Table of Contents

Class CudaManagedMemory

Namespace
JYPPX.CudaSharp
Assembly
JYPPX.CudaSharp.dll

Managed wrapper around a CUDA unified-memory allocation. CUDA unified memory 分配的托管封装。

public sealed class CudaManagedMemory : CudaMemory, IDisposable
Inheritance
CudaManagedMemory
Implements
Inherited Members

Constructors

CudaManagedMemory(int, CudaManagedMemoryAttachmentFlags)

Allocates CUDA managed memory with the requested attachment flags. 使用指定的附着标志分配 CUDA managed memory。

public CudaManagedMemory(int sizeInBytes, CudaManagedMemoryAttachmentFlags flags = CudaManagedMemoryAttachmentFlags.Global)

Parameters

sizeInBytes int

The allocation size in bytes. 分配大小，单位为字节。

flags CudaManagedMemoryAttachmentFlags

The managed-memory attachment flags. managed memory 附着标志。

Properties

AttachmentFlags

Gets the managed-memory attachment flags used for the allocation. 获取该分配使用的 managed memory 附着标志。

public CudaManagedMemoryAttachmentFlags AttachmentFlags { get; }

Property Value

CudaManagedMemoryAttachmentFlags

Methods

Advise(CudaMemoryAdvice, CudaMemoryLocation)

Applies CUDA memory advice to this entire managed allocation at a strongly typed location. 对强类型位置上的整个 managed memory 分配应用 CUDA memory advice。

public void Advise(CudaMemoryAdvice advice, CudaMemoryLocation location)

Parameters

advice CudaMemoryAdvice
location CudaMemoryLocation

Advise(int, int, CudaMemoryAdvice, CudaMemoryLocation)

Applies CUDA memory advice to a managed-memory range at a strongly typed location. 对强类型位置上的 managed memory 范围应用 CUDA memory advice。

public void Advise(int offset, int count, CudaMemoryAdvice advice, CudaMemoryLocation location)

Parameters

offset int

The byte offset within this allocation. 当前分配内的字节偏移。

count int

The number of bytes covered by the advice. advice 覆盖的字节数。

advice CudaMemoryAdvice

The CUDA memory advice. CUDA 内存建议。

location CudaMemoryLocation

The CUDA memory location associated with the advice. 与 advice 关联的 CUDA memory 位置。

PrefetchAsync(CudaMemoryLocation, CudaStream)

Asynchronously prefetches this entire managed allocation to a strongly typed CUDA location. 异步将整个 managed memory 分配预取到强类型 CUDA 位置。

public void PrefetchAsync(CudaMemoryLocation location, CudaStream stream)

Parameters

location CudaMemoryLocation
stream CudaStream

PrefetchAsync(int, int, CudaMemoryLocation, CudaStream)

Asynchronously prefetches a managed-memory range to a strongly typed CUDA location. 异步将 managed memory 范围预取到强类型 CUDA 位置。

public void PrefetchAsync(int offset, int count, CudaMemoryLocation location, CudaStream stream)

Parameters

offset int

The byte offset within this allocation. 当前分配内的字节偏移。

count int

The number of bytes to prefetch. 要预取的字节数。

location CudaMemoryLocation

The destination CUDA memory location. 目标 CUDA memory 位置。

stream CudaStream

The CUDA stream that orders the prefetch. 用于排序预取操作的 CUDA stream。

Remarks

Keep this allocation and stream alive until the stream has synchronized. 在 stream 完成同步前，调用方必须保持此分配和 stream 存活。

Edit this page