high performance computing on graphics processing units: hgpu.org

Data centers today are the backbone of the modern economy, from the server rooms that power small to midsize organizations to the enterprise data centers that support U.S. corporations and provide access to cloud computing services. According to the Natural Resources Defense Council, data centers are one of the largest and fastest-growing consumers of electricity in the United States. In 2013, U.S. data centers consumed an estimated 91 billion kilowatt-hours of electricity-enough to power all the households in New York City twice over-and are on track to reach 140 billion kWh by 2020 [1]. Clearly, lowering power is essential for the scaling of data centers to improve reliability and lower operating costs. Data center servers vary, depending on the server application. Many servers run for long periods without interruption, making hardware reliability and durability extremely important. Although servers can be built from commodity computer parts, mission-critical enterprise servers often use specialized hardware for application acceleration, including graphics processing units (GPUs) and digital signal processors (DSPs). Now, many companies are looking to add field-programmable gate arrays (FPGAs) for their highly parallel architecture and relatively low power consumption. Xilinx’s new SDAccel development environment removes programming as a gating issue to FPGA utilization in this application by providing developers with a familiar CPU/GPU-like environment.