Mila::Dnn::Compute::CpuMemoryResource Class Reference

CPU memory resource for host-accessible memory allocation. More...

Inheritance diagram for Mila::Dnn::Compute::CpuMemoryResource:

Public Member Functions
	CpuMemoryResource (int device_id=0)
	Constructs CPU memory resource.
Public Member Functions inherited from Mila::Dnn::Compute::MemoryResource
virtual	~MemoryResource ()=default
	Virtual destructor for proper cleanup of derived classes.

Static Public Attributes
static constexpr DeviceType	device_type = DeviceType::Cpu
static constexpr bool	is_device_accessible = false
	Indicates CPU memory is not accessible from device code.
static constexpr bool	is_host_accessible = HostAccessible::is_host_accessible
	Indicates CPU memory is accessible from host code.
Static Public Attributes inherited from Mila::Dnn::Compute::MemoryResource
static constexpr bool	is_device_accessible = false
	Checks if the memory is accessible from device code.
static constexpr bool	is_host_accessible = true
	Checks if the memory is accessible from host code.

Protected Member Functions
void *	do_allocate (std::size_t bytes, std::size_t alignment) override
	Allocates aligned CPU memory.
void	do_deallocate (void *ptr, std::size_t, std::size_t) override
	Deallocates CPU memory using platform-specific functions.
bool	do_is_equal (const std::pmr::memory_resource &other) const noexcept override
	Compares CPU memory resources for equality.

Detailed Description

CPU memory resource for host-accessible memory allocation.

Provides optimized CPU memory allocation with proper alignment support for tensor data. CPU memory allocation doesn't require device selection, so the device_id parameter is unused but maintained for interface consistency with other memory resource types.

Constructor & Destructor Documentation

CpuMemoryResource()

Mila::Dnn::Compute::CpuMemoryResource::CpuMemoryResource ( int device_id = 0 )

inlineexplicit

Constructs CPU memory resource.

Parameters

device_id

Device identifier (unused for CPU, maintained for interface consistency)

Note

CPU memory allocation doesn't require device selection

Parameter is kept for consistency with CUDA memory resources

Member Function Documentation

do_allocate()

void * Mila::Dnn::Compute::CpuMemoryResource::do_allocate ( std::size_t bytes, std::size_t alignment )

inlineoverrideprotectedvirtual

Allocates aligned CPU memory.

Uses platform-specific aligned allocation functions to ensure proper memory alignment for optimal CPU performance and SIMD operations.

Parameters

bytes	Number of bytes to allocate
alignment	Memory alignment requirement

Returns

Pointer to allocated memory

Exceptions

std::bad_alloc

If allocation fails

Implements Mila::Dnn::Compute::MemoryResource.

do_deallocate()

void Mila::Dnn::Compute::CpuMemoryResource::do_deallocate ( void * ptr, std::size_t , std::size_t  )

inlineoverrideprotectedvirtual

Deallocates CPU memory using platform-specific functions.

Properly releases memory allocated with platform-specific aligned allocation functions.

Parameters

ptr	Pointer to memory to deallocate

The size and alignment arguments are unused (kept for the memory-resource interface) and therefore intentionally unnamed.

Implements Mila::Dnn::Compute::MemoryResource.

do_is_equal()

bool Mila::Dnn::Compute::CpuMemoryResource::do_is_equal ( const std::pmr::memory_resource & other ) const

inlineoverrideprotectedvirtualnoexcept

Compares CPU memory resources for equality.

CPU memory resources are equal if they are both CpuMemoryResource instances, since they all manage the same underlying host memory pool.

Parameters

other

The other memory resource to compare with

Returns

true if both are CpuMemoryResource instances

Implements Mila::Dnn::Compute::MemoryResource.

---

The documentation for this class was generated from the following file:

• Mila/Src/Dnn/Compute/Devices/Cpu/CpuMemoryResource.ixx