Hi Brainer,
A while ago I was fiddling around with creating software OpenGL routines (for my gp2x handheld), and I DID get something basic working.

I could emulate exactly how OpenGL did its rendering with regards to transformations.

I have a unit you could look at which includes all my OpenGL transformations, etc. but it is 1307 lines long...

Is it ok if I post it here?

Perhaps it may help you with your matrix issues...

cheers,
Paul