Gone are the days when processors were single core or multiple cores based. In the days of high computing requirements, an alternative technology is being developed. Its being developed in the area as taken by the processors of present times. It can theoretically do the computation of hundreds of processors. It has an element of parallelism in it. Its called as System On Chip.
Each of these SoC’s are made up of 2D mesh network. A module is present at each of the intersections in the mesh. In turn each of these modules consists of a controller which routes the data to be processed . Also modules are the processing unit of the setup. They contain the algorithm of a particular application which is used for analysing the data. Typically a mesh consists of a NxN network of modules.
The data to be processed is initially split up into small bits. They are assigned a module address and using a main buffer, they are sent across to the network where the modules receive them and analyze .Then they come up with a result. This result is sent across to the main buffer which would reorganize the results. As the number of modules increase, the processing capability of a SoC would also increase.
I’ve been fortunate to have been working on it for past one year where as a team we have been trying to implement it on an FPGA . Also the designs developed would be tested in next 2 months. With all the efforts we have put in we are hoping for the best 🙂