Hi,

There is a issue on recent Intel graphics (SKL+) on seascape demo.

Screenshot (occurs after couple of seconds):

https://imgur.com/a/TiTj1dzThe reason is a float overflow if order of operations is changed by shader compiler.

There is a simple workaround for this by changing one dot operation to use precise float.

Instead of using:

`float hash(vec2 p)`

{

float h = dot(p, vec2(127.09999847412109375, 311.70001220703125));

return fract(sin(h) * 43758.546875);

}

This can be used

`float dot_precise(vec2 a, vec2 b)`

{

precise float result = 0;

result += a.x * b.x;

result += a.y * b.y;

return result;

}

float hash(vec2 p)

{

float h = dot_precise(p, vec2(127.09999847412109375, 311.70001220703125));

return fract(sin(h) * 43758.546875);

}

in s01-ps.frag.