Shader language version in ARM GLES3.1 Emulator ver 2.2?

Hi Sunil,

Let's see. We are aware about some issues with 'mix' using uvec types.

As 'mix' comes with overloaded versions, can you point which particular one is troublesome in your case?

Is it possible you share the whole shader source that causes the problem?

cheers,

Adam

Hi Sunil,

Let's see. We are aware about some issues with 'mix' using uvec types.

As 'mix' comes with overloaded versions, can you point which particular one is troublesome in your case?

Is it possible you share the whole shader source that causes the problem?

cheers,

Adam

Hi Adam,

Built in function 'mix' is not working even for very basic example of mixing colors from two textures., it works if I remove outColor = mix(texture(tex1, texOut),texture(tex2,texOut),0.5f); by outColor = texture(tex1, texOut); It means problem lies with 'mix' function only.

----------------------------------------------

#version 310 es

precision highp float;

in vec3 vv3colour;

in vec2 texOut;

out vec4 outColor;

uniform sampler2D tex1;

uniform sampler2D tex2;

void main() {

  outColor = mix(texture(tex1, texOut),texture(tex2,texOut),0.5f);

}

----------------------------------------

In another case where I can not share other share shader codes, I created sample fragment shader to reproduce. It seems mix(vec4,vec4,float) is not working and gives following error while linking

error C1115: unable to find compatible overloaded function "mix(vec3, vec3, float)" as shared in my above post.

-------------------------------------------------

#version 310 es

precision highp float;

in vec3 vv3colour;

out vec4 outColor;

void main() {

  outColor = mix(vec4(vv3colour,1.0),vec4(0.0,0.0,0.0,1.0),1.0);

}

--------------------------------------------------

Please let me know if it helps, and you can run above shaders successfully.

Hi Sunil,

it was helpful indeed, and we have reproduction for the problem.

I confirm this is a valid issue for the Emulator and we'll prepare a fix for it.

The not-so-convenient and temporary workaround for the moment would be dropping the built-in and switching to custom mix(vec4,vec4,float) function in your shader, by declaring it:

-------------------

vec4 mix(vec4 x, vec4 y, float a) { return (1-a)*x + a*y; }

-------------------

You can expect announcement on this thread when this issue gets resolution and the workaround is no longer necessary.

Thanks for your report,

Adam