Delegate TensorRtGpuAllocatorHandler

Namespace

JYPPX.TensorRtSharp

Assembly

JYPPX.TensorRtSharp.dll

Handles copied GPU allocator metadata and approves or rejects allocation/reallocation requests. 处理复制后的 GPU allocator 元数据，并批准或拒绝分配、重分配请求。

public delegate bool TensorRtGpuAllocatorHandler(TensorRtGpuAllocatorCallbackRequest request)

Parameters

request TensorRtGpuAllocatorCallbackRequest

The copied pointer-free request. 复制后的无指针请求。

Returns

bool

true to permit allocation or reallocation. 返回 true 允许分配或重分配。

Remarks

The return value is observational for release callbacks: native code always attempts to release owner-managed memory. 对释放回调，返回值仅用于观察；native 代码始终尝试释放 owner 管理的显存，避免托管异常造成泄漏。