Three key characteristics of RISC (Reduced Instruction Set Computer):
A memory access can take as much as 100 to 500 times the time for the cpu to execute an instruction. If data used in computation is already in a register, then there is no need for a memory access. If the processor does not need to wait for memory, then the processor is busier, which results in better performance.
Instructions are executed in the fetch-execute cycle. If the computer waits till the end of one cycle to start the next, then components are idle during different parts of the cycle. If the cycles are overlapped then the machine can get instructions through more quickly and get more done: while decoding one instruction, fetch the next; while executing one instruction, decode the next.
Here is a conceptual view of pipelining. We will discuss MIPS pipelining in more detail.Instr 1 | fetch instr | decode instr | fetch operands | execute | ||
Instr 2 | fetch instr | decode instr | fetch operands | execute | ||
Instr 3 | fetch instr | decode instr | fetch operands | execute |
CISC | RISC | |||
---|---|---|---|---|
IBM 370/168 | Intel 486 | SPARC | MIPS R4000 | |
Year | 1973 | 1989 | 1987 | 1991 |
Number of Instructions | 208 | 235 | 69 | 94 |
Instruction Size | 2 - 6 | 1 - 11 | 4 | 4 |
Addressing Modes | 4 | 11 | 1 | 1 |
Number of Registers | 16 | 8 | 40 - 520 | 32 |
Email Me |
Office Hours |
My Home Page |
Department Home |
MCC Home Page
© Copyright Emmi Schatz 2009