Eevee etc., which still falls short but has potential moving slowly forward.
They say it will take a couple of weeks to replace the card if their testing shows it has issues, so I will be offline until it is replaced.
Just another shot in the dark (there was another (unrelated -- was about poor rendering performance) report (also using 436.XX drivers).
I cannot test these 436.XX (not offered for my card/system), but just out of curiosity: have you tried if using 430.XX makes a difference?
I'll set this back to "Needs Information from User" since we are waiting for the card to be replaced (or the drivers test).
But everyone reading this: please dont close this for inactivity (even if it takes longer than the usual week for @MasterJames to reply).
I did have the driver from before and did a clean install of it to see as that was fine as far as I could tell a couple of months ago. Alas no difference and had put back the newest one after checking the previous driver. I can't check the other version number in this moment, but it was something like that.
The guy at the shop said it takes 6 to 8 weeks maybe more so I am screwed for 2 months and he said it doesn't mean the card that comes back will be without issues.
Still they will do an in house testing first so if I dont hear from them, then they also had issues, and that he gave no timeline but I would think I would hear from them within a week if it checked out on a different inhouse system (otherwise they will send it back on warranty). I guess I could have similarly tried it here in different hardware, but I am uncertain it would have even fit my old system (biggest card I have ever seen power consumption requirements etc.).
Changed status from 'Open' to: 'Archived'
Seem it was a hardware problem.
I guess the idea of being able to detect a jamming card and notify the user of intermittent hardware functionality is a worthy endeavor from my perspective.