Reliable computing acceleration for AI- and DSP-Applications
Today AI applications are driving technology development. In order to fulfill performance and energy requirements, innovative solutions are needed. One important possibility is the use of in-memory-computing architectures for AI and general DSP applications. IHP has considerable know-how and advanced technology development for systems based on RRAM technology. We are investigating the use of RRAM cells and corresponding arrays in accelerating relevant mathematic operations such as matrix multiplication, emulating neural structures, and performing signal processing algorithms in analog domain. In order to effectively perform these tasks, significant focus is on the modelling of the cells and arrays and reliability management.