CRTP base supplying the synchronous half of the resource concept in terms of the stream-ordered half, so a custom stream-ordered resource only has to write allocate_async and deallocate_async. More...

#include <nanovdb/cuda/DeviceResource.h>

Public Member Functions
void *	allocate (size_t bytes, size_t alignment)
	Allocates `bytes` usable on any stream when this returns.
void	deallocate (void *p, size_t bytes, size_t alignment)
	Frees `p` on the null stream.

Detailed Description

template<class Derived>
struct nanovdb::cuda::SyncFromAsync< Derived >

CRTP base supplying the synchronous half of the resource concept in terms of the stream-ordered half, so a custom stream-ordered resource only has to write allocate_async and deallocate_async.

Template Parameters

Derived the resource deriving from this base

A stream-ordered resource must also model the synchronous concept (is_async_resource implies is_resource), which means writing four methods where two would do. The synchronous pair is not a bare delegate: memory from allocate must be usable immediately on any stream, so the null-stream allocation has to be synchronized before it is returned. Omitting that synchronization yields memory that satisfies the concept but is not actually synchronous – a race rather than a compile error – so it lives here rather than being rewritten per resource.

struct MyResource : nanovdb::cuda::SyncFromAsync<MyResource> {
    static constexpr size_t DEFAULT_ALIGNMENT = 256;
    void* allocate_async(size_t bytes, size_t alignment, cudaStream_t stream);
    void  deallocate_async(void* p, size_t bytes, size_t alignment, cudaStream_t stream);
};

Member Function Documentation

◆ allocate()

template<class Derived>

void * allocate	(	size_t	bytes,
		size_t	alignment )

inline

Allocates bytes usable on any stream when this returns.

Parameters

bytes	number of bytes to allocate
alignment	requested alignment

Note: Every call synchronizes the null stream; on hot paths prefer the stream-ordered pair.

◆ deallocate()