gpufftw.h

基于GPU进行快速科学计算

/******************************************************************************\

  Copyright 2005 The University of North Carolina at Chapel Hill.
  All Rights Reserved.

  Permission to use, copy, modify and distribute this software and its
  documentation for educational, research and non-profit purposes, without
  fee, and without a written agreement is hereby granted, provided that the
  above copyright notice and the following three paragraphs appear in all
  copies. Any use in a commercial organization requires a separate license.

  IN NO EVENT SHALL THE UNIVERSITY OF NORTH CAROLINA AT CHAPEL HILL BE LIABLE
  TO ANY PARTY FOR DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL
  DAMAGES, INCLUDING LOST PROFITS, ARISING OUT OF THE USE OF THIS SOFTWARE AND
  ITS DOCUMENTATION, EVEN IF THE UNIVERSITY OF NORTH CAROLINA HAVE BEEN
  ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.


  Permission to use, copy, modify and distribute this software and its
  documentation for educational, research and non-profit purposes, without
  fee, and without a written agreement is hereby granted, provided that the
  above copyright notice and the following three paragraphs appear in all
  copies.

  THE UNIVERSITY OF NORTH CAROLINA SPECIFICALLY DISCLAIM ANY WARRANTIES,
  INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
  FITNESS FOR A PARTICULAR PURPOSE.  THE SOFTWARE PROVIDED HEREUNDER IS ON AN
  "AS IS" BASIS, AND THE UNIVERSITY OF NORTH CAROLINA HAS NO OBLIGATION TO
  PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.


   ---------------------------------
  |Please send all BUG REPORTS to:  |
  |                                 |
  |   [e-mail]               |
  |                                 |
   ---------------------------------


  The authors may be contacted via:

  US Mail:         N. Govindaraju or D. Manocha
                       Department of Computer Science
                       Sitterson Hall, CB #3175
                       University of North Carolina
                       Chapel Hill, NC 27599-3175

	Add functions manycgpufft1d and manyscgpufft1d

	by			Simon Potvin and Jerome Genest, July 2006
				Centre d'optique, photonique et laser (COPL)
				Universite Laval
				Quebec, Canada
	Mail:		[e-mail]
				

\*****************************************************************************/

#include <arbfprog.h>
enum FFT_ERROR {FFT_SUCCESS, FFT_ARRAY_TOO_BIG, FFT_ARRAY_TOO_SMALL, FFT_NON_POWER_OF_TWO};


inline int comparePtr(const void *i1, const void *i2);

class GPUFFTW{
public: //methods
	//ctor 
	/*
	 * Parameters:
	 */
	GPUFFTW();

	//dtor
	~GPUFFTW();

	/* Returns max array size that can be handled depending on video RAM */
	int maxSize(){ 
		return(MAX_SIZE); 
	}

	
	/* Sorts an input array of (key,pointer) tuples on the GPU. 
	 * WARNING: The contents of the array will be overwritten with the sorted result.
	 *
	 * Parameters: 
	 *   array: array of tuples to sort
	 *   size: length of array
	 * Return value:
	 *   error code. FFT_SUCCESS is returned in case of no error.
	 *   Use errorToString() to print formatted string of error code.
	 */
	FFT_ERROR cgpufft1d(float *array, float  *ptrs, int size, int sign, int isreal=0);
	FFT_ERROR scgpufft1d(float *real,  int size, int sign);

	FFT_ERROR setFFTParams(int size);

	/*Compute many fft
	* WARNING: The contents of real and imag must be transpose. For example, if you have two real array(fft_1 and fft_2) of 16 points,
	* the contents of real is going to be { fft_1[1], fft_2[1], fft_1[2], fft_2[2], fft_1[3], fft_2[3], ... , fft_1[16], fft2_[16] }.
	*
	* Parameters:
	*    arraywidth : length of fft(must be power of 2)
	*    arrayheight : numbers of fft(must be 1 or power of 2)
	* Return value:
	*   error code. FFT_SUCCESS is returned in case of no error.
	*   Use errorToString() to print formatted string of error code.
	*/
	FFT_ERROR manycgpufft1d(float *real, float  *imag, int arraywidth, int arrayheight, int sign, int isreal);
	FFT_ERROR manyscgpufft1d(float *real, int arraywidth, int arrayheight, int sign);

	/* Upload float data to Video Memory.
	 *
	 * Parameters:
	 *   array: Input data. All the required parameters not explicit here are supposed
	 *          to have been set by a preceeding call to setSortParams(). 
	 * Return Value:
	 *   None
	 *
	 * Assumption: All the pointers are 32-bit (float * and void *)
	 *
	 */
	void uploadData(void *array);

	
	/* Read data back from GPU to CPU memory
	 *
	 * Parameters:
	 *   array: an array of scalars or (key,pointer) tuples depending
	 *          on whether tuples was set to FALSE or TRUE in the call
	 *			to setSortParams()
	 * Return value:
	 *   None
	 *
	 */
	void readbackData(void *array);	

	/* Print a formatted error string for errors returned from sort() */
	char *FFTErrorString(FFT_ERROR error);
	FFT_ERROR cGPUFFTW2d(float *real, float  *imag, int arraywidth, int arrayheight, int sign, int isreal);
	void CheckErrors();

private: //methods	
	inline void PingPong();
	void PingPong2to4();
	void PingPong4to2();
	void cgpufft1d( float sign, int usedfor2dfft=0 );
	
private: //data
	int W, H, MAX_SIZE, MIN_SIZE;
	int Width, Height, N, LOGN;
	unsigned int fb;
	unsigned int textureid[6];//last two as temporary buffers


	ARBFProg copyfp;
	ARBFProg fullscreenrowwidthfp;
	ARBFProg fftstages_twicerowwidth_to_n_fp;
	ARBFProg fftstages3_to_rowwidthfp;
	ARBFProg fullscreenfft1ststage;
	ARBFProg fullscreenfft2ndstagefp;
	int TILE_SIZE;	
	int source, target;
};