comp.lang.idl-pvwave archive: archive » total() vs. rebin() speed for 1D averaging or integration

Home » Public Forums » archive » total() vs. rebin() speed for 1D averaging or integration

Show: Today's Messages :: Show Polls :: Message Navigator
E-mail to friend

total() vs. rebin() speed for 1D averaging or integration [message #85426]

Thu, 01 August 2013 21:06

kagoldberg
Messages: 26
Registered: November 2012

Junior Member

In hopes that this could be useful for anyone trying to optimize calculation speeds, I've noticed that there can be big a difference between two ways of performing a one-direction averaging or 1D integration on a 2D array.

a = randomn(seed, 200, 2048)
a1 = total(a, 2) ;--- this performs a column total
a2 = rebin(a, 200, 1) ;--- this returns a column average

Trying a few different values for the array size, I see that for single-precision floating-point, there can be a 3x speed improvement with rebin as the winner. When the array size gets to be large (2048x2048), or when the array is a double-precision floating-point, then it's nearly a tie on my machine.

In any case, it's worth running a test, if you're choosing one over the other and you care about speed. YMMV.

Report message to a moderator

[Message index]

total() vs. rebin() speed for 1D averaging or integration

By: kagoldberg on Thu, 01 August 2013 21:06

Previous Topic:	New graphics from IDLde and the command line
Next Topic:	How to make scatter plot with colorbar in IDL?

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Sat Nov 29 14:25:59 PST 2025

Total time taken to generate the page: 0.00853 seconds