IP cores are central to the building of today’s systems on a chip. They reduce the time to design significantly. Building today’s artificial intelligence systems require neural networks. These are available as IP cores in IP core libraries. Complicating the matter is that neural networks can be both digital and analog and a mixture of both. Today, memory cells, that is DRAM and SRAM and flash memory cells incorporate analog-to-digital converters, known as mixed signal design blocks. In recent years it was discovered that semiconductor memory processes used to manufacturer memory chips were also well suited for the construction of analog components, allowing for memory cells to include analog-to-digital converters economically. Today memory cells that incorporate analog-to-digital converters are called compute-in-memory (CIM) cells. CIM cells are necessary for the efficient implementation of artificial intelligence systems. And that is because weighted comparisons, used in the process of decision making, can be made quickly by comparing a converted analog weight with a weight stored in CIM memory.
The IPCoreMatrix Neural Network poster along with a number of home décor, office supply, apparel and gifts products based on the design can be found at the IPCoreMatrix Redbubble Store Front.
