High accuracy optical processors: a new performance comparison

Area of Science:

• Computer Engineering
• Optical Computing
• Signal Processing

Background:

• Traditional processors face limitations in computational speed.
• Hybrid processors offer potential for overcoming these limitations.
• Analog-to-digital conversion is a key bottleneck in signal processing.

Purpose of the Study:

• To evaluate the performance of a hybrid time and space integrating processor.
• To compare optical and digital processor performance using a specific metric.
• To analyze the scalability of optical versus digital systems.

Main Methods:

• Utilized a hybrid time and space integrating processor architecture.
• Employed the digital multiplication-by-analog convolution algorithm.
• Measured performance using operations per second per component.

Main Results:

• The hybrid processor achieves more than one operation per analog-to-digital conversion.
• Optical processors exhibit nearly quadratic performance scaling with component speed.
• Digital processors demonstrate only linear performance scaling.

Conclusions:

• Hybrid processors, particularly optical systems, offer superior performance scaling.
• Optical computing presents a promising avenue for high-speed signal processing.
• Component speed significantly impacts the efficiency of optical processing systems.