A system that is to be placed on a wafer is partitioned into reasonably large functions, each provided with a set of I/O and power pads. The wafer design is called ''''design A''''. A second wafer design (design B) that is the mirror image of design A is also constructed. Wafers of designs A and B are tested and divided into two groups: group I wafers have relatively few functions that are inoperative; group II wafers have relatively few functions that are operative. Inoperative functions are removed from group I wafers and discarded; good functions are removed from group II wafers and retained. A given function on wafer A is the mirror image of the same function on wafer B. Therefore, a given function from a group II wafer A (or B) can be inverted and attached to a group I wafer B (or A) that has had the corresponding function removed from it. The I/O and power pads of the function removed from the group II wafer are joined to the I/O and power pads remaining on the group I wafer.