Hi,
Yes, that's probably correct - the joy of switching from C to C++.
Normally switching from old C code to our C++ image class only loses
a small amount of speed (whilst making the code much more
maintainable and reusable etc etc) but in the case of SUSAN the old C
code had been _hugely_ hand-optimised (was originally part of a real-
time computer vision system that I wrote many years ago) and hence
the big slowdown here. Because SUSAN is not used very much in our
general analysis pipelines (except in FILM where the data is always
quite spatially small) we felt that the big slowdown in SUSAN wasn't
a big issue.
Cheers, Steve.
On 26 Aug 2007, at 22:11, Robert Terwilliger wrote:
> Dear FSL,
>
> I have found susan (fsl_4) to require an ~8-fold longer time than
> susan_smooth (fsl_3).
>
> I have pasted sample output below. I am running on a core 2 duo iMac.
>
> Any ideas?
>
> Regards,
>
> Robert Terwilliger
>
>
> ##################
>
> clone$ time /usr/local/fsl_3/bin/susan_smooth sample 1000
> sample_fsl_3 1 3 0 0
>
> real 0m13.918s
> user 0m12.246s
> sys 0m0.263s
> clone$ time /usr/local/fsl/bin/susan sample 1000 1 3 0 0 sample_fsl_4
>
> real 1m41.747s
> user 1m28.331s
> sys 0m2.499s
>
>
------------------------------------------------------------------------
---
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director, Oxford University FMRIB Centre
FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
------------------------------------------------------------------------
---
|