System Architecture; Clocking And Pipelining; P�Og�Am Co�Nte - Holtek BS83B24C Manual

System Architecture

A key factor in the high-performance features of the Holtek range of microcontrollers is attributed
to their internal system architecture. The series of devices take advantage of the usual features found
within RISC microcontrollers providing increased speed of operation and enhanced performance.
The pipelining scheme is implemented in such a way that instruction fetching and instruction
execution are overlapped, hence instructions are effectively executed in one or two cycles for most
of the standard or extended instructions respectively. The exceptions to this are branch or call
instructions which need one more cycle. An 8-bit wide ALU is used in practically all instruction set
operations, which carries out arithmetic operations, logic operations, rotation, increment, decrement,
branch decisions, etc. The internal data path is simplified by moving data through the Accumulator
and the ALU. Certain internal registers are implemented in the Data Memory and can be directly
or indirectly addressed. The simple addressing methods of these registers along with additional
architectural features ensure that a minimum of external components is required to provide a
functional I/O control system with maximum reliability and flexibility. This makes the devices
suitable for low-cost, high-volume production for controller applications.

Clocking and Pipelining

The main system clock, derived from either an LIRC, LXT or HIRC oscillator is subdivided into
four internally generated non-overlapping clocks, T1~T4. The Program Counter is incremented at
the beginning of the T1 clock during which time a new instruction is fetched. The remaining T2~T4
clocks carry out the decoding and execution functions. In this way, one T1~T4 clock cycle forms
one instruction cycle. Although the fetching and execution of instructions takes place in consecutive
instruction cycles, the pipelining structure of the microcontroller ensures that instructions are
effectively executed in one instruction cycle. The exception to this are instructions where the
contents of the Program Counter are changed, such as subroutine calls or jumps, in which case the
instruction will take one more instruction cycle to execute.
For instructions involving branches, such as jump or call instructions, two machine cycles are
required to complete instruction execution. An extra cycle is required as the program takes one
cycle to first obtain the actual jump or call address and then another cycle to actually execute the
branch. The requirement for this extra cycle should be taken into account by programmers in timing
sensitive applications.
f
SYS
(S�stem Clock)
Phase Clock T1
Phase Clock T�
Phase Clock T3
Phase Clock T4
P�og�am Co�nte�
Pipelining
Rev. 1.00
PC
�etch Inst. (PC)
Exec�te Inst. (PC-1) �etch Inst. (PC+1)
Exec�te Inst. (PC)
System Clocking and Pipelining
�4
BS83B24C/BS83C40C
Touch Flash MCU
PC+1 PC+�
�etch Inst. (PC+�)
Exec�te Inst. (PC+1)
�e���a�� 0�� �01�