摘要 |
PURPOSE:To execute high speed vector arithmetic compared with software by inverting mask data used at the time of executing the vector arithmetic corresponding to the instruction address of a vector arithmetic instruction. CONSTITUTION:When an IF statement in a DO loop is executed, each element of a vector S(I) and that of a vector Y(I) ar compared, and a vector arithmetic unit 10 produces the mask data. Then the vector arithmetic instruction of A(I)*B(I) is executed, and is placed at an even address. On the other hand, the vector arithmetic instruction of D(I)*E(I) in an ELSE statement is placed at an odd address. When an instruction counter 9 stands at an even number (the lower-most order bit is '0'), the contents of a mast data register MDR 6 are transmitted to the vector arithmetic unit 10. Whereas said counter stands at an odd number (lowe-most order bit is '1'), said contents are transmitted to the unit 10 by inverting them. As a result the mask data can be inverted without software. |