TensorRT
Also known as: NVIDIA TensorRT
TensorRT is NVIDIA’s toolkit for optimizing and serving deep-learning models on its GPUs.
- Compiles a model into a fast, tuned engine.
- Heavy on kernel fusion and precision control.
- A common target for maximum GPU throughput.
Max speed on NVIDIA: when you want the fastest possible inference on an NVIDIA GPU, TensorRT is the optimization layer you reach for.