A Faster IoT

A faster way to IoT

New Programming Code Called TACO

We live in a world were most data is sparse. By that I mean that most data is made up of huge tables where most of the data is blank. For example, if you think of a matrix of products (X) and customers (Y), most values in your matrix would be blank except for the few cells where a given customer has bought one of your products; those cells may be represented as 1 while the other cells would be 0 or blank.

Now imagine this data matrix applied to all the machines, automobiles, and other IoT devices who are constantly recording data 24 hours a day. Most of the time nothing happens, and all of that time is represented as a 0. The problem with this is that, computers spend much time multiplying and dividing by zero. Until now this was accepted as a cost of big data. however, researchers at MIT presented a new system which optimizes code for sparse data. The new code Is called TACO (Tensor algebra compiler) and offers 100x faster calculations when calculating large data. You can read more about it here.

The beauty of this new code is the ability to sift through immense volumes of data in much shorter time frames. This type of technology is invaluable as everything becomes connected through IoT and our reliance on data to make smarter decisions grows.