Hi all, so I know there is a performance difference between normal loops and unrolled loops and I was wondering if this is also the case with conditional statements. For example many nested if statements vs one big line with AND and ORs, assuming the conditional logic is identical?
For example
If a and b and c then ... end
vs
if a then
if b then
if c then
...
end
end
end
After reading wikipedia about loop unrolling I’m guessing no, but still curious.
I do not believe so BUT don’t take my word for it
@Kirl Here’s a 10 times loop of 10 million iteration of each of your conditional statements. I print out the time it takes for the 10 million iterations for each of the 10 loops. They’re times are close enough that it probable doesn’t matter which one you use.
EDIT: It’s possible that when this gets converted to byte code, they turn out to be about the same code.
function setup()
a=true
b=true
c=true
for g=1,10 do
s=os.clock()
for z=1,10000000 do
if a and b and c then
x=1
end
end
print(os.clock()-s)
s=os.clock()
for z=1,10000000 do
if a then
if b then
if c then
x=1
end
end
end
end
print(os.clock()-s)
print()
end
end
Haha, thanks dave, I’ve expanded your example to 10 conditionals, I guess there is a difference but like you said it probably doesn’t matter in practice. I don’t think it results in the same bytecode though, because there is a definite speed difference. The oneline conditional check is slightly faster.
Tempting to take it to 100 conditionals… =)
function setup()
a=true
b=true
c=true
d=true
e=true
f=true
g=true
h=true
i=true
j=true
k=true
for y=1,10 do
s=os.clock()
for z=1,10000000 do
if a and b and c and d and e and f and g and h and i and j and k then
x=1
end
end
print(os.clock()-s)
s=os.clock()
for z=1,10000000 do
if a then
if b then
if c then
if d then
if e then
if f then
if g then
if h then
if i then
if j then
if k then
x=1
end
end
end
end
end
end
end
end
end
end
end
end
print(os.clock()-s)
print()
end
end
@Kirl I’m not sure how accurate the times are. I noticed when I ran your code, the pairs of time increased in time as they ran.
@Kirl Here’s your above code, but using a different timer. This timer seems to be more accurate.
function setup()
sc=require("socket")
a=true
b=true
c=true
d=true
e=true
f=true
g=true
h=true
i=true
j=true
k=true
for y=1,10 do
s=sc.gettime()
for z=1,10000000 do
if a and b and c and d and e and f and g and h and i and j and k then
x=1
end
end
print(sc.gettime()-s)
s=sc.gettime()
for z=1,10000000 do
if a then
if b then
if c then
if d then
if e then
if f then
if g then
if h then
if i then
if j then
if k then
x=1
end
end
end
end
end
end
end
end
end
end
end
end
print(sc.gettime()-s)
print()
end
end
@Kirl I compared the bytecodes created for the 2 sets of if statements below and the bytecodes for each were exactly the same. The difference in times for the above programs comparing the 2 loops was probably just other things running in the Codea code.
----------
if a then
if b then
if c then
c=5
end
end
end
----------
if a and b and c then
c=5
end
----------
Huh, interesting, thanks!
How did you convert to bytecode btw?
string.dump and a whole lot of work.