I noticed a topic brought up in 2015 about this. The device version of this function is not available in CUDA 7 (in 2015).
It’s 2022 now, and we already have CUDA 11.6. May I know why there is still no device path available for this particular function? Is LU factorisation faster when run on CPU instead of GPU?
I am currently dealing with circuit matrices and I’m trying to parallelise LU linear solver using GPU. And while other linear solvers have a GPU version, LU still doesn’t. And I was wondering if it is possible to run LU solver on GPU? And if it does, do I need to do the algorithm myself?