libtrilinos-kokkos-dev 12.4.2-2 / usr / include / trilinos / Kokkos_CudaSpace.hpp

/*
//@HEADER
// ************************************************************************
// 
//                        Kokkos v. 2.0
//              Copyright (2014) Sandia Corporation
// 
// Under the terms of Contract DE-AC04-94AL85000 with Sandia Corporation,
// the U.S. Government retains certain rights in this software.
// 
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are
// met:
//
// 1. Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer.
//
// 2. Redistributions in binary form must reproduce the above copyright
// notice, this list of conditions and the following disclaimer in the
// documentation and/or other materials provided with the distribution.
//
// 3. Neither the name of the Corporation nor the names of the
// contributors may be used to endorse or promote products derived from
// this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY SANDIA CORPORATION "AS IS" AND ANY
// EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
// PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL SANDIA CORPORATION OR THE
// CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
// EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
// PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
// LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
// NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
// SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
//
// Questions? Contact  H. Carter Edwards (hcedwar@sandia.gov)
// 
// ************************************************************************
//@HEADER
*/

#ifndef KOKKOS_CUDASPACE_HPP
#define KOKKOS_CUDASPACE_HPP

#include <Kokkos_Core_fwd.hpp>

#if defined( KOKKOS_HAVE_CUDA )

#include <iosfwd>
#include <typeinfo>
#include <string>

#include <Kokkos_HostSpace.hpp>

#include <impl/Kokkos_AllocationTracker.hpp>

#include <Cuda/Kokkos_Cuda_abort.hpp>
#include <Cuda/Kokkos_Cuda_BasicAllocators.hpp>

/*--------------------------------------------------------------------------*/

namespace Kokkos {

/** \brief  Cuda on-device memory management */

class CudaSpace {
public:

  //! Tag this class as a kokkos memory space
  typedef CudaSpace             memory_space ;
  typedef Kokkos::Cuda          execution_space ;
  typedef Kokkos::Device<execution_space,memory_space> device_type;

  typedef unsigned int          size_type ;

  /*--------------------------------*/

#if ! defined( KOKKOS_USING_EXPERIMENTAL_VIEW )

  typedef Impl::CudaMallocAllocator allocator;

  /** \brief  Allocate a contiguous block of memory.
   *
   *  The input label is associated with the block of memory.
   *  The block of memory is tracked via reference counting where
   *  allocation gives it a reference count of one.
   */
  static Impl::AllocationTracker allocate_and_track( const std::string & label, const size_t size );

  /*--------------------------------*/
  /** \brief  Cuda specific function to attached texture object to an allocation.
   *          Output the texture object, base pointer, and offset from the input pointer.
   */
#if defined( __CUDACC__ )
  static void texture_object_attach(  Impl::AllocationTracker const & tracker
                                    , unsigned type_size
                                    , ::cudaChannelFormatDesc const & desc
                                   );
#endif

#endif /* #if ! defined( KOKKOS_USING_EXPERIMENTAL_VIEW ) */

  /*--------------------------------*/

  CudaSpace();
  CudaSpace( const CudaSpace & rhs ) = default ;
  CudaSpace & operator = ( const CudaSpace & rhs ) = default ;
  ~CudaSpace() = default ;

  /**\brief  Allocate untracked memory in the cuda space */
  void * allocate( const size_t arg_alloc_size ) const ;

  /**\brief  Deallocate untracked memory in the cuda space */
  void deallocate( void * const arg_alloc_ptr
                 , const size_t arg_alloc_size ) const ;

  /*--------------------------------*/
  /** \brief  Error reporting for HostSpace attempt to access CudaSpace */
  static void access_error();
  static void access_error( const void * const );

private:

  int  m_device ; ///< Which Cuda device

  // friend class Kokkos::Experimental::Impl::SharedAllocationRecord< Kokkos::CudaSpace , void > ;
};

namespace Impl {
/// \brief Initialize lock array for arbitrary size atomics.
///
/// Arbitrary atomics are implemented using a hash table of locks
/// where the hash value is derived from the address of the
/// object for which an atomic operation is performed.
/// This function initializes the locks to zero (unset).
void init_lock_array_cuda_space();

/// \brief Retrieve the pointer to the lock array for arbitrary size atomics.
///
/// Arbitrary atomics are implemented using a hash table of locks
/// where the hash value is derived from the address of the
/// object for which an atomic operation is performed.
/// This function retrieves the lock array pointer.
/// If the array is not yet allocated it will do so.
int* lock_array_cuda_space_ptr(bool deallocate = false);
}
} // namespace Kokkos

/*--------------------------------------------------------------------------*/
/*--------------------------------------------------------------------------*/

namespace Kokkos {

/** \brief  Cuda memory that is accessible to Host execution space
 *          through Cuda's unified virtual memory (UVM) runtime.
 */
class CudaUVMSpace {
public:

  //! Tag this class as a kokkos memory space
  typedef CudaUVMSpace          memory_space ;
  typedef Cuda                  execution_space ;
  typedef Kokkos::Device<execution_space,memory_space> device_type;
  typedef unsigned int          size_type ;

  /** \brief  If UVM capability is available */
  static bool available();

  /*--------------------------------*/

#if ! defined( KOKKOS_USING_EXPERIMENTAL_VIEW )

  typedef Impl::CudaUVMAllocator allocator;

  /** \brief  Allocate a contiguous block of memory.
   *
   *  The input label is associated with the block of memory.
   *  The block of memory is tracked via reference counting where
   *  allocation gives it a reference count of one.
   */
  static Impl::AllocationTracker allocate_and_track( const std::string & label, const size_t size );


  /** \brief  Cuda specific function to attached texture object to an allocation.
   *          Output the texture object, base pointer, and offset from the input pointer.
   */
#if defined( __CUDACC__ )
  static void texture_object_attach(  Impl::AllocationTracker const & tracker
                                    , unsigned type_size
                                    , ::cudaChannelFormatDesc const & desc
                                   );
#endif

#endif /* #if ! defined( KOKKOS_USING_EXPERIMENTAL_VIEW ) */

  /*--------------------------------*/

  CudaUVMSpace();
  CudaUVMSpace( const CudaUVMSpace & rhs ) = default ;
  CudaUVMSpace & operator = ( const CudaUVMSpace & rhs ) = default ;
  ~CudaUVMSpace() = default ;

  /**\brief  Allocate untracked memory in the cuda space */
  void * allocate( const size_t arg_alloc_size ) const ;

  /**\brief  Deallocate untracked memory in the cuda space */
  void deallocate( void * const arg_alloc_ptr
                 , const size_t arg_alloc_size ) const ;

  /*--------------------------------*/

private:

  int  m_device ; ///< Which Cuda device
};

} // namespace Kokkos

/*--------------------------------------------------------------------------*/
/*--------------------------------------------------------------------------*/

namespace Kokkos {

/** \brief  Host memory that is accessible to Cuda execution space
 *          through Cuda's host-pinned memory allocation.
 */
class CudaHostPinnedSpace {
public:

  //! Tag this class as a kokkos memory space
  /** \brief  Memory is in HostSpace so use the HostSpace::execution_space */
  typedef HostSpace::execution_space  execution_space ;
  typedef CudaHostPinnedSpace         memory_space ;
  typedef Kokkos::Device<execution_space,memory_space> device_type;
  typedef unsigned int                size_type ;

  /*--------------------------------*/

#if ! defined( KOKKOS_USING_EXPERIMENTAL_VIEW )

  typedef Impl::CudaHostAllocator allocator ;

  /** \brief  Allocate a contiguous block of memory.
   *
   *  The input label is associated with the block of memory.
   *  The block of memory is tracked via reference counting where
   *  allocation gives it a reference count of one.
   */
  static Impl::AllocationTracker allocate_and_track( const std::string & label, const size_t size );

#endif /* #if ! defined( KOKKOS_USING_EXPERIMENTAL_VIEW ) */

  /*--------------------------------*/

  CudaHostPinnedSpace();
  CudaHostPinnedSpace( const CudaHostPinnedSpace & rhs ) = default ;
  CudaHostPinnedSpace & operator = ( const CudaHostPinnedSpace & rhs ) = default ;
  ~CudaHostPinnedSpace() = default ;

  /**\brief  Allocate untracked memory in the space */
  void * allocate( const size_t arg_alloc_size ) const ;

  /**\brief  Deallocate untracked memory in the space */
  void deallocate( void * const arg_alloc_ptr
                 , const size_t arg_alloc_size ) const ;

  /*--------------------------------*/
};

} // namespace Kokkos

/*--------------------------------------------------------------------------*/
/*--------------------------------------------------------------------------*/

namespace Kokkos {
namespace Impl {

void DeepCopyAsyncCuda( void * dst , const void * src , size_t n);

template<> struct DeepCopy< CudaSpace , CudaSpace , Cuda>
{
  DeepCopy( void * dst , const void * src , size_t );
  DeepCopy( const Cuda & , void * dst , const void * src , size_t );
};

template<> struct DeepCopy< CudaSpace , HostSpace , Cuda >
{
  DeepCopy( void * dst , const void * src , size_t );
  DeepCopy( const Cuda & , void * dst , const void * src , size_t );
};

template<> struct DeepCopy< HostSpace , CudaSpace , Cuda >
{
  DeepCopy( void * dst , const void * src , size_t );
  DeepCopy( const Cuda & , void * dst , const void * src , size_t );
};

template<class ExecutionSpace> struct DeepCopy< CudaSpace , CudaSpace , ExecutionSpace >
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< CudaSpace , CudaSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};

template<class ExecutionSpace> struct DeepCopy< CudaSpace , HostSpace , ExecutionSpace >
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< CudaSpace , HostSpace , Cuda>( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};

template<class ExecutionSpace>
struct DeepCopy< HostSpace , CudaSpace , ExecutionSpace >
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< HostSpace , CudaSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};

template<class ExecutionSpace>
struct DeepCopy< CudaSpace , CudaUVMSpace , ExecutionSpace >
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< CudaSpace , CudaSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};

template<class ExecutionSpace>
struct DeepCopy< CudaSpace , CudaHostPinnedSpace , ExecutionSpace>
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< CudaSpace , HostSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};


template<class ExecutionSpace>
struct DeepCopy< CudaUVMSpace , CudaSpace , ExecutionSpace>
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< CudaSpace , CudaSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};

template<class ExecutionSpace>
struct DeepCopy< CudaUVMSpace , CudaUVMSpace , ExecutionSpace>
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< CudaSpace , CudaSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};

template<class ExecutionSpace>
struct DeepCopy< CudaUVMSpace , CudaHostPinnedSpace , ExecutionSpace>
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< CudaSpace , HostSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};

template<class ExecutionSpace> struct DeepCopy< CudaUVMSpace , HostSpace , ExecutionSpace >
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< CudaSpace , HostSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};


template<class ExecutionSpace> struct DeepCopy< CudaHostPinnedSpace , CudaSpace , ExecutionSpace >
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< HostSpace , CudaSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};

template<class ExecutionSpace> struct DeepCopy< CudaHostPinnedSpace , CudaUVMSpace , ExecutionSpace >
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< HostSpace , CudaSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};

template<class ExecutionSpace> struct DeepCopy< CudaHostPinnedSpace , CudaHostPinnedSpace , ExecutionSpace >
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< HostSpace , HostSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};

template<class ExecutionSpace> struct DeepCopy< CudaHostPinnedSpace , HostSpace , ExecutionSpace >
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< HostSpace , HostSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};


template<class ExecutionSpace> struct DeepCopy< HostSpace , CudaUVMSpace , ExecutionSpace >
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< HostSpace , CudaSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};

template<class ExecutionSpace> struct DeepCopy< HostSpace , CudaHostPinnedSpace , ExecutionSpace >
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< HostSpace , HostSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};

} // namespace Impl
} // namespace Kokkos

//----------------------------------------------------------------------------
//----------------------------------------------------------------------------

namespace Kokkos {
namespace Impl {

/** Running in CudaSpace attempting to access HostSpace: error */
template<>
struct VerifyExecutionCanAccessMemorySpace< Kokkos::CudaSpace , Kokkos::HostSpace >
{
  enum { value = false };
  KOKKOS_INLINE_FUNCTION static void verify( void )
    { Kokkos::abort("Cuda code attempted to access HostSpace memory"); }

  KOKKOS_INLINE_FUNCTION static void verify( const void * )
    { Kokkos::abort("Cuda code attempted to access HostSpace memory"); }
};

/** Running in CudaSpace accessing CudaUVMSpace: ok */
template<>
struct VerifyExecutionCanAccessMemorySpace< Kokkos::CudaSpace , Kokkos::CudaUVMSpace >
{
  enum { value = true };
  KOKKOS_INLINE_FUNCTION static void verify( void ) { }
  KOKKOS_INLINE_FUNCTION static void verify( const void * ) { }
};

/** Running in CudaSpace accessing CudaHostPinnedSpace: ok */
template<>
struct VerifyExecutionCanAccessMemorySpace< Kokkos::CudaSpace , Kokkos::CudaHostPinnedSpace >
{
  enum { value = true };
  KOKKOS_INLINE_FUNCTION static void verify( void ) { }
  KOKKOS_INLINE_FUNCTION static void verify( const void * ) { }
};

/** Running in CudaSpace attempting to access an unknown space: error */
template< class OtherSpace >
struct VerifyExecutionCanAccessMemorySpace<
  typename enable_if< ! is_same<Kokkos::CudaSpace,OtherSpace>::value , Kokkos::CudaSpace >::type ,
  OtherSpace >
{
  enum { value = false };
  KOKKOS_INLINE_FUNCTION static void verify( void )
    { Kokkos::abort("Cuda code attempted to access unknown Space memory"); }

  KOKKOS_INLINE_FUNCTION static void verify( const void * )
    { Kokkos::abort("Cuda code attempted to access unknown Space memory"); }
};

//----------------------------------------------------------------------------
/** Running in HostSpace attempting to access CudaSpace */
template<>
struct VerifyExecutionCanAccessMemorySpace< Kokkos::HostSpace , Kokkos::CudaSpace >
{
  enum { value = false };
  inline static void verify( void ) { CudaSpace::access_error(); }
  inline static void verify( const void * p ) { CudaSpace::access_error(p); }
};

/** Running in HostSpace accessing CudaUVMSpace is OK */
template<>
struct VerifyExecutionCanAccessMemorySpace< Kokkos::HostSpace , Kokkos::CudaUVMSpace >
{
  enum { value = true };
  inline static void verify( void ) { }
  inline static void verify( const void * ) { }
};

/** Running in HostSpace accessing CudaHostPinnedSpace is OK */
template<>
struct VerifyExecutionCanAccessMemorySpace< Kokkos::HostSpace , Kokkos::CudaHostPinnedSpace >
{
  enum { value = true };
  KOKKOS_INLINE_FUNCTION static void verify( void ) {}
  KOKKOS_INLINE_FUNCTION static void verify( const void * ) {}
};

} // namespace Impl
} // namespace Kokkos

//----------------------------------------------------------------------------
//----------------------------------------------------------------------------

namespace Kokkos {
namespace Experimental {
namespace Impl {

template<>
class SharedAllocationRecord< Kokkos::CudaSpace , void >
  : public SharedAllocationRecord< void , void >
{
private:

  friend class SharedAllocationRecord< Kokkos::CudaUVMSpace , void > ;

  typedef SharedAllocationRecord< void , void >  RecordBase ;

  SharedAllocationRecord( const SharedAllocationRecord & ) = delete ;
  SharedAllocationRecord & operator = ( const SharedAllocationRecord & ) = delete ;

  static void deallocate( RecordBase * );

  static ::cudaTextureObject_t
  attach_texture_object( const unsigned sizeof_alias
                       , void * const   alloc_ptr
                       , const size_t   alloc_size ); 

  static RecordBase s_root_record ;

  ::cudaTextureObject_t   m_tex_obj ;
  const Kokkos::CudaSpace m_space ;

protected:

  ~SharedAllocationRecord();
  SharedAllocationRecord() : RecordBase(), m_tex_obj(0), m_space() {}

  SharedAllocationRecord( const Kokkos::CudaSpace        & arg_space
                        , const std::string              & arg_label
                        , const size_t                     arg_alloc_size
                        , const RecordBase::function_type  arg_dealloc = & deallocate
                        );

public:

  std::string get_label() const ;

  static SharedAllocationRecord * allocate( const Kokkos::CudaSpace &  arg_space
                                          , const std::string       &  arg_label
                                          , const size_t               arg_alloc_size );

  /**\brief  Allocate tracked memory in the space */
  static
  void * allocate_tracked( const Kokkos::CudaSpace & arg_space
                         , const std::string & arg_label
                         , const size_t arg_alloc_size );

  /**\brief  Reallocate tracked memory in the space */
  static
  void * reallocate_tracked( void * const arg_alloc_ptr
                           , const size_t arg_alloc_size );

  /**\brief  Deallocate tracked memory in the space */
  static
  void deallocate_tracked( void * const arg_alloc_ptr );

  static SharedAllocationRecord * get_record( void * arg_alloc_ptr );

  template< typename AliasType >
  inline
  ::cudaTextureObject_t attach_texture_object()
    {
      static_assert( ( std::is_same< AliasType , int >::value ||
                       std::is_same< AliasType , ::int2 >::value ||
                       std::is_same< AliasType , ::int4 >::value )
                   , "Cuda texture fetch only supported for alias types of int, ::int2, or ::int4" );

      if ( m_tex_obj == 0 ) {
        m_tex_obj = attach_texture_object( sizeof(AliasType)
                                         , (void*) RecordBase::m_alloc_ptr
                                         , RecordBase::m_alloc_size );
      }

      return m_tex_obj ;
    }

  template< typename AliasType >
  inline
  int attach_texture_object_offset( const AliasType * const ptr )
    {
      // Texture object is attached to the entire allocation range
      return ptr - reinterpret_cast<AliasType*>( RecordBase::m_alloc_ptr );
    }

  static void print_records( std::ostream & , const Kokkos::CudaSpace & , bool detail = false );
};


template<>
class SharedAllocationRecord< Kokkos::CudaUVMSpace , void >
  : public SharedAllocationRecord< void , void >
{
private:

  typedef SharedAllocationRecord< void , void >  RecordBase ;

  SharedAllocationRecord( const SharedAllocationRecord & ) = delete ;
  SharedAllocationRecord & operator = ( const SharedAllocationRecord & ) = delete ;

  static void deallocate( RecordBase * );

  static RecordBase s_root_record ;

  ::cudaTextureObject_t      m_tex_obj ;
  const Kokkos::CudaUVMSpace m_space ;

protected:

  ~SharedAllocationRecord();
  SharedAllocationRecord() : RecordBase(), m_tex_obj(0), m_space() {}

  SharedAllocationRecord( const Kokkos::CudaUVMSpace     & arg_space
                        , const std::string              & arg_label
                        , const size_t                     arg_alloc_size
                        , const RecordBase::function_type  arg_dealloc = & deallocate
                        );

public:

  std::string get_label() const ;

  static SharedAllocationRecord * allocate( const Kokkos::CudaUVMSpace &  arg_space
                                          , const std::string          &  arg_label
                                          , const size_t                  arg_alloc_size
                                          );

  /**\brief  Allocate tracked memory in the space */
  static
  void * allocate_tracked( const Kokkos::CudaUVMSpace & arg_space
                         , const std::string & arg_label
                         , const size_t arg_alloc_size );

  /**\brief  Reallocate tracked memory in the space */
  static
  void * reallocate_tracked( void * const arg_alloc_ptr
                           , const size_t arg_alloc_size );

  /**\brief  Deallocate tracked memory in the space */
  static
  void deallocate_tracked( void * const arg_alloc_ptr );

  static SharedAllocationRecord * get_record( void * arg_alloc_ptr );


  template< typename AliasType >
  inline
  ::cudaTextureObject_t attach_texture_object()
    {
      static_assert( ( std::is_same< AliasType , int >::value ||
                       std::is_same< AliasType , ::int2 >::value ||
                       std::is_same< AliasType , ::int4 >::value )
                   , "Cuda texture fetch only supported for alias types of int, ::int2, or ::int4" );

      if ( m_tex_obj == 0 ) {
        m_tex_obj = SharedAllocationRecord< Kokkos::CudaSpace , void >::
          attach_texture_object( sizeof(AliasType)
                               , (void*) RecordBase::m_alloc_ptr
                               , RecordBase::m_alloc_size );
      }

      return m_tex_obj ;
    }

  template< typename AliasType >
  inline
  int attach_texture_object_offset( const AliasType * const ptr )
    {
      // Texture object is attached to the entire allocation range
      return ptr - reinterpret_cast<AliasType*>( RecordBase::m_alloc_ptr );
    }

  static void print_records( std::ostream & , const Kokkos::CudaUVMSpace & , bool detail = false );
};

template<>
class SharedAllocationRecord< Kokkos::CudaHostPinnedSpace , void >
  : public SharedAllocationRecord< void , void >
{
private:

  typedef SharedAllocationRecord< void , void >  RecordBase ;

  SharedAllocationRecord( const SharedAllocationRecord & ) = delete ;
  SharedAllocationRecord & operator = ( const SharedAllocationRecord & ) = delete ;

  static void deallocate( RecordBase * );

  static RecordBase s_root_record ;

  const Kokkos::CudaHostPinnedSpace m_space ;

protected:

  ~SharedAllocationRecord();
  SharedAllocationRecord() : RecordBase(), m_space() {}

  SharedAllocationRecord( const Kokkos::CudaHostPinnedSpace     & arg_space
                        , const std::string              & arg_label
                        , const size_t                     arg_alloc_size
                        , const RecordBase::function_type  arg_dealloc = & deallocate
                        );

public:

  std::string get_label() const ;

  static SharedAllocationRecord * allocate( const Kokkos::CudaHostPinnedSpace &  arg_space
                                          , const std::string          &  arg_label
                                          , const size_t                  arg_alloc_size
                                          );
  /**\brief  Allocate tracked memory in the space */
  static
  void * allocate_tracked( const Kokkos::CudaHostPinnedSpace & arg_space
                         , const std::string & arg_label
                         , const size_t arg_alloc_size );

  /**\brief  Reallocate tracked memory in the space */
  static
  void * reallocate_tracked( void * const arg_alloc_ptr
                           , const size_t arg_alloc_size );

  /**\brief  Deallocate tracked memory in the space */
  static
  void deallocate_tracked( void * const arg_alloc_ptr );


  static SharedAllocationRecord * get_record( void * arg_alloc_ptr );

  static void print_records( std::ostream & , const Kokkos::CudaHostPinnedSpace & , bool detail = false );
};

} // namespace Impl
} // namespace Experimental
} // namespace Kokkos

//----------------------------------------------------------------------------
//----------------------------------------------------------------------------

#endif /* #if defined( KOKKOS_HAVE_CUDA ) */
#endif /* #define KOKKOS_CUDASPACE_HPP */