Laikh Tewari

@peri044 can you help fill in the instructions for the export flow?

@gs-olive would you ever want to set `construct_live=False` in the compile path? It sounds like this feature reduces device memory pressure between compilation and execution at the cost of added...

Looks like it, I have tensorrt==8.6.1.post1 Shouldn't that be installed automatically as a dependency when I pip install torch_tensorrt?

https://developer.nvidia.com/blog/cuda-pro-tip-the-fast-way-to-query-device-properties/ --- This check was added that likely caused perf issue: https://github.com/pytorch/TensorRT/blob/bf4474dc7816c184489d3985ce892315f5e0cc42/core/runtime/runtime.cpp#L81 This check invokes a constructor for a TensorRT wrapper object RTDevice::RTDevice https://github.com/pytorch/TensorRT/blob/bf4474dc7816c184489d3985ce892315f5e0cc42/core/runtime/RTDevice.cpp#L16 And this is invoking cudaGetDeviceProperties which...