That is right. I started using fields::use_real_fields and it got
about 2 times faster.
Could you tell me how much it influences the results ?
One more funny thing I noticed about fields::use_real_fields is that now
the problem with not working HDF fields output that I described in an
other mail on this forum ("Unable to output fields to HDF in MPI meep.")
seems to be solved. When using real fields I can output
HDF fields in MPI mode without problems.
Thank you.
Thomas Jefferson
2008/7/13 Steven G. Johnson <[EMAIL PROTECTED]>:
> On Jul 7, 2008, at 4:37 AM, Jefferson Thomas wrote:
>> I am developing an application using C++ meep. I have just done some
>> speed tests
>> in the same environment using meep in Scheme mode and as a C++
>> program.
>> Here are the files i used:
> [...]
>>
>> As you see both descibe the same experimental conditions with a 5x10x2
>> volume, resolution of 20 and a continuous lightsource with frequency
>> 0.7 in 1,1,1.
>> Running them on the same system gives some strange result - the
>> simulation using Scheme finished around 5 times faster that the one
>> using C++ !
>> Is there something I dont understand here ? As far as I can see both
>> solutions should operate exactly with about the same speed.
>> Why does it differ so much ? What can I do to make the C++ program run
>> as fast as the Scheme one ?
>
> Well, for one thing the Scheme interface defaults to using real
> fields, whereas in the C++ interface you need to specify this
> explicity by calling fields::use_real_fields ...
>
> Note also that your source is not in the same place in the two cases,
> because in the C++ interface the origin defaults to being at the
> corner of the cell rather than at the center.
>
> Steven
>
> _______________________________________________
> meep-discuss mailing list
> meep-discuss@ab-initio.mit.edu
> http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss
>
_______________________________________________
meep-discuss mailing list
meep-discuss@ab-initio.mit.edu
http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss
