Autotuning Numerical Dense Linear Algebra for Batched Computation With GPU Hardware Accelerators | IEEE Journals & Magazine | IEEE Xplore