Optimization of Fusion Kernels on Accelerators with Indirect or Strided Memory Access Patterns

Publication
IEEE Transactions on Parallel and Distributed Systems