[ARCHIVED] Cooperative primitives for CUDA C++. See https://github.com/NVIDIA/cccl

Code supporting the RAL paper "Accelerating Robot Dynamics Gradients on a CPU, GPU, and FPGA"