Can the GPU on Colibri T20 512MB V1.2a support dynamic for loops in the shader source

We use Colibri T20 board and the SDK built based on Microsoft QFE Updates February 2015, While I compile the shader source code, a error cause it failed. The error message is:
error C5013: profile does not support “for” statements and “for” could not be unrolled.
There are for loops in my code like this:

float computeMinValueWithValidity() {\n"
" vec2 startPos = vFragTexCoord;\n"
" vec2 endPos = vFragTexCoord+vSamplesPerPixel;\n"
" for (float y = startPos.y; y <= endPos.y; y += vSampleSize.y) {\n"
" for (float x = startPos.x; x <= endPos.x; x += vSampleSize.x) {\n"
" ……"
" }\n"
" }\n"
" }\n"

I want to know if the GPU on T20 support dynamic for loops? Is it necessary that the iteration is CONST? Does it support NON-CONST iteration in for loops?

Great thanks

Did you already checked the the details in the Khronos OpenGL ES 2.0 documentation? There is also a docuemnt from NVidia with some details specific for the Tegra 2 processor.