Search Issue Tracker
Third Party Issue
Votes
1
Found in
2019.4
2020.3
2021.1
2021.1.14f1
2021.2
2022.1
Issue ID
1351770
Regression
No
Same calculation gives different results when running Compute Shader due to Shader optimiser
How to reproduce:
1. Open the attached project
2. Open the Test Runner (Window -> General -> Test Runner)
3. In the Test Runner window, select 'Run All'
4. Inspect the results
Expected result: All tests pass
Actual result: Tests with Compute Shader fail
Reproducible with: 2019.4.29f1, 2020.3.16f1, 2021.1.17f1, 2021.2.0b7, 2022.1.0a5
Add comment
All about bugs
View bugs we have successfully reproduced, and vote for the bugs you want to see fixed most urgently.
Latest issues
- Keywords on Material Variants aren't automatically saved when changed on original Material through Shader Graph
- Light Probes get baked when calling LightProbes.Tetrahedralize
- Shadows flicker and cause visual artifacts when modifying a GameObject's bounds using Swizzle (Y Mask) and Sine Time nodes
- [WebGL] Frame rate drops by 5-20 fps when moving cursor or touch input in the Player
- Light bleeds when using box shaped spotlight with specific Emission Range values
Resolution Note (2022.1.X):
The compiler we use (FXC) erroneously thinks that bit 8 (mask 0x100) is always set. Unfortunately, there's no way to fix this.
Using DXC (by adding "#pragma use_dxc" in the compute shader) is a potential workaround.