Q.ANT adds optical Fourier transforms to NPU Gen 2+
Q.ANT unveiled its NPU Gen 2+ at SC26 in Chicago, adding optical Discrete Fourier Transforms (DFTs) to its Native Processing Unit technology. The addition broadens the mathematical functions developers can execute directly in light for AI and high-performance computing applications.
DFTs can help AI models operate with fewer parameters and less data. Computing these transforms on digital hardware, however, takes time and energy when handling high data volumes. The new processor instead performs DFTs natively in light, addressing the cost of executing the transform itself.
Q.ANT positions the capability as a way for developers to build faster, leaner models and computing applications. The company presents the addition as a demonstration of photonic computing’s potential algorithmic and performance advantages, without providing benchmark results or quantified improvements.