Documentation

The current public debate on the future of the semiconductor industry has turned to discussions about a growing selection of technologies that focuses instead on new system architectures and better use of available silicon through new concepts in circuit, device, and packaging design. The emergence of embedded FPGA is, in fact, not only essential at this juncture of the microelectronics history, but also inevitable. To understand this, a review of the history of FPGA technology is in order.

Since the initial introduction of FPGAs decades ago, each new architecture has continued to employ a bit-wise routing structure. While this approach has been successful, the rise of high-speed communication standards has required ever increasing on-chip bus widths to be able to support these new data rates. Achronix's solution was to create a revolutionary 2D high-speed network on chip (NoC) on top of the traditional segmented FPGA routing structure for its new Speedster7t FPGA family.

The Speedcore design and integration methodology has been defined with intimate awareness of the difficulties ASIC engineering teams must contend with. All the necessary files and flows for capturing the functional, timing and power characteristics of a user-defined and programmed Speedcore instance, along with support for successfully reconfiguring an already field-deployed Speedcore IP embedded in an ASIC, are available to an ASIC development team either as products of the ACE design tools or as deliverables provided by Achronix. This methodology has already been proven in silicon and readily accommodates variations and preferences in company-specific ASIC development methodologies.

Artificial intelligence (AI) and machine learning (ML) technologies now power a rapidly expanding range of product and applications from deeply embedded systems to hyperscale data-center deployments. Although there is a huge degree of diversity in the hardware designs supporting these applications, all require hardware acceleration. Data orchestration encompasses the pre- and post-processing operations that ensure the data seen by a machine learning engine arrives at an optimal speed and in the most suitable form for efficient processing.

Thanks to the rise of big data and the rapid increase in computing power, machine learning technology has experienced revolutionary development in recent years. Machine learning tasks such as image classification, speech recognition, and natural language processing, operate on Euclidean data with a certain size, dimension, and an orderly arrangement. However, in many realistic scenarios, data is represented by complex non-Euclidean data such as graphs. In this context, many new graph-based machine learning algorithm, or graph neural networks (GNNs), are constantly emerging in academia and industry.