| Re: Most Common IDL Programming Errors [message #59897 is a reply to message #59761] |
Fri, 11 April 2008 14:23   |
Vince Hradil
Messages: 574
Registered: December 1999
|
Senior Member |
|
|
On Apr 11, 3:52 pm, "R.G. Stockwell" <notha...@noemail.com> wrote:
> "Vince Hradil" <hrad...@yahoo.com> wrote in message
>
> news:b4e816fa-1c96-44a1-96bb-1a4e7395d376@24g2000hsh.googleg roups.com...
>
>> On Apr 11, 12:41 pm, "R.G. Stockwell" <notha...@noemail.com> wrote:
>>> <sav...@nsidc.org> wrote in message
>
>>> news:ywkufxtss6b9.fsf@snowblow.colorado.edu...
>
>>>> "R.G. Stockwell" <notha...@noemail.com> writes:
>
>>>> > a = fltarr(len) + !values.f_nan
>
>>>> What about
>>>> a = make_array(len, value=!values.f_nan)
>
>>>> That's what I use. I'm sure it saves me some computational power by not
>>>> having
>>>> to compute those values I'm just going to overwrite anyway. :)
>
>>> You know, I am just the kind of person who will end up time testing
>>> those variations. And seeing what operation is faster: fltarr(len) +
>>> !values.f_nan,
>>> fltarr(len) - !values.f_nan, fltarr(len) * !values.f_nan, or
>>> fltarr(len)/!values.f_nan
>
>>> :)
>
>>> Cheers,
>>> bob
>
>> I'm waiting with bated breath. 8^)
>
> Here they are (10k iterations, ran loops over the different
> commands, repeated them in different orders, all functions and arrays
> were previously created):
>
> (and by the way WOW! - make array wins)
>
> plus 89.172000 Seconds.
>
> minus 89.188000 Seconds.
>
> multiply 90.531000 Seconds.
>
> divide 89.485000 Seconds.
>
> makearray 2.0000000 Seconds.
>
> makearray 1.9840000 Seconds.
>
> divide 90.250000 Seconds.
>
> multiply 90.219000 Seconds.
>
> minus 88.797000 Seconds.
>
> plus 88.125000 Seconds.
"Wow!" is right. I was being facetious originally, but I think I
really learned something. Thanks for the effort.
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
