[Insight-developers] Some casting improvements.

[Bradley Lowekamp]

Kent,

I thought you were going to take this some place else....

What do you mean by "std::algorithm copy semantics"?

Are you taking about having a begin and end method to out array-like classes?

The classes derived from itk::Array are derived from vnl_vector, which have a begin/end method.
FixedArray has Begin and End.
Size, VariableLengthVector, and Offset have nothing similar


Where I thought you were going to go was that the array-like objects derived from FixedArray, are simple arrogate classes. So we car refer to the buffer as just a pointer to it's components instead of the arrogate structure. So we could just get the efficiency of memcpy like performance for float to float elements for example.

Brad

On May 9, 2013, at 4:12 PM, "Williams, Norman K" <norman-k-williams at uiowa.edu> wrote:

> That's pretty sweet.
> 
> I wondered about this the other day:  You can't use std::copy directly on
> types that could in theory benefit from it.  For example:
> 
> #include <algorithm>
> #include <iostream>
> #include "itkVector.h"
> 
> int main(int argc, char *argv[])
> {
>  typedef itk::Vector<double, 32> VecType;
>  VecType a, b;
>  // this won't work
>  std::copy(a,a.end(),b);
>  // this works but is fugly
>  std::copy(a.GetDataPointer(),
>            a.GetDataPointer()+VecType::Dimension,
>            b.GetDataPointer());
> }
> 
> 
> Would it add unacceptable ugliness to support std::algorithm copy
> semantics?
> --
> Kent Williams norman-k-williams at uiowa.edu
> 
> 
> 
> 
> 
> 
> On 5/8/13 7:37 AM, "Bradley Lowekamp" <blowekamp at mail.nih.gov> wrote:
> 
>> 
>> 
>> 
>> Hello,
>> 
>> 
>> I know the CastImageFilter is no the most exciting filter but, it's
>> probably the most frequently used and should be as snappy as possible.
>> And the lessons learn here should be applicable to many other places.
>> 
>> 
>> I am clamming that I have speed this guy up by 5-25X from ITK v4.3.2 when
>> the following two patch are combined:
>> 
>> 
>> http://review.source.kitware.com/#/c/11174/
>> http://review.source.kitware.com/#/c/11163/
>> 
>> 
>> 
>> I have to thank Kent for motivating the use of using std:copy for
>> conversion to get most of this gain, in ImageAlgorithm::Copy.
>> 
>> 
>> The following performance was measure by converting short to float,  with
>> the following loop:
>> 
>> 
>> const unsigned int cnt = 5;
>> 
>> 
>> for ( unsigned int j = 0; j < 10; ++j )
>>   {
>>   t.Start();
>>   for ( unsigned int i = 0; i < cnt; ++i )
>>     {
>>     inImage1->Modified();
>>     f->Update();
>>     }
>>   t.Stop();
>>   }
>> 
>> 
>> std::cout << "\t+Add Average Time: " << t.GetMean()/cnt << t.GetUnit()
>> << std::endl;
>> 
>> 
>> 
>> 
>> 
>> I hope the follow table get formatted OK over e-mail:
>> 
>> 
>> gcc 4.4.74.3.2masterw/ std::copymaster speedupincremental speeduptotal
>> speedup[64, 64,
>> 64]0.00230210.00072730.00012913.165104825.63317275717.82958224[128, 128,
>> 128]0.01696210.00433210.00101023.9154090134.2882285416.79016867[256, 256,
>> 256]0.14184400.03067570.02385984.6239857611.2856645915.94489476[32,
>> 32]0.00001410.00000640.00000372.1828232231.7320512033.780761589[64,
>> 64]0.00003640.00001160.00000343.128445423.41794490910.6928541[128,
>> 128]0.00012510.00003190.00000773.9209226224.16303764816.32294849[256,
>> 256]0.00046880.00010400.00001904.509772995.46965017624.66688063gcc
>> 4.7.1[64, 64,
>> 64]0.001622900.00049710.00012433.2649981893.99942067713.05810127[128,
>> 128,
>> 128]0.011172400.00222670.00095295.0174247322.33688300211.72513457[256,
>> 256, 256]0.089189000.02281310.02360303.9095519680.9665339153.77871457[32,
>> 32]0.000011200.00000580.00000361.9167268711.6311584593.126485249[64,
>> 64]0.000026280.00000750.00000323.483567442.3718157528.262380129[128,
>> 128]0.000087600.00001400.00000746.2571732661.89174800111.83699502[256,
>> 256]0.000317160.00003170.000019110.005110391.65661205616.5745865icc
>> 13.0.1[64, 64,
>> 64]0.00187010.0012090.0001301621.5468320939.28842519314.36763418[128,
>> 128,
>> 128]0.01223670.0079420.0009915831.5408278398.00905219212.34057058[256,
>> 256, 256]0.09650670.0603450.01903721.5992387163.1698674175.069374698[32,
>> 32]0.00001140.0000093.28E-061.2467412812.7921503123.481089056[64,
>> 64]0.00003010.0000183.22E-061.7153699545.4444565279.339257142[128,
>> 128]0.00010300.0000629.04E-061.6631539176.85224533711.39633867[256,
>> 256]0.00039610.0002212.05E-051.79448405810.7557825119.30108025
> 
> 
> 
> ________________________________
> Notice: This UI Health Care e-mail (including attachments) is covered by the Electronic Communications Privacy Act, 18 U.S.C. 2510-2521, is confidential and may be legally privileged.  If you are not the intended recipient, you are hereby notified that any retention, dissemination, distribution, or copying of this communication is strictly prohibited.  Please reply to the sender that you have received the message in error, then delete it.  Thank you.
> ________________________________