Vector Processing Rises to the Challenges
of AI and Machine Learning
In the complex and evolving world of computing, processor technologies tend to be designed for specific product applications and align well with the market trend curves. Often times however, these processor technologies are adapted for other use cases and applications that stretch the processors’ capabilities. Sometimes that adaptation can last for many years or even decades until a physical or performance limitation is recognized. On the other hand, it is also common for processor technologies to be developed and made available too early, and not be adapted at all. However, it is rare that two processor technologies, designed for fundamentally different product applications in two vastly different fields, converge to solve the needs of an emerging, market-driven, application. Together, the two processor technologies perform better than trying to solve the challenges independently.
This convergence is precisely what is happening with vector and scalar processing, two key technologies that have coexisted to serve distinctly different computing markets. Once reserved solely for supercomputers, vector processing is now being combined with scalar processing to solve some of the most critical statistical machine learning and collaborative filtering problems facing the AI industry.
This white paper discusses the evolution of these two approaches to computing, and how the technology challenges of artificial intelligence (AI) and machine learning are calling for them to come together in an unprecedented solution.