Make wider range of numbers representable with FPNum on the cost of smaller precision
#ifdef GL_ES
precision mediump float;
#endif
attribute vec2 vertex;
attribute vec2 texcoord;
attribute vec4 colors;
varying vec2 tex;
uniform mat4 mvp;
void main()
{
vec4 p = mvp * vec4(vertex, 0.0, 1.0);
gl_Position = p;
tex = texcoord;
}