GPU computing is the use of graphics processing units (GPU) and central processing units (CPU) to perform scientific calculations, engineering computation, and the processing of large volumes of data. GPUs have massive parallel architecture. By using GPUs, the completion of computationally intensive assignments that benefit from parallel computation is faster compared with conventional central processing units (CPUs).
Due to the increased clock speeds now present in modern GPUs, coupled with optimized parallel architectures, GPU computing is now being utilized by computer scientists and researchers for high-performance computing (HPC). This work was traditionally performed by a CPU using specialized chips known as general-purpose computing on graphics processing units (GPGPU).
John D. Owens, Mike Houston, David Luebke, Simon Green, John E. Stone and James C. Phillips