Artificial Intelligence continues to evolve in complexity, creating growing demand for additional computing resources, which in turn creates a growing demand for additional power. Hewlett-Packard Labs is exploring analog solutions to increasing computational efficiency through more calculations per second over GPUs without needing to constantly grow the hardware footprint required for these complex use cases. In the interview here from HPE Discover, Martin Foltin explains how this analog solution works and highlights some of the performance gains realized in early lab testing.
0 comments on “An Analog Solution to Accelerating Artificial Intelligence”