During the recent SC19 supercomputing conference, the top semiconductor and systems vendors discussed and demoed the highest-performance computing solutions in the world. While it’s easy to imagine these platforms solving some of the most challenging problems, and simulating everything from the human genome to climate change, there are thousands of other applications that can benefit from HPC as well.
Enterprises are rushing to push data, applications and IP services to the cloud, but perhaps more companies should consider how they can use HPC and AI to improve their organization’s operational and competitive advantage.
Rapid Pace of Advancement
With the ever-increasing amount of data from the environment — sensors, connected machines, and a variety of other sources — a new challenge has arisen: how to process and understand all this information in a meaningful way.
There are early examples across a wide range of applications:
- Banks and financial institutions are using HPC and AI for threat detection;
- Internet and advertising companies are using it for targeted ads;
- Insurance companies are using it for risk assessment;
- Healthcare organizations are using it for detection and treatment of cancer and other ailments;
- Chip companies are using it for new chip designs; and
- farmers are using it to improve crop yields.
Because of the rapid pace of technology, the chip and systems companies that do not offer HPC or AI solutions may not be relevant in the next five years. Because of the technology race, HPC and AI solutions and intellectual property are more accessible now than ever before.
Similarly, those research organizations and enterprises that do not use HPC and AI also will become irrelevant in the future. Even universities must consider using HPC and AI to remain competitive.
One of the major sources of federal funding for university research is the National Science Foundation, which specifies that when it evaluates project proposals it looks for radically different approaches, applies new expertise, or engages novel disciplinary or interdisciplinary perspectives.
Many of the recent grants have targeted projects leveraging HPC and AI.
IBM, Nvidia Use Cases
IBM, one of the world’s largest IT technology and services providers and a leader in HPC and AI, published several use cases on how customers are leveraging HPC and AI. In each case, the organizations leverage IBM’s AC922 server platform with Nvidia GPU acceleration.
Vision Blanco, a Paraguay financial intuition, is using AI to extend the ability of its existing infrastructure to support more small- and medium-sized businesses by better evaluating credit risk.
Bridger Pipeline is using it to improve response time to oil pipeline leaks. The benefit is an improvement of the detection of leaks by one-fifth, while reducing the number of false detections. This not only improves the performance of the pipeline, but also reduces the risk to the environment and the general public.
Oxford Cancer Biomarkers are using AI to detect colorectal cancer sooner to save lives and to reduce medical costs.
One of the most interesting applications is The Harry Butler Institute’s use of image detection and recognition with AI to detect threat species on Barrow Island, a nature reserve off the west coast of Australia.
AI also improves AI — as we are starting to see. Many semiconductor companies have indicated that they are using AI in the layout/design of new chips. AI may be used in the specification of new chips in the future.
In another example, the Oak Ridge National Laboratory is using data from the 4,608 node Summit supercomputer, one of the most powerful supercomputers in the world, to improve the cooling and overall efficiency of the platform.
By combining real-time data from the IBM AC922 servers with local weather and data from the cooling plant, both the performance of Summit and the cooling plant are optimized for maximum efficiency and peak performance.
SC19 offered the latest information about new chips, new systems, and the Top-500 supercomputer rankings. However, at the end of the day, what matters most is not the performance crown, but what we can do with HPC and AI. Ultimately, it’s those innovations that will change businesses and society.