Reducing the programming burden

The rising adoption of remote, edge computing and embedded systems has meant increasing workloads are moving to the edge. And with these technologies come even greater demand for versatile programming.

“The need for programmability has never been greater,” said Dermot Hargaden, vice president and general manager for Intel’s Programmable Solutions Group, at the company’s FPGA Technology Day in November. “The need for pervasive compute, constant connectivity, for the parallelization that FPGAs can provide, is only going to grow.”

In fact, the FPGA market grew from $6 billion in 2020 to $7 billion in 2021, according to Intel, and a projected compound annual growth rate of 12% through 2027 could bring the market to a value of $13 billion. There’s good reason for the sunny projections. A field-programmable array, or FPGA, gives programmers a powerful and versatile tool. Allowing for fast rewiring and consuming little power, FPGAs are suited for emerging applications such as AI at the edge, where developers must be nimble and innovative.

In this edition’s cover story here, “The Rise of Embedded Vision Puts FPGAs in the Spotlight,” contributing editor Hank Hogan surveys the FPGA market and its leading players, including Lattice Semiconductor, Intel, Efinix, Microchip, Concurrent EDA, and Avnet. He details novel applications that are becoming possible thanks to FPGAs. Such applications include implementing the JPEG XS compression standard, which allows for imaging and video encoding at a more than tenfold reduction of the data size of a video stream — making it possible to run high-bandwidth video over lower-power networks — all with a negligible effect on quality.

In our next feature, we turn to an area of the spectrum that continues to garner interest from systems integrators and end users alike: shortwave infrared. In recent years, advancements in sensor technology — notably, those incorporating quantum dots and Sony’s SenSWIR sensor — have expanded options for integrators. But converting a camera incorporating a sensor that can detect in the visible, NIR, and SWIR to “SWIR only” can be a tall order. In “Selecting the Right Filter Is Critical to SWIR Imaging”, Midwest Optical Systems’ James Gardiner examines the respective role for longpass and bandpass filters in blocking unwanted portions of the spectrum or targeting specific spectral signatures.

Finally, CMOS image sensors must be clean and defect-free to perform properly in an automatic optical inspection. David Harel of Carmel Forge Ltd. and Doron Nakar of Nova Ltd. outline the series of cleanliness and defect checks that systems integrators should follow, along with the proper cleaning steps. “Inspecting CMOS Image Sensors for Dirt and Defects” begins here.

We hope you enjoy the issue!