In-Memory Computing (IMC) hardware accelerators for Deep Neural Networks (DNNs) require massive quantity of coefficients stored in a single device to limit performance losses in multicore clusters.