OpenEuphoria

Re: Which is faster ?

• Posted by mattlewis (admin) Feb 17, 2009
• 977 views

Do you have any benchmarks that demonstrate this? I ran the following on r1380, and find() always won. I changed the length of the sequence, changed the values to be all integers (and used = instead of equal()), etc. Actually, the loop won one time, when s = {""}. There's just a lot more overhead in a euphoria loop than the native C loop in the backend find function (not to mention your optimizations for non-integer data).

 
sequence s = { 1,"2",3,4.4,5,6,7} 
constant REPS = 1000000 
 
atom t = time() 
for i = 1 to REPS do 
	for j = 1 to length(s) do 
		object x = s[j] 
		integer ix = find( j, s ) 
	end for 
end for 
atom find_time = time() - t 
 
t = time() 
for i = 1 to REPS do 
	for j = 1 to length(s) do 
		for k = 1 to length(s) do 
			object x = s[j] 
			if equal( s[k], j ) then 
				exit 
			end if 
		end for 
	end for 
end for 
atom loop_time = time() - t 
 
printf(1, "%d repetitions:\nfind: %1.3f\nloop: %1.3f\n", { REPS, find_time, loop_time }) 

Matt