Optimizing Your Code

Table of Contents

Previous Slide

Software Development in the UNIX Environment
Optimizing Your Code

How Code Gets Optimized

Compiler performs classical optimizations (see list below) and architecture-specific optimizations
Linker places routines to minimize instruction cache misses
Programmer uses the highest level of compiler optimization that gives accurate results

Classical Optimizations: (taken from High Performance Computing by Kevin Dowd published by O'Reilly )

**Examples of Classical Optimizations**
Type of Optimization	Unoptimized Code	Optimized Code
Copy propagation	x=y z=1+x	x=y z=1+y
Constant folding (evaluates at compile time	integer i,k parameter(i=100) k=200 j=i+k	j is a constant as well as i,k
Dead code Removal	Program main i=2 write(,) i stop i=4 write(,) i end	Program main i=2 write(,) i stop
Strength reduction	y=x*2 j=k2	x*x k+k
Variable renaming	x=y*z; q=r+x+x; x=a+b;	x0=y*z; q=r+x0+x0; x=a+b;
Common Subexpression Elimination	d=c*(a+b); e=(a+b)/2;	temp=a+b; d=c*temp; e=temp/2;
Loop Invariant Code Motion	do 10 i=1,n a(i)=b(i)+c+d e=g(k) 10 continue	temp=c+d do 10 i=1,n a(i)=b(i)+temp 10 continue e=g(k)