Fugu is a spiking neural network library designed to produce algorithms that are composable, scalable and portable. Developers define spiking neural network functions (“Bricks”) using procedural code to generate platform-agnostic networks. These networks are then complied to specific execution backends, either spiking simulators or spiking neuromorphic hardware. This approach allows components from different authors to work easily together and frees the user from worrying about platform-specific requirements.
Predictive Quantum Simulator for state-of-the-art and beyond CMOS device technologies
Modern transistors have a gate length of tens of nanometers or smaller. At this scale, effects of quantum confinement, atomic defects, and the environment play crucial role in the device performance, necessitating the process variation and reliability analysis. This can be achieved by creating a Device Digital Twin through predictive first principle simulations.
CBR3D is a Sandia’s quantum transport simulator that can accurately predict electrical characteristics and reveal device physics of modern transistors from first-principles, i.e. without the use of fitting parameters. The simulator is capable of modelling devices of arbitrary geometry and multiple contacts and scales linearly with the volume size, which makes it suitable for predictive simulation of realistic nanoscale-FET structures. Relevant forms of elastic and inelastic scatterings are included in the simulator, including electron-electron interaction, and scattering on discrete charged impurities. The validity and predictive ability of the simulator has been demonstrated: in GAAFETs, low- and high- bias electrical characteristics reported by IBM has been reproduced; a novel Gate-Induced-Drain-Leakage effect has been predicted1; in beyond-CMOS devices, the sheet resistance values for highly doped sheets have been accurately reproduced across a wide range of dopant densities2, and the tunneling resistance in Atomically Precise Advanced Manufactured (APAM) tunnel junctions3 has been confirmed by experimental measurements4, successfully validating the theory5.
CBR3D predicts a novel Gate-Induced-Drain-Leakage effect
