Is there support for 8bits per channel and 16bits per channel UAV in a compute shader?

Question

I'm working on a compute shader. It needs to output some textures as UAVs. Some of them have 8bits components and some have 16bits per channel. Consider this line:

RWTexture2D<float4> _watNormTex;

I bind a R8G8B8A8_UNORM texture to it and the output is black but when I bind a R32G32B32A32_FLOAT texture it saves the correct values. So it seems it has a problem with texture format.

What can I do to output to 8bit or 16bit per channel textures?

PS: My test device has a Geforce 710 GT GPU.

Do you have debug layer enabled? It should provide you some information in case something is wrong, also without posting any code it's gonna be difficult to source the error (as a side note I tried your case and it worked with both formats, so I would think something is wrong in your pipeline setup). — mrvux
@catflier You are absolutely right. It was a framework setup problem. Thanks — morteza khosravi

morteza khosravi morteza khosravi · Accepted Answer · 2017-02-05T07:28:14

0

votes

It was a setup problem as @catflier suggested. The formats should work as expected.

Is there support for 8bits per channel and 16bits per channel UAV in a compute shader?

1 Answers