Interface class representing AI/DL inference results to be consumed by NVIDIA IndeX. More...
#include <iinference_result.h>
Inherits mi::base::Interface_declare< 0xd4374014, ... >.
Public Member Functions | |
virtual void | bind_inference_result (mi::neuraylib::Tag_struct xac_program, mi::Uint32 slot_id, nv::index::ICuda_memory_buffer *inference_results)=0 |
Binds the results of the inference operation to an XAC program's slot. More... | |
Interface class representing AI/DL inference results to be consumed by NVIDIA IndeX.
|
pure virtual |
Binds the results of the inference operation to an XAC program's slot.
The inference operation shall write into a CUDA buffer which is wrapped by an instance of the class ICuda_memory_buffer
. This buffer shall be passed back to NVIDIA IndeX compute and rendering system and shall be bind to a specific XAC program. In the context of NVIDIA IndeX's Advanced Computing (XAC), uniform data, or user buffers are bound to an XAC program using slots, which represent unique identifier. These enable an XAC program to retrieve the uniform data or user buffer and make use of them inside the XAC program. While the NVIDIA IndeX system transparently binds the CUDA buffers, the user or application writer is responsible for appropriate use of the CUDA buffers inside the XAC program, e.g., memory buffer overflows are not managed by NVIDIA IndeX but the user has to take care of those.
[out] | xac_program | The XAC program that shall make use of the inference results. The inference will be bound to this program and the program is responsible to grab and apply the inference results when rendering or computing the dataset. |
[out] | slot_id | The unique identifier that binds the inference results to an XAC program. The slot id needs to be available in the XAC program so that the program can make use of the bound data. |
[out] | inference_results | The CUDA memory buffer containing the inference results to be bound to the XAC program. |