segmented_scan_cta.cuh File Reference

CUDPP CTA-level scan routines. More...

#include "cta/scan_cta.cuh"
#include <cudpp_globals.h>
#include <math.h>
#include <cstdio>

Classes
class	SegmentedScanTraits< T, Oper, backward, exclusivity, doShiftFlags, fullBlock, sums, sm12OrBetter >
	Template class containing compile-time parameters to the segmented scan functions. More...

Functions
Segmented scan Functions
template<class T , typename traits >
__device__ void	loadForSegmentedScanSharedChunkFromMem4 (T *s_odata, T threadScan0[4], T threadScan1[4], unsigned int &threadFlag, unsigned int *s_oflags, unsigned int *s_oindices, const T *d_idata, const unsigned int *d_iflags, int numElements, int iDataOffset, int &ai, int &bi, int &aiDev, int &biDev)
	Handles loading input s_data from global memory to shared memory (vec4 version) More...
template<class T , class traits >
__device__ void	storeForSegmentedScanSharedChunkToMem4 (T *d_odata, T threadScan0[4], T threadScan1[4], unsigned int threadFlag, T *s_idata, unsigned int numElements, int oDataOffset, int ai, int bi, int aiDev, int biDev)
	Handles storing result s_data from shared memory to global memory (vec4 version) More...
template<class T , class traits , unsigned int blockSize>
__device__ T	reduceCTA (volatile T *s_data)
template<class T , class traits , bool isExclusive, unsigned int log_simd_threads>
__device__ void	warpSegScan (T val, unsigned int flag, volatile T *s_data, volatile unsigned int *s_flags, T &oVal, unsigned int &oFlag, bool print=false)
template<class T , class traits >
__device__ void	segmentedScanWarps (T val1, unsigned int flag1, T val2, unsigned int flag2, T *s_data, unsigned int *s_flags)
template<class T , class traits >
__device__ void	segmentedScanCTA (T *s_data, unsigned int *s_flags, unsigned int *s_indices, T *d_blockSums=0, unsigned int *d_blockFlags=0, unsigned int *d_blockIndices=0)
	CTA-level segmented scan routine;. More...

Detailed Description

CUDPP CTA-level scan routines.

segmented_scan_cta.cu