I recently saw a post about using nothing instead of pairs in loops, I thought it would be a slow and/or deprecated method and wanted to try it myself. If the measurements are true, pairs is the worst iterator in terms of performance.
Is there a reason to this performance impact? Here is the test code:
local dictionary = {
["aad"] = 4,
["hello"] = "a",
["yes"] = "no",
[653] = 1,
}
local array = {"43242","aadf","tgergre","gfgegf"}
local Frequency = 10^7
print("\n\n\n\n")
c = os.clock()
for i = 1, Frequency do
for i,v in ipairs(array) do
local index = i
local value = v
end
end
print("Ipairs with array: "..tostring(os.clock()-c))
print()
c = os.clock()
for i = 1, Frequency do
for i,v in pairs(dictionary) do
local index = i
local value = v
end
end
print("Pairs with dictionary: "..tostring(os.clock()-c))
c = os.clock()
for i = 1, Frequency do
for i,v in pairs(array) do
local index = i
local value = v
end
end
print("Pairs with array: "..tostring(os.clock()-c))
print()
c = os.clock()
for i = 1, Frequency do
for i,v in next, dictionary do
local index = i
local value = v
end
end
print("next with dictionary: "..tostring(os.clock()-c))
c = os.clock()
for i = 1, Frequency do
for i,v in next, array do
local index = i
local value = v
end
end
print("next with array: "..tostring(os.clock()-c))
print()
c = os.clock()
for i = 1, Frequency do
for i,v in dictionary do
local index = i
local value = v
end
end
print("Nothing with dictionary:"..tostring(os.clock()-c))
c = os.clock()
for i = 1, Frequency do
for i,v in array do
local index = i
local value = v
end
end
print("Nothing with array: "..tostring(os.clock()-c))
Any information is appreciated
Edit: Forgor ipairs doesnt work with dictionaries, fixed a few mistakes
I think that os.time() is a better way to measure performance impact with these sorts of things, though I think that tick() is pretty much the same. Honestly, other than this, I don’t know why pairs is so slow.
Oops, os.clock() is actually better, not os.time()
Yeah, those results seem more accurate. Still takes up a surprising amount of performance. And why it takes up more with an array than it does with a dictionary is a mystery to me…
I don’t know if I’m going to go back and edit all my coding now, I didn’t realize that having nothing was actually faster. I’ve always used pairs.
ipairs is still the best for dictionaries, but it looks like nothing is better on arrays. I’ll mark this comment as the solution until someone more knowledged explains it
local dictionary = {
["aad"] = 4,
["hello"] = "a",
["yes"] = "no",
[653] = 1,
}
print("ipairs:")
for i,v in ipairs(dictionary) do
print("\t",i,v)
end
print("pairs:")
for i,v in pairs(dictionary) do
print("\t",i,v)
end
print("next:")
for i,v in next, dictionary do
print("\t",i,v)
end
print("nothing:")
for i,v in dictionary do
print("\t",i,v)
end
prints this:
Ipairs does not work with dictionaries.
Also with the next test you used the dictionary twice, not the array
Here are my results:
(which is fastest seems to change everytime I run it)
It’s still surprising that next and nothing are the fastest.
Has anyone actually had code where this made a huge difference? Normally it’s more about what happens inside of the loop instead of how I iterate so I just use pairs because I know it will work with any table
Yeah, I made a few errors in the test code and completely forgot ipairs doesn’t work with dictionaries. However nothing and next are still faster than pairs.
