Table of Contents

Enum ManagedMemoryOptions

Namespace
DotCompute.Backends.CUDA.Types
Assembly
DotCompute.Abstractions.dll

Flags for CUDA managed memory allocation and behavior.

[Flags]
public enum ManagedMemoryOptions

Fields

EnablePrefetch = 32

Enable prefetching optimizations.

MultiDevice = 16

Memory is shared across multiple devices.

None = 0

Default managed memory behavior.

PreferDevice = 1

Memory should be initially resident on the device.

PreferHost = 2

Memory should be initially resident on the host.

ReadMostly = 4

Enable read-mostly optimization.

SingleDevice = 8

Memory is accessed by a single device.

TrackAccess = 64

Track access patterns for optimization.