Report - A Survey of CPU-GPU Heterogeneous Computing Techniques
28 May 2015 Oakridge - Sparsh Mittal and Jeffrey S. Vetter from Oak Ridge National Laboratory published a "Survey of CPU-GPU Heterogeneous Computing Techniques". The report reviews heterogeneous computing approaches at runtime, algorithm, programming, compiler and application level.
Both discrete and fused CPU-GPU systems are discussed as are bench-mark suites designed for evaluating heterogeneous computing systems. The survey provides insights into working and scope of applications of heterogeneous computing approaches to researchers and allow them to further harness the computational powers of CPUs and GPUs to achieve the goal of exascale supercomputer performance.