of Microprocessors

MUSIC SYNTHESIS SOFTWARE 653
1. Double-word load accumulator
2. Double-word store accumulator
3. Double-word addition to accumulator
4. Double-word subtraction from accumulator
5. Double-word negation of accumulator
6. Double-word left shifting of accumulator
7. Double-word signed and unsigned right shifting of accumulator
8. Unsigned multiplication 8 X 8 = 16
9. Signed multiplication (uses unsigned multiplication as a subroutine)
8 X 8= 15
10. Unsigned division 16/8=8
11. Signed division (uses unsigned division as a subroutine) 15/8=8
Double-word addition is necessary for forming sums of products that will in
general be double-word values. Shifting operations are needed for scaling. Of
course, single-word addition, subtraction, negation, and shifting are also
necessary. It is often helpful to have a "load with sign extend" operation,
which is used to translate a single-word signed value into a double-word one.
It is surprising how many computers, both micro and mini, that claim to
have complete hardware arithmetic capability fail to include all of these
necessary operations.
A Fixed-Point Arithmetic Package
for the 6502
Here we will briefly describe the 8-bit arithmetic package for the 6502
shown in Fig. 18-3. A pseudo-16-bit accumulator is assumed to exist in
page zero memory at location ACCH for the high order half and ACCL for
the low order half. All subroutines that require a double-byte second operand
expect the address of the high-order half of the operand, which must be in
page zero, to be in the X index register. Single-byte second operands are
expected to be in the hardware accumulator.
Load and store are trivially simple. Although they can be easily done by
in-line code in the calling program, using the subroutines will make a
signal-processing program easier to read and later convert to a machine with
hardware to replace this package.
Double-precision addition and subtraction are equally straightforward.
Both process the low bytes first and then use the carry flag to transmit carry
information to the high bytes. Note that the 6502 does not have add and
subtract in which the carry flag is ignored. Therefore, it is frequently necessary
to clear or set the carry before an addition or subtraction, respectively.
Negation is performed by subtracting the pseudoaccumulator from zero,
which turns out to be faster than complementing all of the bits and then
incrementing.