CUDA utilities. More...

#include <iostream>
#include <cuda.h>
#include <cuda_runtime.h>
#include "../parameter.h"
#include <assert.h>

Include dependency graph for cuda_utilities.cuh:

This graph shows which files directly or indirectly include this file:

Namespaces
namespace	CudaUtils

namespace	CudaUtils::Kernel

namespace	CudaUtils::Kernel::Launch

namespace	cuda

namespace	cuda::math
	Device/CUDA math functions.

Macros
#define	MILUPHPC_CUDAUTILITIES_CUH

#define	CUDA_CALLABLE_MEMBER

#define	safeCudaCall(call) checkCudaCall(call, #call, __FILE__, __LINE__)
	check CUDA call More...

#define	gpuErrorcheck(ans) { gpuAssert((ans), __FILE__, __LINE__); }
	check CUDA call More...

#define	cudaAssert(...)

#define	cudaTerminate(...)

Functions
void	gpuAssert (cudaError_t code, const char *file, int line, bool abort=true)
	Check CUDA error codes. More...

void	checkCudaCall (cudaError_t command, const char commandName, const char fileName, int line)
	Check CUDA call. More...

__global__ void	CudaUtils::Kernel::collectValues (integer indices, real entries, real *collector, integer count)

__global__ void	CudaUtils::Kernel::checkValues (integer indices, real entry1, real entry2, real entry3, integer count)

template<typename T >
__global__ void	CudaUtils::Kernel::findDuplicates (T array, integer duplicateCounter, int length)

template<typename T >
__global__ void	CudaUtils::Kernel::findDuplicateEntries (T array1, T array2, integer *duplicateCounter, int length)

template<typename T >
__global__ void	CudaUtils::Kernel::findDuplicateEntries (T array1, T array2, T array3, integer duplicateCounter, int length)

template<typename T , typename U >
__global__ void	CudaUtils::Kernel::findDuplicates (T array, U entry1, U entry2, integer duplicateCounter, int length)

template<typename T >
__global__ void	CudaUtils::Kernel::markDuplicates (T array, integer duplicateCounter, int length)

template<typename T , typename U >
__global__ void	CudaUtils::Kernel::markDuplicates (T array, U entry1, U entry2, U entry3, integer duplicateCounter, integer child, int length)

template<typename T >
__global__ void	CudaUtils::Kernel::removeDuplicates (T array, T removedArray, integer *duplicateCounter, int length)

real	CudaUtils::Kernel::Launch::collectValues (integer indices, real entries, real *collector, integer count)

real	CudaUtils::Kernel::Launch::checkValues (integer indices, real entry1, real entry2, real entry3, integer count)

template<typename T >
real	CudaUtils::Kernel::Launch::findDuplicates (T array, integer duplicateCounter, int length)

template<typename T >
real	CudaUtils::Kernel::Launch::findDuplicateEntries (T array1, T array2, integer *duplicateCounter, int length)

template<typename T >
real	CudaUtils::Kernel::Launch::findDuplicateEntries (T array1, T array2, T array3, integer duplicateCounter, int length)

template<typename T , typename U >
real	CudaUtils::Kernel::Launch::findDuplicates (T array, U entry1, U entry2, integer duplicateCounter, int length)

template<typename T >
real	CudaUtils::Kernel::Launch::markDuplicates (T array, integer duplicateCounter, int length)

template<typename T , typename U >
real	CudaUtils::Kernel::Launch::markDuplicates (T array, U entry1, U entry2, U entry3, integer duplicateCounter, integer child, int length)

template<typename T >
real	CudaUtils::Kernel::Launch::removeDuplicates (T array, T removedArray, integer *duplicateCounter, int length)

__device__ real	cuda::math::min (real a, real b)
	Minimum value out of two floating point values. More...

__device__ real	cuda::math::min (real a, real b, real c)
	Minimum value out of three floating point values. More...

__device__ real	cuda::math::max (real a, real b)
	Maximum value out of two floating point values. More...

__device__ real	cuda::math::max (real a, real b, real c)
	Maximum value out of three floating point values. More...

__device__ real	cuda::math::abs (real a)
	Absolute value of a floating point value. More...

__device__ real	cuda::math::sqrt (real a)
	Square root of a floating point value. More...

__device__ real	cuda::math::rsqrt (real a)
	Inverse square root of a floating point value. More...

Detailed Description

CUDA utilities.

CUDA utilities like:

preprocessor directives
preprocessor macros
CUDA math functions
general CUDA kernels like
- finding duplicates
- copying values from device to device
- ...

Author: Michael Staneker

Bug:: no

Definition in file cuda_utilities.cuh.

Macro Definition Documentation

◆ CUDA_CALLABLE_MEMBER

#define CUDA_CALLABLE_MEMBER

Definition at line 30 of file cuda_utilities.cuh.

◆ cudaAssert

#define cudaAssert ( ... )

Value:

    {    \
    printf(__VA_ARGS__);     \
}

Somewhat of an assertion, printing warning and/or terminating in dependence of SAFETY_LEVEL

Definition at line 50 of file cuda_utilities.cuh.

◆ cudaTerminate

#define cudaTerminate ( ... )

Value:

    { \
    printf(__VA_ARGS__);     \
    assert(0);               \
}

Terminate from within CUDA kernel using assert(0)

Definition at line 70 of file cuda_utilities.cuh.

◆ gpuErrorcheck

#define gpuErrorcheck ( ans ) { gpuAssert((ans), __FILE__, __LINE__); }

check CUDA call

Definition at line 41 of file cuda_utilities.cuh.

◆ MILUPHPC_CUDAUTILITIES_CUH

#define MILUPHPC_CUDAUTILITIES_CUH

Definition at line 19 of file cuda_utilities.cuh.

◆ safeCudaCall

#define safeCudaCall ( call ) checkCudaCall(call, #call, __FILE__, __LINE__)

check CUDA call

Definition at line 36 of file cuda_utilities.cuh.

Function Documentation

◆ checkCudaCall()

void checkCudaCall	(	cudaError_t	command,
		const char *	commandName,
		const char *	fileName,
		int	line
	)

Check CUDA call.

Parameters

command
commandName
fileName
line

Definition at line 30 of file cuda_utilities.cu.

◆ gpuAssert()

void gpuAssert	(	cudaError_t	code,
		const char *	file,
		int	line,
		bool	abort = `true`
	)

Check CUDA error codes.