The Central Processing Unit (CPU) has been at the heart of High Performance Computing (HPC) for decades. However, in recent years, advances in parallel processing technology mean the landscape has changed dramatically. Modern Graphics Processing Units(GPUs), which feature 1,000s of compute cores and high capacity, high-bandwidth memory, have become extremely adept at processing massively parallel HPC code — delivering results very quickly. The GPU’s shift from dedicated graphics processor to powerful compute processor has been made possible by OpenCL™, the open standard for parallel programming (recently updated to version 2.1). With OpenCL, HPC code can be optimized to run on virtually any processor, be it GPU, CPU,
even a mobile processor. AMD offers a whole family of dedicated server-grade GPUs, designed specifically for heavyduty compute operations. The AMD FirePro™ S9150 server
GPU, for example, can deliver exceptional processing power inside some HPE ProLiant servers and clusters. GPU compute touches many applications, from engineering analysis and seismic processing to medical imaging and ray trace rendering.
HPC-capable GPUs broadly fall into two camps — those that are optimized for single precision operations and those that are optimized for both single and double precision operations. Single precision operations are used in applications where high performance takes precedence over extreme accuracy. Examples include video enhancement, signal processing, video transcoding, deep neural networks (DNN) and digital rendering applications. Double precision operations are used in applications where numerical precision is critical. Examples include Computational Fluid Dynamics (CFD), Structural Mechanics, Reservoir Simulation and Aerodynamics applications.
Error: Contact form not found.
All information that you supply is protected by our privacy policy. By submitting your information you agree to our Terms of Use.
* All fields required.