Moore’s law has finally run out of steam for CPUs. The number of x86 cores that can be placed cost-effectively on a single chip has reached a practical limit, making higher densities prohibitively expensive for most applications. Fortunately, for big data analytics, machine learning, and database applications, a more capable and cost-effective alternative for scaling compute performance is already available: the graphics processing unit, or GPU. In this report, executives at Kinetica and Sierra Communications explain how incorporating GPUs is ideal for keeping pace with the relentless growth in streaming, complex, and large data confronting organizations today. Technology professionals, business analysts, and data scientists will learn how their organizations can begin implementing GPU-accelerated solutions either on premise or in the cloud. This report explores: How GPUs supplement CPUs to enable continued price/performance gains The many database and data analytics applications that can benefit from GPU acceleration Why GPU databases with user-defined functions (UDFs) can simplify and unify the machine learning/deep learning pipeline How GPU-accelerated databases can process streaming data from the Internet of Things and other sources in real time The performance advantage of GPU databases in demanding geospatial analytics applications How cognitive computing—the most compute-intensive application currently imaginable—is now within reach, using GPUs
