JoinHashTableGpuUtils.h File Reference

#include "DataMgr/Allocators/DeviceAllocator.h"
#include "QueryEngine/GpuMemUtils.h"

Include dependency graph for JoinHashTableGpuUtils.h:

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

Go to the source code of this file.

Functions
template<class T >
T *	transfer_vector_of_flat_objects_to_gpu (const std::vector< T > &vec, DeviceAllocator &allocator)
template<class T >
T *	transfer_flat_object_to_gpu (const T &object, DeviceAllocator &allocator)

Function Documentation

template<class T >

T* transfer_flat_object_to_gpu	(	const T &	object,
		DeviceAllocator &	allocator
	)

Definition at line 35 of file JoinHashTableGpuUtils.h.

References Allocator::alloc(), DeviceAllocator::copyToDevice(), and heavydb.dtypes::T.

Referenced by RangeJoinHashTable::approximateTupleCount(), BoundingBoxIntersectJoinHashTable::approximateTupleCount(), BaselineJoinHashTable::approximateTupleCount(), anonymous_namespace{BoundingBoxIntersectJoinHashTable.cpp}::compute_bucket_sizes(), and BaselineJoinHashTableBuilder::initHashTableOnGpu().

                                                                            {
  static_assert(std::is_standard_layout<T>::value,
                "Transferring an object to GPU only works for standard layout elements");
  const auto bytes = sizeof(T);
  auto gpu_ptr = allocator.alloc(bytes);
  allocator.copyToDevice(gpu_ptr, reinterpret_cast<const int8_t*>(&object), bytes);
  return reinterpret_cast<T*>(gpu_ptr);
}

Here is the call graph for this function:

Here is the caller graph for this function:

template<class T >

T* transfer_vector_of_flat_objects_to_gpu	(	const std::vector< T > &	vec,
		DeviceAllocator &	allocator
	)

Definition at line 24 of file JoinHashTableGpuUtils.h.

References Allocator::alloc(), DeviceAllocator::copyToDevice(), and heavydb.dtypes::T.

Referenced by RangeJoinHashTable::approximateTupleCount(), BoundingBoxIntersectJoinHashTable::approximateTupleCount(), BaselineJoinHashTable::approximateTupleCount(), anonymous_namespace{BoundingBoxIntersectJoinHashTable.cpp}::compute_bucket_sizes(), and BaselineJoinHashTable::initHashTableForDevice().

                                                                      {
  static_assert(std::is_trivially_copyable<T>::value && std::is_standard_layout<T>::value,
                "Transferring a vector to GPU only works for flat object elements");
  const auto vec_bytes = vec.size() * sizeof(T);
  auto gpu_vec = allocator.alloc(vec_bytes);
  allocator.copyToDevice(gpu_vec, reinterpret_cast<const int8_t*>(vec.data()), vec_bytes);
  return reinterpret_cast<T*>(gpu_vec);
}

Here is the call graph for this function:

Here is the caller graph for this function: