VisionWorks Toolkit Reference

December 18, 2015 | 1.2 Release

 All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
CUDA Interoperability

Detailed Description

Adds CUDA interoperability support.

The VisionWorks framework can be used along with existing user CUDA code or with other CUDA libraries (like NPP, Thrust, etc.). The framework provides functionality to cooperate with existing CUDA code in 2 ways:

The following VisionWorks objects can be accessed in the CUDA address space:

The following VisionWorks objects can be imported from existing CUDA memory:

See also
CUDA Interoperability Tutorial

Enumerations

enum  nvx_accessor_e {
  NVX_READ_ONLY_CUDA = VX_ENUM_BASE(VX_ID_NVIDIA, VX_ENUM_ACCESSOR) + 0x0,
  NVX_WRITE_ONLY_CUDA = VX_ENUM_BASE(VX_ID_NVIDIA, VX_ENUM_ACCESSOR) + 0x1,
  NVX_READ_AND_WRITE_CUDA = VX_ENUM_BASE(VX_ID_NVIDIA, VX_ENUM_ACCESSOR) + 0x2
}
 Extended memory accessors. More...
 
enum  nvx_import_type_e { NVX_IMPORT_TYPE_CUDA = VX_ENUM_BASE(VX_ID_NVIDIA, VX_ENUM_IMPORT_MEM) + 0x0 }
 Extended import type. More...
 

Enumeration Type Documentation

Extended memory accessors.

Enumerator
NVX_READ_ONLY_CUDA 

RO with CUDA pointer.

NVX_WRITE_ONLY_CUDA 

WO with CUDA pointer.

NVX_READ_AND_WRITE_CUDA 

RW with CUDA pointer.

Definition at line 370 of file nvx.h.

Extended import type.

Enumerator
NVX_IMPORT_TYPE_CUDA 

The CUDA import memory type.

Definition at line 383 of file nvx.h.