RISC vs. CISC
T he simplest way t o examine t he advant ages and disadvant ages of RISC arc hit ec t ure is by c ont rast ing it wit h it 's predec essor: CISC (Complex Inst ruc t ion Set Comput ers) arc hit ec t ure.
Multiply ing Two Numbers in Memory
On t he right is a diagram represent ing t he
st orage sc heme f or a generic c omput er. T he
main memory is divided int o loc at ions
numbered f rom (row) 1: (c olumn) 1 t o (row)
6: (c olumn) 4. T he exec ut ion unit is
responsible f or c arrying out all c omput at ions.
However, t he exec ut ion unit c an only operat e
on dat a t hat has been loaded int o one of t he
six regist ers (A, B, C, D, E, or F ). Let 's say
we want t o f ind t he produc t of t wo numbers
- one st ored in loc at ion 2:3 and anot her
st ored in loc at ion 5:2 - and t hen st ore t he
produc t bac k in t he loc at ion 2:3.
The C ISC Approach
T he primary goal of CISC arc hit ec t ure is t o
c omplet e a t ask in as f ew lines of assembly
as possible. T his is ac hieved by building
proc essor hardware t hat is c apable of
underst anding and exec ut ing a series of
operat ions. F or t his part ic ular t ask, a CISC
proc essor would c ome prepared wit h a
spec if ic inst ruc t ion (we'll c all it "MULT ").
When exec ut ed, t his inst ruc t ion loads t he t wo values int o separat e regist ers, mult iplies t he operands in t he exec ut ion unit , and t hen st ores t he produc t in t he appropriat e regist er. T hus, t he ent ire t ask of mult iplying t wo numbers c an be c omplet ed wit h one inst ruc t ion:
ML 23 52
UT :, :
MULT is what is known as a "c omplex inst ruc t ion." It operat es direc t ly on t he c omput er's memory banks and does not require t he programmer t o explic it ly c all any loading or www- cs- faculty.stanfor d.edu/~ er ober ts/cour ses/soco/pr ojects/r isc/r isccisc/
RISC vs. CISC
st oring f unc t ions. It c losely resembles a c ommand in a higher level language. F or inst anc e, if we let "a" represent t he value of 2:3 and "b" represent t he value of 5:2, t hen t his c ommand is ident ic al t o t he C st at ement "a = a * b." One of t he primary advant ages of t his syst em is t hat t he c ompiler has t o do very lit t le work t o t ranslat e a high- level language st at ement int o assembly. Bec ause t he lengt h of t he c ode is relat ively short , very lit t le RAM is required t o st ore inst ruc t ions. T he emphasis is put on building c omplex inst ruc t ions direc t ly int o t he hardware. The RISC Approach
RISC proc essors only use simple inst ruc t ions t hat c an be exec ut ed wit hin one c loc k c yc le. T hus, t he "MULT " c ommand desc ribed above c ould be divided int o t hree separat e c ommands: "LOAD," whic h moves dat a f rom t he memory bank t o a regist er, "PROD," whic h f inds t he produc t of t wo operands loc at ed wit hin t he regist ers, and "ST ORE," whic h moves dat a f rom a regist er t o t he memory banks. In order t o perf orm t he exac t series of st eps desc ribed in t he CISC approac h, a programmer would need t o c ode f our lines of assembly:
LA A 23
OD , :
LA B 52
OD , :
PO A B
At f irst , t his may seem like a muc h less ef f ic ient way of c omplet ing t he operat ion. Bec ause t here are more lines of c ode, more RAM is needed t o st ore t he assembly level inst ruc t ions. T he c ompiler must also perf orm more work t o c onvert a high- level language st at ement int o c ode of t his f orm.
However, t he RISC st rat egy
also brings some very import ant
Emphasis on hardware
Emphasis on sof t ware
advant ages. Bec ause eac h
Inc ludes mult i- c loc k
Single- c loc k,
inst ruc t ion requires only one
c omplex inst ruc t ions
reduc ed inst ruc t ion only
c loc k c yc le t o exec ut e, t he
Memory- t o- memory:
Regist er t o regist er:
ent ire program will exec ut e in
"LOAD" and "ST ORE"
"LOAD" and "ST ORE"
approximat ely t he...
Please join StudyMode to read the full document