Cycles render different between GPU and CPU on transparent shader #39885

Closed
opened 2014-04-24 22:49:35 +02:00 by Patrice Bertrand · 10 comments

In most cases, Cycles render is exactly identical be it on GPU or CPU. I noticed a slight difference when rendering on a home-made 3 server farm, 2 with GPU, 1 with CPU, it gave a flickering effect because frames made with CPU are a bit different. I narrowed it down to the transparent shader, as the attached file bug-transparent-cycle.blend demonstrates. Render it once on GPU, then again on CPU with another slot and compare. Now I don't know if it's an absolute rule that Cycles should render identically on GPU and CPU, is it ?

In most cases, Cycles render is exactly identical be it on GPU or CPU. I noticed a slight difference when rendering on a home-made 3 server farm, 2 with GPU, 1 with CPU, it gave a flickering effect because frames made with CPU are a bit different. I narrowed it down to the transparent shader, as the attached file [bug-transparent-cycle.blend](https://archive.blender.org/developer/F85853/bug-transparent-cycle.blend) demonstrates. Render it once on GPU, then again on CPU with another slot and compare. Now I don't know if it's an absolute rule that Cycles should render identically on GPU and CPU, is it ?

Changed status to: 'Open'

Changed status to: 'Open'

Added subscriber: @PatriceBertrand

Added subscriber: @PatriceBertrand

I tested the file on a 2.68RC1, and the 2 renders are identical, so something happened between 2.68 and 2.70, I don't have many versions at hand to try.
My specs: Windows 64bits, GeForce GTX560Ti.

I tested the file on a 2.68RC1, and the 2 renders are identical, so something happened between 2.68 and 2.70, I don't have many versions at hand to try. My specs: Windows 64bits, GeForce GTX560Ti.

Added subscriber: @mont29

Added subscriber: @mont29
Brecht Van Lommel was assigned by Bastien Montagne 2014-04-25 08:29:36 +02:00

This looks like a duplicate of #39761?

This looks like a duplicate of #39761?

Not exactly a duplicate, since #39761 states that cycles render in view port is not right, while it is correct in a real render. In my case, it is the full render that is different.

Not exactly a duplicate, since #39761 states that cycles render in view port is not right, while it is correct in a real render. In my case, it is the full render that is different.

Added subscriber: @lopataasdf

Added subscriber: @lopataasdf

Changed status from 'Open' to: 'Archived'

Changed status from 'Open' to: 'Archived'

This file has (near) overlapping faces, which is always problematic for raytracers. Here different float precision on the GPU and CPU makes the results different. The suggestion for now is to avoid this case and increase the distance a bit. We might improve the accuracy in the future, but for now it's not considered a bug.

This file has (near) overlapping faces, which is always problematic for raytracers. Here different float precision on the GPU and CPU makes the results different. The suggestion for now is to avoid this case and increase the distance a bit. We might improve the accuracy in the future, but for now it's not considered a bug.

Thank you for your analysis. Still it seems strange to call this anything else than bug. Particularly since it worked fine before. I do agree it is not a major problem, it is not very often that cpu and gpu are used together and must match. I think using shrinkwrap with a very small offset is pretty commonplace.

Thank you for your analysis. Still it seems strange to call this anything else than bug. Particularly since it worked fine before. I do agree it is not a major problem, it is not very often that cpu and gpu are used together and must match. I think using shrinkwrap with a very small offset is pretty commonplace.
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
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
4 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#39885
No description provided.