Fastest way to check divisibility?

PullJosh

Which is faster?

or

Or is there an even better way I haven't thought of?

Znapi

Mod would probably be faster. Profile it yourself?

bobbybee

Znapi wrote:
Mod would probably be faster. Profile it yourself?

PullJosh

Assuming this is a fair test, mod is faster (matching your prediction).

With the list n showing, mod time was 41.611 and round time was 44.897. With the list hidden, it went much faster (as expected). However, the ratios between the times were quite similar to the original test.
(Not sure why I didn't think to test it myself… Thanks for reminding me that that's an option. :3)

BookOwl

In my testing, it took 1.18 secs to compute

100,000 times, and 1.45 secs to compute

100,000 times. That means it took around 0.0000118 secs to do each mod test, and 0.0000145 to do the other test.
I think that the mod test is clearer though, as long as you know what ‘mod’ is.

bobbybee

Also, microoptimizations are badddd.

Thepuzzlegame

bobbybee wrote:
Also, microoptimizations are badddd.

More like unnecessary.

PullJosh

Thepuzzlegame wrote:
bobbybee wrote:
Also, microoptimizations are badddd.
More like unnecessary.

I wouldn't have done it if I thought it would mean so little.

bobbybee

PullJosh wrote:
Thepuzzlegame wrote:
bobbybee wrote:
Also, microoptimizations are badddd.
More like unnecessary.
I wouldn't have done it if I thought it would mean so little.

As a rule of thumb, reporters are fast – this is true in a lot of high level languages, with some gotchas once in a while.

As long as this code isn't being called in a loop or something, you should be good.

PullJosh

bobbybee wrote:
PullJosh wrote:
Thepuzzlegame wrote:
bobbybee wrote:
Also, microoptimizations are badddd.
More like unnecessary.
I wouldn't have done it if I thought it would mean so little.
As a rule of thumb, reporters are fast – this is true in a lot of high level languages, with some gotchas once in a while.

As long as this code isn't being called in a loop or something, you should be good.

Good to know.

MartinBraendli

The conclusion “it probably doesnt matter which one you pick” is right, however your speed tests underestimate the margin of difference (div-round takes more than 50% longer than modulo). Do your tests inside a function which is run without screen refresh and leave away constant time factors like random(), setVar(), and addElementToList().

WooHooBoy

MartinBraendli wrote:
The conclusion “it probably doesnt matter which one you pick” is right, however your speed tests underestimate the margin of difference (div-round takes more than 50% longer than modulo). Do your tests inside a function which is run without screen refresh and leave away constant time factors like random(), setVar(), and addElementToList().

For that matter, test it on every programming language you can think of!

MartinBraendli

Why? Modulo (integer operation) will allways be faster than Division (floating point operation), especially since you need 3 flops to check for divisibility. So I think the Modulo thing will be faster in any programming language (except If you run it on a super exotic processeor, which for reasons unclear to me, processes floating point operations faster than integer operations).

WooHooBoy

MartinBraendli wrote:
Why? Modulo (integer operation) will allways be faster than Division (floating point operation), especially since you need 3 flops to check for divisibility. So I think the Modulo thing will be faster in any programming language (except If you run it on a super exotic processeor, which for reasons unclear to me, processes floating point operations faster than integer operations).

It would be interesting to see exactly how much difference it makes in other languages.

__init__

fastest =

totes

Last edited by __init__ (Aug. 20, 2015 21:50:02)

ev3coolexit987654

__init__ wrote:

fastest =

totes

fixed
Edit: Is i necessary?

Last edited by ev3coolexit987654 (Aug. 20, 2015 21:42:29)

bobbybee

ev3coolexit987654 wrote:
__init__ wrote:
fastest =
totes
fixed
Edit: Is i necessary?

…. Pretty much none of that script is necessary

MartinBraendli

__init__ wrote:

fastest =

totes

I think think this is a joke, like “esotheric algorithms”. If you have never heard of esotheric algorithms, go check out bogobogosort. The site also has other intersting esotheric stuff (programming context, not the karma-yoga stuff).

PullJosh

MartinBraendli wrote:
__init__ wrote:
fastest =
totes
I think think this is a joke, like “esotheric algorithms”. If you have never heard of esotheric algorithms, go check out bogobogosort. The site also has other intersting esotheric stuff (programming context, not the karma-yoga stuff).

Darn… I was hoping for the karma-yoga.

Discuss Scratch