ROHINI COLLEGE OF ENGINEERING & TECHNOLOGY

5.2. ARCHITECTURE OF DIGITAL SIGNAL PROCESSOR

The TMS320C5X processor has an advanced version of Hardware architecture,
with separate buses for program and data, which facilitate simultaneous access of
program and data. The program bus has separate lines to transmit data and address.
Similarly the data has separate lines to transmit data and address.
.

Fig.5.2.1.Internal architecture of TMS320C5X

[Source: ‘Digital Signal Processing Principles, Algorithms and Applications’ by
J.G. Proakis and D.G. Manolakis page-674]
1.32-bit ALU/accumulator
The 32-bit ALU and accumulator implement a wide range of arithmetic and
logical functions, the majority of which execute in a single cycle. The ALU is a
general-purpose arithmetic/logic unit that operates on 16-bit words taken from data
memory or derived from immediate instructions. In addition to the usual arithmetic
instructions, the ALU can perform Boolean operations, facilitating the bit manipulation
ability required of a high-speed controller. One input to the ALU always is supplied by

EE8591-DIGITAL SIGNAL PROCESSING

 ROHINI COLLEGE OF ENGINEERING & TECHNOLOGY

the accumulator, and the other input can be furnished from the product register (PREG)
of the multiplier, the accumulator buffer (ACCB), or the output of the scaling shifter
[which has been read from data memory or from the accumulator (ACC)]. After the
ALU performs the arithmetic or logical operation, the result is stored in the ACC where
additional operations, such as shifting, can be performed. Data input to the ALU can be
scaled by the scaling shifter.
The 32-bit ACC is split into two 16-bit segments for storage in data memory.
Shifters at the output of the ACC provide a left shift of 0 to 7 places. This shift is
performed while the data is being transferred to the data bus for storage. The contents of
the ACC remain unchanged. When the post scaling shifter is used on the high word of
the ACC (bits 31–16), the most significant bits (MSBs) are lost and the least significant
bits (LSBs) are filled with bits shifted in from the low word (bits 15–0). When the post
scaling shifter is used on the low word, the LSBs are filled with zeros.
2.Scaling shifters
The ’C5x provides a scaling shifter that has a 16-bit input connected to the data
bus and a 32-bit output connected to the ALU. This scaling shifter produces a left shift
of 0 to 16 bits on the input data. The shift count is specified by a constant embedded in
the instruction word or by the value in TREG1. The LSBs of the output are filled with
zeros; the MSBs may be either filled with zeros or sign extended, depending upon the
value of the sign-extension mode (SXM) bit of status register ST1. The ’C5x also
contains several other shifters that allow it to perform numerical scaling, bit extraction,
extended-precision arithmetic, and overflow prevention. These shifters are connected to
the output of the product register and the ACC.
3.Parallel logic unit
The parallel logic unit (PLU) is a second logic unit, additional to the main ALU,
that executes logic operations on data without affecting the contents of the ACC. The
PLU provides the bit-manipulation ability required of a high-speed controller and
simplifies control/status register operations. The PLU provides a direct logic operation
path to data memory space and can set, clear, test, or toggle multiple bits directly in a
data memory location, a control/status register, or any register that is mapped into data
memory space.

EE8591-DIGITAL SIGNAL PROCESSING

 ROHINI COLLEGE OF ENGINEERING & TECHNOLOGY

4.16 × 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.
5. 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

EE8591-DIGITAL SIGNAL PROCESSING

 ROHINI COLLEGE OF ENGINEERING & TECHNOLOGY

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.
6. Memory
The ’C5x implements three separate address spaces for program memory, data
memory, and I/O. Each space accommodates a total of 64K 16-bit words (see Figures 1
through 7). Within the 64K words of data space, the 256 to 32K words at the top of the
address range can be defined to be external global memory in increments of powers of
two, as specified by the contents of the global memory allocation register (GREG).
Access to global memory is arbitrated using the global memory bus request (BR) signal.
The ’C5x devices include a considerable amount of on-chip memory to aid in system
performance and integration including ROM, single-access RAM (SARAM), and dual-
access RAM (DARAM). The amount and types of memory available on each device are
shown in Table 1. On the ’C5x, the first 96 (0–5Fh) data-memory locations are
allocated for memory-mapped registers. This memory-mapped register space contains
various control and status registers including those for the CPU, serial port, timer, and
software wait-state generators. Additionally, the first 16 I/O port locations are mapped
into this data-memory space, allowing them to be accessed either as data memory using
single-word instructions or as I/O locations with two-word instructions. Two-word
instructions allow access to the full 64K words of I/O space. The mask-programmable
ROM is located in program memory space. Customers can arrange to have this ROM
programmed with contents unique to to any particular application.
The ROM is enabled or disabled by the state of the MP/MC control input upon
resetting the device or by manipulating the MP/MC bit in the PMST status register after

EE8591-DIGITAL SIGNAL PROCESSING

 ROHINI COLLEGE OF ENGINEERING & TECHNOLOGY

reset. The ROM occupies the lowest block of program memory when enabled. When
disabled, these addresses are located in the device’s external program-memory space.
The ’C5x also has a mask-programmable option that provides security protection for the
contents of on-chip ROM. When this internal option bit is programmed, no externally-
originating instruction can access the on-chip ROM. This feature can be used to provide
security for proprietary algorithms.
7. Interrupts and subroutines
The ’C5x implements four general-purpose interrupts, INT4–INT1, along with
reset (RS) and the nonmaskable interrupt (NMI) which are available for external
devices to request the attention of the processor. Internal interrupts are generated by the
serial port (RINT and XINT), by the timer (TINT), and by the software-interrupt
(TRAP, INTR, and NMI) instructions. Interrupts are prioritized with RS having the
highest priority, followed by NMI, and INT4 having the lowest priority. Additionally,
any interrupt except RS and NMI can be masked individually with a dedicated bit in the
interrupt mask register (IMR) and can be cleared, set, or tested using its own dedicated
bit in the interrupt flag register (IFR). The reset and NMI functions are not maskable.
All interrupt vector locations are on two-word boundaries so that branch instructions
can be accommodated in those locations. While normally located at program memory
address 0, the interrupt vectors can be remapped to the beginning of any 2K-word page
in program memory by modifying the contents of the interrupt vector pointer (IPTR)
located in the PMST status register. A built-in mechanism protects multicycle
instructions from interrupts. If an interrupt occurs during a multicycle instruction, the
interrupt is not processed until the instruction completes execution. This mechanism
applies to instructions that are repeated (using the RPT instruction) and to instructions
that become multicycle because of wait states. Each time an interrupt is serviced or a
subroutine is entered, the PC is pushed onto an internal hardware stack, providing a
mechanism for returning to the previous context. The stack contains eight locations,
allowing interrupts or subroutines to be nested up to eight levels deep.
8.Power-down modes
The ’C5x implements several power-down modes in which the ’C5x core enters
a dormant state and dissipates considerably less power. A power-down mode is invoked

EE8591-DIGITAL SIGNAL PROCESSING

 ROHINI COLLEGE OF ENGINEERING & TECHNOLOGY

either by executing the IDLE/IDLE2 instructions or by driving the HOLD input low.
When the HOLD signal initiates the power-down mode, on-chip peripherals continue to
operate; this power-down mode is terminated when HOLD goes inactive. While the
’C5x is in a power-down mode, all internal contents are maintained; this allows
operation to continue unaltered when the power-down mode is terminated. All CPU
activities are halted when the IDLE instruction is executed, but the CLKOUT1 pin
remains active. The peripheral circuits continue to operate, allowing peripherals such as
serial ports and timers to take the CPU out of its powered-down state. A power-down
mode, when initiated by an IDLE instruction, is terminated upon receipt of an interrupt.

EE8591-DIGITAL SIGNAL PROCESSING