public holoscan::CudaAllocator (Class CudaAllocator)

Class Documentation

class RMMAllocator : public holoscan::CudaAllocator

RMM (RAPIDS memory manager) allocator.

This is a memory pool which provides a user-specified number of equally sized blocks of memory.

This allocator supports simultaneous memory pools for CUDA device memory and pinned host memory.

Because it is a CudaAllocator it supports both synchronous (allocate, free) and asynchronous (allocate_async and free_async) APIs for memory allocation.

The values for the memory parameters, such as device_memory_initial_size must be specified in the form of a string containing a non-negative integer value followed by a suffix representing the units. Supported units are B, KB, MB, GB and TB where the values are powers of 1024 bytes (e.g., MB = 1024 * 1024 bytes). Examples of valid units are “512MB”, “256 KB”, “1 GB”. If a floating point number is specified that decimal portion will be truncated (i.e. the value is rounded down to the nearest integer).

==Parameters==

device_memory_initial_size (std::string, optional): The initial size of the device memory pool. See above for the format accepted. Defaults to “8MB” on aarch64 and “16MB” on x86_64.
device_memory_max_size (std::string, optional): The maximum size of the device memory pool. See above for the format accepted. The default is to use twice the value set for device_memory_initial_size.
host_memory_initial_size (std::string, optional): The initial size of the host memory pool. See above for the format accepted. Defaults to “8MB” on aarch64 and “16MB” on x86_64.
host_memory_max_size (std::string, optional): The maximum size of the host memory pool. See above for the format accepted. The default is to use twice the value set for device_memory_initial_size.
dev_id (int32_t, optional): The CUDA device id specifying which device the memory pool will use (Default: 0).

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 RMMAllocator(ArgT &&arg, ArgsT&&... args)

RMMAllocator() = default

inline RMMAllocator(const std::string &device_memory_initial_size, const std::string &device_memory_max_size, const std::string &host_memory_initial_size, const std::string &host_memory_max_size, int32_t dev_id = 0)

RMMAllocator(const std::string &name, nvidia::gxf::RMMAllocator *component)

inline virtual const char *gxf_typename() const override

virtual void setup(ComponentSpec &spec) override

Define the resource specification.

Parameters: spec – The reference to the component specification.

nvidia::gxf::RMMAllocator *get() const

Previous Class Resource

Next Class RuntimeError

Class RMMAllocator

Inheritance Relationships

Base Type

Class Documentation