TMS320C5X_16 Datasheet, PDF (21/94 Pages) Texas Instruments – DIGITAL SIGNAL PROCESSORS

TMS320C5x, TMS320LC5x
DIGITAL SIGNAL PROCESSORS
SPRS030A – APRIL 1995 – REVISED APRIL 1996
16 × 16-bit parallel multiplier
The ’C5x uses a 16 × 16-bit hardware multiplier that is capable of computing a signed or an unsigned 32-bit
product in a single machine cycle. All multiply instructions, except the MPYU (multiply unsigned) instruction,
perform a signed multiply operation in the multiplier. That is, two numbers being multiplied are treated as
2s-complement numbers, and the result is a 32-bit 2s-complement number.
There are two registers associated with the multiplier: TREG0, a 16-bit temporary register that holds one of the
operands for the multiplier, and PREG, the 32-bit product register that holds the product. Four product shift
modes (PM) are available at the PREG’s output. These shift modes are useful for performing
multiply/accumulate operations, performing fractional arithmetic, or justifying fractional products. The PM field
of status register ST1 specifies the PM shift mode.
The product can be shifted one bit to compensate for the extra sign bit gained in multiplying two 16-bit
2s-complement numbers (MPY). A 4-bit shift is used in conjunction with the MPY instruction with a short
immediate value (13 bits or less) to eliminate the four extra sign bits gained in multiplying a 16-bit number by
a 13-bit number. Finally, the output of PREG can, instead, be right-shifted 6 bits to enable the execution of up
to 128 consecutive multiply/accumulates without the possibility of overflow.
The load-TREG0 (LT) instruction normally loads TREG0 to provide one operand (from the data bus), and the
MPY instruction provides the second operand (also from the data bus). A multiplication also can be performed
with a short or long immediate operand by using the MPY instruction with an immediate operand. A product is
obtained every two cycles except when a long immediate operand is used.
Four multiply/accumulate instructions (MAC, MACD, MADD, and MADS as defined in Table 7) fully utilize the
computational bandwidth of the multiplier, allowing both operands to be processed simultaneously. The data
for these operations is transferred to the multiplier during each cycle through the program and data buses. This
facilitates single-cycle multiply/accumulates when used with repeat ( RPT and RPTZ ) instructions. In these
instructions, the coefficient addresses are generated by the PC, while the data addresses are generated by the
ARAU. This allows the repeated instruction to access the values sequentially from the coefficient table and step
through the data in any of the indirect addressing modes. The RPTZ instruction also clears the accumulator and
the product register to initialize the multiply/accumulate operation.
The MACD and MADD instructions, when repeated, support filter constructs (weighted running averages) so
that as the sum-of-products is executed, the sample data is shifted in memory to make room for the next sample
and to eliminate the oldest sample. Circular addressing with MAC and MADS instructions also can be used to
support filter implementation.
auxiliary registers and auxiliary-register arithmetic unit (ARAU)
The ’C5x provides a register file containing eight auxiliary registers (AR0 – AR7). The auxiliary registers are used
for indirect addressing of the data memory or for temporary data storage. Indirect auxiliary-register addressing
allows placement of the data memory address of an instruction operand into one of the auxiliary registers. These
registers are referenced with a 3-bit auxiliary register pointer (ARP) that is loaded with a value from 0 through
7, designated AR0 through AR7, respectively. The auxiliary registers and the ARP can be loaded from data
memory, the ACC, the product register, or by an immediate operand defined in the instruction. The contents of
these registers can be stored in data memory or used as inputs to the central arithmetic logic unit (CALU). These
registers are accessible as memory-mapped locations within the ’C5x data-memory space.
The auxiliary register file (AR0 – AR7) is connected to the auxiliary register arithmetic unit (ARAU). The ARAU
can autoindex the current auxiliary register while the data memory location is being addressed. Indexing can
be performed either by ±1 or by the contents of the INDX register. As a result, accessing tables of information
does not require the CALU for address manipulation; thus, the CALU is free for other operations in parallel.
• POST OFFICE BOX 1443 HOUSTON, TEXAS 77251–1443
21