Search Issue Tracker
Fixed in 2022.2.X
Votes
1
Found in
2019.4
2020.3
2020.3.26f1
2021.2
2022.1
2022.2
Issue ID
1398163
Regression
No
Full floating precision register is used when compiling Shader to GLES3
How to reproduce:
1. Open the user's attached project
2. In the Project window, under Assets, select the Shader "Modx-CharacterDissolve"
3. In the Inspector window, next to "Compile and show code", select the downward arrow to open the platform menu
4. Select the GLES3x platform
5. Select "Compile and show code"
6. Search for the line "vec3 u_xlat0" in the compiled Shader
Expected result: vec3 register is not used in the compiled Shader
Actual result: vec3 register is used in the compiled Shader
Reproducible with: 2019.4.36f1, 2020.3.30f1, 2021.2.15f1, 2022.1.0b11, 2022.2.0a7
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
- Other threads get resized when trying to resize thread groups on the Timeline of the Profiler window
- Green highlight for selected panel lacks contrast when Editor theme is set to Light
- [Android] AR Camera Background has a white/bright tint when deployed with Gamma Color Space and Vulkan API
- Crash on TextCore::OTL_GPOS_TableReader::GetLookupList when changing to a specific Font in the UI Builder
- Reverting [SerializeReference] List property creates persistent element overrides in Prefab Variants
Resolution Note (fix version 2022.2):
Fixed in 2022.2.0a9
We would recommend to use DX11-style separate texture/sampler state declaration with explicit half type information:
SamplerState sampler_MainTex;
Texture2D<half4> _MainTex;
Without explicit data type info on the texture, we cannot guarantee the most efficient half-precision utilization on the resulting shader. This syntax variant also provides a workaround for this particular bug case.