NVIDIA Holoscan SDK v2.7.0
Holoscan v2.7.0

Class CudaAllocator

Base Type

Derived Types

class CudaAllocator : public holoscan::Allocator

Base class for all CUDA allocators.

CudaAllocators are allocators for CUDA memory that support asynchronous allocation.

Subclassed by holoscan::RMMAllocator, holoscan::StreamOrderedAllocator

Public Functions

template<typename ArgT, typename ...ArgsT, typename = std::enable_if_t<!std::is_base_of_v<::holoscan::Resource, std::decay_t<ArgT>> && (std::is_same_v<::holoscan::Arg, std::decay_t<ArgT>> || std::is_same_v<::holoscan::ArgList, std::decay_t<ArgT>>)>>
inline explicit CudaAllocator(ArgT &&arg, ArgsT&&... args)
CudaAllocator() = default
CudaAllocator(const std::string &name, nvidia::gxf::CudaAllocator *component)
inline virtual const char *gxf_typename() const override
nvidia::byte *allocate_async(uint64_t size, cudaStream_t stream)
void free_async(byte *pointer, cudaStream_t stream)
size_t pool_size(MemoryStorageType type) const
nvidia::gxf::CudaAllocator *get() const

Previous Class CsvDataExporter
Next Class CudaBufferAvailableCondition
© Copyright 2022-2024, NVIDIA. Last updated on Dec 2, 2024.