2026-01-28 06:16:04 +00:00

35 lines
1001 B
C++

/**
* \brief An implementation of {@see AbstractInferencer} using Microsoft CUDA ONNX Runtime to do inference
* \author Nan Zhou, nanzhou at kneron dot us
* \copyright Kneron Inc. All right reserved.
*/
#ifndef PIANO_DYNASTY_FLOATING_POINT_INCLUDE_MSFT_GPU_MSFTINFERENCER_H_
#define PIANO_DYNASTY_FLOATING_POINT_INCLUDE_MSFT_GPU_MSFTINFERENCER_H_
#include <memory>
#include <string>
#include <limits.h>
#include "msft/MSFTInferencer.h"
#include "msft-gpu/InferencerPImpl.h"
namespace dynasty {
namespace inferencer {
namespace msftgpu {
class InferencerImpl : public dynasty::inferencer::msftgpu::Inferencer<float> {
friend class dynasty::inferencer::msftgpu::Inferencer<float>::Builder;
explicit InferencerImpl(std::shared_ptr<Ort::Env> enviroment, std::shared_ptr<Ort::Session> session) :
msftgpu::Inferencer<float>(enviroment, session) {}
public:
~InferencerImpl() override = default;
};
}
}
}
#endif //PIANO_DYNASTY_FLOATING_POINT_INCLUDE_MSFT_GPU_MSFTINFERENCER_H_