As televisions grow in size and the technology involved with video media improves, the resolution of televisions also grows too. Soon we may be seeing ultrahigh-definition (UHD) TVs hitting store shelves with four times the pixels of modern Full HD televisions with 1080 lines. With four times the pixels, UHD televisions need to be able to do as much as four times the work to process a signal, which is why advanced processing chips are being developed, and now MIT has designed one.
To keep the amount of data needed down, High-Efficiency Video Coding (HEVC) is being used to encode UHD video streams, and this codec, like others currently in use, operates on the principle that most of the pixels in an image do not change during a scene. That means only one frame, the key frame, needs to be completely encoded for a scene, while motion data and correction information is what defines the changes throughout the scene. An HEVC chip needs to be able to take in this information and very quickly process it to create the UHD video stream, which is why MIT designed its with multiple efficiency-increase technologies such as simplifying a 32x32 matrix to just 32 unique numbers, instead of 1024 values, and addressing pixel data within memory in squares, instead of rows which separate pixels above and below each other.
Designed by MIT, fabricated by TSMC, and funded by TI, the new chip is not for commercial use. However, the researchers' work will likely influence others as they build HEVC chips for the TVs and other devices you and I buy.