Skip to content
GitLab
Commit d281e78e authored by Christian Schulte zu Berge's avatar Christian Schulte zu Berge
Browse files

closes Issue #42: Completely removed ViennaCL dependency from RandomWalksLib.

parent 9fe8a51f
Expand all Hide whitespace changes
Inline Side-by-side
modules/randomwalk/ext/RandomWalksLib/CMakeLists.txt
View file @ d281e78e
......@@ -33,29 +33,13 @@ SET(RandomWalksLibSources ConfidenceMaps2D.cpp
SparseSolverEigenLLT.cpp
SparseSolverFactory.cpp)
# Make OpenCL dependency optional
IF(OPENCL_FOUND)
ADD_DEFINITIONS("-DRANDOMWALKSLIB_HAS_OPENCL")
LIST(APPEND RandomWalksLibExternalLibs ${OPENCL_LIBRARY})
LIST(APPEND RandomWalksLibHeaders SparseSolverViennaCPU.h SparseSolverViennaGPU.h)
LIST(APPEND RandomWalksLibSources SparseSolverViennaCPU.cpp SparseSolverViennaGPU.cpp)
ENDIF()
################################################################################
# define library target
################################################################################
# Eigen RandomWalksLibHeaders are already in CAMPVis/ext
INCLUDE_DIRECTORIES(${CampvisGlobalIncludeDirs})
#include ViennaCl RandomWalksLibHeaders
INCLUDE_DIRECTORIES(${CMAKE_CURRENT_SOURCE_DIR})
ADD_LIBRARY(RandomWalksLib STATIC ${RandomWalksLibSources} ${RandomWalksLibHeaders})
# We don't want this within CAMPVis...
#put debug target to x64/Debug and all other configurations to x64/Release
#SET_TARGET_PROPERTIES( RandomWalksLib PROPERTIES
# ARCHIVE_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/x64/Release #static libs are archives
# ARCHIVE_OUTPUT_DIRECTORY_DEBUG ${CMAKE_BINARY_DIR}/x64/Debug
# ARCHIVE_OUTPUT_NAME RandomWalksLib
# ARCHIVE_OUTPUT_NAME_DEBUG RandomWalksLibd ) #debug build should end with a 'd'
TARGET_LINK_LIBRARIES(RandomWalksLib ${RandomWalksLibExternalLibs})
modules/randomwalk/ext/RandomWalksLib/SparseSolverFactory.cpp
View file @ d281e78e
......@@ -35,18 +35,6 @@ SparseSolverInterface * SparseSolverFactory::createSolver(std::string type, int
SparseSolverInterface * solver = new SparseSolverEigenCustom(iterations,tolerance);
return solver;
}
#ifdef RANDOMWALKSLIB_HAS_OPENCL
else if(type.compare("Vienna-CG-CPU")==0)
{
SparseSolverInterface * solver = new SparseSolverViennaCPU(iterations,tolerance);
return solver;
}
else if(type.compare("Vienna-CG-GPU")==0)
{
SparseSolverInterface * solver = new SparseSolverViennaGPU(iterations,tolerance);
return solver;
}
#endif
else
{
return new SparseSolverEigenLLT();
......
modules/randomwalk/ext/RandomWalksLib/SparseSolverViennaCPU.cpp deleted 100644 → 0
View file @ 9fe8a51f
#include "SparseSolverViennaCPU.h"
#define VIENNACL_HAVE_EIGEN
#include <viennacl\scalar.hpp>
#include <viennacl\matrix.hpp>
#include <viennacl\compressed_matrix.hpp>
#include <viennacl\linalg\cg.hpp>
SparseSolverViennaCPU::SparseSolverViennaCPU(int iterations, double tolerance)
{
this->iterations = iterations;
this->tolerance = tolerance;
}
std::vector<double> SparseSolverViennaCPU::solve_Ax_b(SparseMatrix<double> A, SparseVector<double> b, int numel, std::vector<int> & uidx, const std::vector<int> * labels, const std::vector<int> * seeds, int active_label)
{
SparseMatrix<double> A_sparse_matrix(A);
VectorXd b_dense = b;
viennacl::linalg::cg_tag custom_tag(tolerance, iterations);
VectorXd x_dense = viennacl::linalg::solve(A_sparse_matrix, b_dense, custom_tag); // CPU Solver with dense
std::vector<double> xmat(numel);
for (int i=0; i<x_dense.rows(); i++)
{
xmat[uidx[i]] = x_dense(i);
}
for (size_t i=0; i<seeds->size(); i++)
{
if((*labels)[i] == active_label)
xmat[(*seeds)[i]] = 1.0;
else
xmat[(*seeds)[i]] = 0.0;
}
return xmat;
}
\ No newline at end of file
modules/randomwalk/ext/RandomWalksLib/SparseSolverViennaCPU.h deleted 100644 → 0
View file @ 9fe8a51f
/* This Source Code Form is subject to the terms of the Mozilla Public
* License, v. 2.0. If a copy of the MPL was not distributed with this
* file, You can obtain one at http://mozilla.org/MPL/2.0/. */
#ifndef SPARSE_SOLVER_VIENNA_CPU_H__
#define SPARSE_SOLVER_VIENNA_CPU_H__
#include "SparseSolverInterface.h"
/** \brief Conjugate Gradient Vienna solver on CPU for random walks system
* \author Athanasios Karamalis
* \date 11.06.2012
*/
class SparseSolverViennaCPU : public SparseSolverInterface
{
public:
SparseSolverViennaCPU(int iterations, double tolerance);
/// Solver with CG
virtual std::vector<double> solve_Ax_b(SparseMatrix<double> A, SparseVector<double> b, int numel, std::vector<int> & uidx, const std::vector<int> * labels, const std::vector<int> * seeds, int active_label);
private:
int iterations; ///< CG iterations
double tolerance; ///< CG tolerance
};
#endif
\ No newline at end of file
modules/randomwalk/ext/RandomWalksLib/SparseSolverViennaGPU.cpp deleted 100644 → 0
View file @ 9fe8a51f
#include "SparseSolverViennaGPU.h"
#define VIENNACL_HAVE_EIGEN
#include <viennacl\scalar.hpp>
#include <viennacl\matrix.hpp>
#include <viennacl\compressed_matrix.hpp>
#include <viennacl\linalg\cg.hpp>
SparseSolverViennaGPU::SparseSolverViennaGPU(int iterations, double tolerance)
{
this->iterations = iterations;
this->tolerance = tolerance;
}
std::vector<double> SparseSolverViennaGPU::solve_Ax_b(SparseMatrix<double> A, SparseVector<double> b, int numel, std::vector<int> & uidx, const std::vector<int> * labels, const std::vector<int> * seeds, int active_label)
{
SparseMatrix<double> A_sparse_matrix(A);
VectorXd b_dense = b;
viennacl::linalg::cg_tag custom_tag(tolerance, iterations);
viennacl::compressed_matrix<double> viennacl_A(A_sparse_matrix.rows(), A_sparse_matrix.cols());
viennacl::vector<double> viennacl_b(b_dense.rows());
viennacl::vector<double> viennacl_x(b_dense.rows());
viennacl::copy(A_sparse_matrix, viennacl_A);
viennacl::copy(b_dense, viennacl_b);
viennacl_x = viennacl::linalg::solve(viennacl_A, viennacl_b, custom_tag); // GPU Solver
VectorXd x_dense(b.rows());
viennacl::copy(viennacl_x, x_dense);
std::vector<double> xmat(numel);
for (int i=0; i<x_dense.rows(); i++)
{
xmat[uidx[i]] = x_dense(i);
}
for (size_t i=0; i<seeds->size(); i++)
{
if((*labels)[i] == active_label)
xmat[(*seeds)[i]] = 1.0;
else
xmat[(*seeds)[i]] = 0.0;
}
return xmat;
}
\ No newline at end of file
modules/randomwalk/ext/RandomWalksLib/SparseSolverViennaGPU.h deleted 100644 → 0
View file @ 9fe8a51f
/* This Source Code Form is subject to the terms of the Mozilla Public
* License, v. 2.0. If a copy of the MPL was not distributed with this
* file, You can obtain one at http://mozilla.org/MPL/2.0/. */
#ifndef SPRASE_SOLVER_VIENNA_GPU_H__
#define SPRASE_SOLVER_VIENNA_GPU_H__
#include "SparseSolverInterface.h"
/** \brief Conjugate Gradient Vienna solver on GPU for random walks system
* \author Athanasios Karamalis
* \date 11.06.2012
*/
class SparseSolverViennaGPU : public SparseSolverInterface
{
public:
SparseSolverViennaGPU(int iterations, double tolerance);
/// Solver with CG
virtual std::vector<double> solve_Ax_b(SparseMatrix<double> A, SparseVector<double> b, int numel, std::vector<int> & uidx, const std::vector<int> * labels, const std::vector<int> * seeds, int active_label);
private:
int iterations; ///< CG iterations
double tolerance; ///< CG tolerance
};
#endif
\ No newline at end of file
modules/randomwalk/ext/RandomWalksLib/viennacl/backend/cpu_ram.hpp deleted 100644 → 0
View file @ 9fe8a51f
#ifndef VIENNACL_BACKEND_CPU_RAM_HPP_
#define VIENNACL_BACKEND_CPU_RAM_HPP_
/* =========================================================================
Copyright (c) 2010-2012, Institute for Microelectronics,
Institute for Analysis and Scientific Computing,
TU Wien.
Portions of this software are copyright by UChicago Argonne, LLC.
-----------------
ViennaCL - The Vienna Computing Library
-----------------
Project Head: Karl Rupp rupp@iue.tuwien.ac.at
(A list of authors and contributors can be found in the PDF manual)
License: MIT (X11), see file LICENSE in the base directory
============================================================================= */
/** @file viennacl/backend/cpu_ram.hpp
@brief Implementations for the OpenCL backend functionality
*/
#include <vector>
#include "viennacl/tools/shared_ptr.hpp"
namespace viennacl
{
namespace backend
{
namespace cpu_ram
{
typedef viennacl::tools::shared_ptr<char> handle_type;
// Requirements for backend:
// * memory_create(size, host_ptr)
// * memory_copy(src, dest, offset_src, offset_dest, size)
// * memory_write_from_main_memory(src, offset, size,
// dest, offset, size)
// * memory_read_to_main_memory(src, offset, size
// dest, offset, size)
// *
//
namespace detail
{
template<class U>
struct array_deleter
{
void operator()(U* p) const { delete[] p; }
};
}
inline handle_type memory_create(std::size_t size_in_bytes, const void * host_ptr = NULL)
{
if (!host_ptr)
return handle_type(new char[size_in_bytes], detail::array_deleter<char>());
handle_type new_handle(new char[size_in_bytes], detail::array_deleter<char>());
// copy data:
char * raw_ptr = new_handle.get();
const char * data_ptr = static_cast<const char *>(host_ptr);
for (std::size_t i=0; i<size_in_bytes; ++i)
raw_ptr[i] = data_ptr[i];
return new_handle;
}
inline void memory_copy(handle_type const & src_buffer,
handle_type & dst_buffer,
std::size_t src_offset,
std::size_t dst_offset,
std::size_t bytes_to_copy)
{
assert( (dst_buffer.get() != NULL) && bool("Memory not initialized!"));
assert( (src_buffer.get() != NULL) && bool("Memory not initialized!"));
for (std::size_t i=0; i<bytes_to_copy; ++i)
dst_buffer.get()[i+dst_offset] = src_buffer.get()[i + src_offset];
}
inline void memory_write(handle_type & dst_buffer,
std::size_t dst_offset,
std::size_t bytes_to_copy,
const void * ptr)
{
assert( (dst_buffer.get() != NULL) && bool("Memory not initialized!"));
for (std::size_t i=0; i<bytes_to_copy; ++i)
dst_buffer.get()[i+dst_offset] = static_cast<const char *>(ptr)[i];
}
inline void memory_read(handle_type const & src_buffer,
std::size_t src_offset,
std::size_t bytes_to_copy,
void * ptr)
{
assert( (src_buffer.get() != NULL) && bool("Memory not initialized!"));
for (std::size_t i=0; i<bytes_to_copy; ++i)
static_cast<char *>(ptr)[i] = src_buffer.get()[i+src_offset];
}
}
} //backend
} //viennacl
#endif
modules/randomwalk/ext/RandomWalksLib/viennacl/backend/cuda.hpp deleted 100644 → 0
View file @ 9fe8a51f
#ifndef VIENNACL_BACKEND_CUDA_HPP_
#define VIENNACL_BACKEND_CUDA_HPP_
/* =========================================================================
Copyright (c) 2010-2012, Institute for Microelectronics,
Institute for Analysis and Scientific Computing,
TU Wien.
Portions of this software are copyright by UChicago Argonne, LLC.
-----------------
ViennaCL - The Vienna Computing Library
-----------------
Project Head: Karl Rupp rupp@iue.tuwien.ac.at
(A list of authors and contributors can be found in the PDF manual)
License: MIT (X11), see file LICENSE in the base directory
============================================================================= */
/** @file viennacl/backend/cuda.hpp
@brief Implementations for the CUDA backend functionality
*/
#include <iostream>
#include <vector>
#include "viennacl/tools/shared_ptr.hpp"
// includes CUDA
#include <cuda_runtime.h>
#define VIENNACL_CUDA_ERROR_CHECK(err) detail::cuda_error_check (err, __FILE__, __LINE__)
namespace viennacl
{
namespace backend
{
namespace cuda
{
typedef viennacl::tools::shared_ptr<char> handle_type;
// Requirements for backend:
// * memory_create(size, host_ptr)
// * memory_copy(src, dest, offset_src, offset_dest, size)
// * memory_write_from_main_memory(src, offset, size,
// dest, offset, size)
// * memory_read_to_main_memory(src, offset, size
// dest, offset, size)
// *
//
namespace detail
{
inline void cuda_error_check(cudaError error_code, const char *file, const int line )
{
if(cudaSuccess != error_code)
{
std::cerr << file << "(" << line << "): " << ": CUDA Runtime API error " << error_code << ": " << cudaGetErrorString( error_code ) << std::endl;
throw "CUDA error";
}
}
template <typename U>
struct cuda_deleter
{
void operator()(U * p) const
{
//std::cout << "Freeing handle " << reinterpret_cast<void *>(p) << std::endl;
cudaFree(p);
}
};
}
inline handle_type memory_create(std::size_t size_in_bytes, const void * host_ptr = NULL)
{
void * dev_ptr = NULL;
VIENNACL_CUDA_ERROR_CHECK( cudaMalloc(&dev_ptr, size_in_bytes) );
//std::cout << "Allocated new dev_ptr " << dev_ptr << " of size " << size_in_bytes << std::endl;
if (!host_ptr)
return handle_type(reinterpret_cast<char *>(dev_ptr), detail::cuda_deleter<char>());
handle_type new_handle(reinterpret_cast<char*>(dev_ptr), detail::cuda_deleter<char>());
// copy data:
//std::cout << "Filling new handle from host_ptr " << host_ptr << std::endl;
cudaMemcpy(new_handle.get(), host_ptr, size_in_bytes, cudaMemcpyHostToDevice);
return new_handle;
}
inline void memory_copy(handle_type const & src_buffer,
handle_type & dst_buffer,
std::size_t src_offset,
std::size_t dst_offset,
std::size_t bytes_to_copy)
{
assert( (dst_buffer.get() != NULL) && bool("Memory not initialized!"));
assert( (src_buffer.get() != NULL) && bool("Memory not initialized!"));
cudaMemcpy(reinterpret_cast<void *>(dst_buffer.get() + dst_offset),
reinterpret_cast<void *>(src_buffer.get() + src_offset),
bytes_to_copy,
cudaMemcpyDeviceToDevice);
}
inline void memory_write(handle_type & dst_buffer,
std::size_t dst_offset,
std::size_t bytes_to_copy,
const void * ptr)
{
assert( (dst_buffer.get() != NULL) && bool("Memory not initialized!"));
cudaMemcpy(reinterpret_cast<char *>(dst_buffer.get()) + dst_offset,
reinterpret_cast<const char *>(ptr),
bytes_to_copy,
cudaMemcpyHostToDevice);
}
inline void memory_read(handle_type const & src_buffer,
std::size_t src_offset,
std::size_t bytes_to_copy,
void * ptr)
{
assert( (src_buffer.get() != NULL) && bool("Memory not initialized!"));
cudaMemcpy(reinterpret_cast<char *>(ptr),
reinterpret_cast<char *>(src_buffer.get()) + src_offset,
bytes_to_copy,
cudaMemcpyDeviceToHost);
}
} //cuda
} //backend
} //viennacl
#endif
modules/randomwalk/ext/RandomWalksLib/viennacl/backend/mem_handle.hpp deleted 100644 → 0
View file @ 9fe8a51f
#ifndef VIENNACL_BACKEND_MEM_HANDLE_HPP
#define VIENNACL_BACKEND_MEM_HANDLE_HPP
/* =========================================================================
Copyright (c) 2010-2012, Institute for Microelectronics,
Institute for Analysis and Scientific Computing,
TU Wien.
Portions of this software are copyright by UChicago Argonne, LLC.
-----------------
ViennaCL - The Vienna Computing Library
-----------------
Project Head: Karl Rupp rupp@iue.tuwien.ac.at
(A list of authors and contributors can be found in the PDF manual)
License: MIT (X11), see file LICENSE in the base directory
============================================================================= */
/** @file viennacl/backend/mem_handle.hpp
@brief Implements the multi-memory-domain handle
*/
#include <vector>
#include <cassert>
#include "viennacl/forwards.h"
#include "viennacl/tools/shared_ptr.hpp"
#include "viennacl/backend/cpu_ram.hpp"
#ifdef VIENNACL_WITH_OPENCL
#include "viennacl/backend/opencl.hpp"
#endif
#ifdef VIENNACL_WITH_CUDA
#include "viennacl/backend/cuda.hpp"
#endif
namespace viennacl
{
namespace backend
{
// if a user compiles with CUDA, it is reasonable to expect that CUDA should be the default
#ifdef VIENNACL_WITH_CUDA
inline memory_types default_memory_type() { return CUDA_MEMORY; }
#elif defined(VIENNACL_WITH_OPENCL)
inline memory_types default_memory_type() { return OPENCL_MEMORY; }
#else
inline memory_types default_memory_type() { return MAIN_MEMORY; }
#endif
class mem_handle
{
public:
typedef viennacl::tools::shared_ptr<char> ram_handle_type;
typedef viennacl::tools::shared_ptr<char> cuda_handle_type;
mem_handle() : active_handle_(MEMORY_NOT_INITIALIZED), size_in_bytes_(0) {}
ram_handle_type & ram_handle() { return ram_handle_; }
ram_handle_type const & ram_handle() const { return ram_handle_; }
#ifdef VIENNACL_WITH_OPENCL
viennacl::ocl::handle<cl_mem> & opencl_handle() { return opencl_handle_; }
viennacl::ocl::handle<cl_mem> const & opencl_handle() const { return opencl_handle_; }
#endif
#ifdef VIENNACL_WITH_CUDA
cuda_handle_type & cuda_handle() { return cuda_handle_; }
cuda_handle_type const & cuda_handle() const { return cuda_handle_; }
#endif
memory_types get_active_handle_id() const { return active_handle_; }
void switch_active_handle_id(memory_types new_id)
{
if (new_id != active_handle_)
{
if (active_handle_ == MEMORY_NOT_INITIALIZED)
active_handle_ = new_id;
else if (active_handle_ == MAIN_MEMORY)
{
active_handle_ = new_id;
}
else if (active_handle_ == OPENCL_MEMORY)
{
#ifdef VIENNACL_WITH_OPENCL
active_handle_ = new_id;
#else
throw "compiled without OpenCL suppport!";
#endif
}
else if (active_handle_ == CUDA_MEMORY)
{
#ifdef VIENNACL_WITH_CUDA
active_handle_ = new_id;
#else
throw "compiled without CUDA suppport!";
#endif
}
else
throw "invalid new memory region!";
}
}
bool operator==(mem_handle const & other) const
{
if (active_handle_ != other.active_handle_)
return false;
switch (active_handle_)
{
case MAIN_MEMORY:
return ram_handle_.get() == other.ram_handle_.get();
#ifdef VIENNACL_WITH_OPENCL
case OPENCL_MEMORY:
return opencl_handle_.get() == other.opencl_handle_.get();
#endif
#ifdef VIENNACL_WITH_CUDA
case CUDA_MEMORY:
return cuda_handle_.get() == other.cuda_handle_.get();
#endif
default: break;
}