Cooperative Kernels: GPU Multitasking for Blocking Algorithms Tyler SorensenImperial College London London, UKt.sorensen15@imperial.ac.ukHugues Evrard Imperial College LondonLondon, UK h.evrard@imperial.ac.ukAlastair F. Donaldson Imperial College LondonLondon, UK alastair.donaldson@imperial.ac.ukABSTRACT There is growing interest in accelerating irregular data-parallel algorithms on GPUs. These algorithms are typically blocking, so they require fair scheduling. But GPU programming models (e.g. O