Analog In-Memory Computing with Multilevel Memristive Devices for High Performance Computing
Watch a technical presentation exploring TetraMem's innovative solution to the von Neumann bottleneck through analog in-memory computing with multilevel memristive devices. Learn how pre-trained synaptic weights from cloud-based training are programmed directly into multi-level memristors/RRAMs for edge deployment, with capabilities for scenario-specific post-tuning. Discover the significance of achieving stable 8-bit and higher multi-level conductance in individual memristor devices, including breakthrough 11-bit/cell performance. Explore how these advances enable monolithically integrated semiconductor chips with large crossbar arrays on CMOS circuits, manufactured in commercial foundries for high-performance computing applications. Understand the latest developments in arbitrary precision computing based on analog computing, recently accepted for publication in a major journal.