Transform matrix elements are represented in fixed point notation, that is, as a 16-bit signed integer and a 16-bit fractional part. These precision limits apply during graphics engine matrix multiplication for all initial, intermediate, and final matrix element values.