Use Bazel to Compile Torch-TensorRT into Windows Binaries
Budget: $10 – $30 CAD
I am building a Windows program using Pytorch C++ (Libtorch). I want to use the Torch-TensorRT library, but it is only officially supported on Linux. Following some other peoples's instructions, I have actually managed to compile a torch_tensorrt.dll file (and turn that into a .lib), but that isn't quite what I need as it doesn't include the "core" and "torchscript" namespaces.
I need the .dll and .lib files for the entire Torch-TensorRT library, compiled successfully and working with Libtorch 1.10.1+cu113, as well as detailed instructions on how you got them so I can compile them myself for future updates.
Ideal candidate is familiar with using Bazel (as that's what is used to compile it), and experience with Libtorch, CUDA, and especially Torch-TensorRT would be a huge asset. For the right person, this should be a relatively easy gig.
Here is my Github issue on the topic, which references the other Github issue threads that I followed to get the torch_tensorrt.dll compiled: https://github.com/NVIDIA/Torch-TensorRT/issues/1014
I need the .dll and .lib files for the entire Torch-TensorRT library, compiled successfully and working with Libtorch 1.10.1+cu113, as well as detailed instructions on how you got them so I can compile them myself for future updates.
Ideal candidate is familiar with using Bazel (as that's what is used to compile it), and experience with Libtorch, CUDA, and especially Torch-TensorRT would be a huge asset. For the right person, this should be a relatively easy gig.
Here is my Github issue on the topic, which references the other Github issue threads that I followed to get the torch_tensorrt.dll compiled: https://github.com/NVIDIA/Torch-TensorRT/issues/1014