Slides
It has been widely accepted that 50 ms latency is sufficient to maintain a good interactive experience because human beings cannot perceive any delay in causal actions below 50 ms. While this has been good for mouse and keyboard interactions, the use of touchscreen on tablets and smartphones has made delay shorter than 10 ms necessary. In some devices such as head-mounted display with augmented reality, much shorter latency than 10 ms is required.. We analyzed where this latency comes from, and tried to reduce it. Our observation is that faster CPU does not effect on reducing latency too much. Even a processor with multiple cores is not helpful too much. In this talk, I introduce our approach, making your device predict your future and act in advance. By doing so, we can conceal major latency caused by application and graphic stack.