Expand description
Raw FFI + dynamic loader for NVIDIA TensorRT (C API surface).
TensorRT’s native public API is C++; NVIDIA ships a partial C-ABI surface
suitable for language bindings in NvInferRuntimeCAPI.h (TRT 10+). This
crate wraps that surface for runtime deserialization and inference. The
builder side of TensorRT remains C++-only; use the TRT trtexec tool or
the Python bindings to produce serialized engines, then load them here.
Structs§
- Tensor
Rt - trtDims_
t - trtStatus_
t - TensorRT C API returns
bool(0/1) orint32_tstatus codes depending on the function. We provide a thintrtStatus_tnewtype for the error-reporting subset so it implementsCudaStatus.
Enums§
Constants§
- TRT_
MAX_ DIMS - Analog of
nvinfer1::Dims— up to 8 dimensions.
Functions§
Type Aliases§
- PFN_
context Enqueue V3 - PFN_
context GetTensor Address - PFN_
context GetTensor Shape - PFN_
context SetInput Shape - PFN_
context SetTensor Address - PFN_
create Infer Runtime - PFN_
deserialize Cuda Engine - PFN_
destroy Cuda Engine - PFN_
destroy Execution Context - PFN_
destroy Infer Runtime - PFN_
engine Create Execution Context - PFN_
engine Create Execution Context With Strategy - PFN_
engine GetIO Tensor Name - PFN_
engine GetName - PFN_
engine GetNbIO Tensors - PFN_
engine GetNb Optimization Profiles - PFN_
engine GetTensor Bytes PerComponent - PFN_
engine GetTensor Data Type - PFN_
engine GetTensorIO Mode - PFN_
engine GetTensor Shape - PFN_
engine Serialize - PFN_
getInfer LibVersion - PFN_
host Memory Data - PFN_
host Memory Destroy - PFN_
host Memory Size - trtI
Cuda Engine_ t - trtI
Execution Context_ t - trtI
Host Memory_ t - trtI
Logger_ t - trtI
Plugin Registry_ t - trtI
Runtime_ t - trtLog
Callback - Logger callback signature (matches
nvinfer1::ILogger::log).