Artificial Intelligence (Ai) is quickly becoming the backbone of many software features and services we use every day. Anytime you search for specific image in Google Photos or read through Facebook’s NewsFeed, you’re using a form of AI software and algorithms. Another example is virtual assistants, from SIRI to Amazon’s ECHO – potentially millions of people can ask directions, answer general questions, find out about the weather, and even order a pizza of Uber.
Typically, these kind of use cases have requirements for a large number of fine-grained workloads that, individually, have small computational demands compared to their data feed I/O processing and general management overhead. TruStream’s programming model and associated core fabric, TruFabric, improves performance and latency of fine-grained workloads by dynamically and efficiently allocating processor resources so they match the ever changing real-time demands of these systems.