Ep. 21 Graphics Processors and Reducing Cyber Risk for Federal Agencies

The federal government is subjected to thousands of cyberattacks a day. When you combine that statistic with the tremendous gap in people that need to be hired for cyber defense, you can see the problem that is developing.

One way is to gain a better understanding of malicious actors move from network to cloud and then expand into a system. The log information is all there, the problem is that there is so much of it, a human doesn't have a chance of trying to get a handle on the attack vectors taking place.

When it comes to processing enormous amounts of information, the classic Central Processing Unit has limitations. One way around this limiting factor is to use a Graphical Processing Unit, or a GPU. Today, systems architects are designing systems that can cluster thousands of GPUs to accomplish this tremendous task. There are some systems with 10,000 GPUs processing large amounts of data.

During the interview, Bartley Richarson talks about the role nVidia has in understanding the people attacking the federal government. He outlines basic concepts like data preparation, model training, and visualization. When presented with mountains of data and an incredible demand on compute, systems can be structured to help federal managers accelerate time to insight.

The basic example used is determining a best path for a firefighter to follow. There is a better example in outer space. For example, a traditional satellite will gather data and relay that data to a ground station. From there, it will be relayed to a place where the analysis is done. Each step along the way takes time.

One approach is to have a satellite that can do autonomous board data fusion in space. From there, it can use artificial intelligence to relay information to the federal government. This can be as pedestrian as a traffic report to a wide range of military intelligence.