General Electric (GE) has been sharing stories about its successes with high performance computing at its Advanced Computing Lab and plans to offer a window into how the company makes use of strategic technologies.
The Advanced Computing Lab is, according to Chris McConnell from the Edison Engineering Development Program, not focused just on hardware, but more on gaining a deeper understanding of “how computer architectures work for getting the algorithms and applications to meet performance specs.” He points to CUDA and OpenCL as programming tools that are among those he and his team work with.
GE also focuses on MPI programming, data intensive simulation, algorithm optimization and evaluation of new architectures. He says that these technologies are assisting GE with their work on radiation detection systems, next-gen computers for aircraft, large scale simulations of alloys and turbine designs and biomedical imaging. McConnell points to related work that also goes on in their Computational Intelligence Lab, which is geared toward analytics and machine learning.
McConnell noted that his company has been able to realize tremendous benefits from the use of HPC. He said this week that “the idea of high performance and scientific computing is ever growing with the use of GPUs, commodity clusters and high end supercomputers” and that “the work done using these machines and techniques have enabled GE to make decisions in a fraction of the time that it once took.”
For those with the time, past recorded webinars that share insights about the intersection of supercomputing, engineering and data-intensive computing are available here—a site that contains other sessions on diverse topics relevant to the HPC community, including a detailed exploration of the “Internet of Things.”
The company will be holding another series of webinars in August on topic that include the use of social media and collaboration tools for research and another that considers how advanced automation and analytics can create new ways to operate complex energy and other networks.