Driving Innovation in Memory Architecture of Consumer Hardware with Digital Photography and Machine Intelligence Use Cases
Eric Shiu, SukHwan Lim · 2017
Mobile devices have been the driving force behind some of the most important semiconductor innovations in the past decade. In fact, iPhone's 10 year anniversary is coming up in June 2017. Machine intelligence and digital photography will continue to be main differentiating features that influence consumers' purchasing choice of mobile devices. It is also known that the rate of scaling (Moore's law) has been slowing down since 20nm process nodes were introduced for both logic CMOS and memory (DRAM, flash) processes. Memory and thermal constraints have been the main bottlenecks for users' perceived performance. Here we look at the memory architecture in consumer devices, such as Chromebooks and Pixel phones. We start from high-level use cases, such as interfacing with Google's personal assistant ("OK Google"), taking high dynamic range (HDR) pictures or recording 4K videos ("camera app"). We then dive into full-system architecture simulation, performance and power analysis. Finally, we discuss a few future research areas in architecture and technology.