Long-form writing from members of the lab — research notes accompanying recent papers, engineering write-ups, and the occasional reflection on the practice of running an architecture group.
Graph algorithms often spend more time moving data than processing it. Our accelerators address this bottleneck by performing graph computation directly in resistive memory arrays. They keep the graph in its compact representation as it moves to the crossbars and reuse the same configurations for recurring connection patterns
Our group presents three works at IEEE NEWCAS 2026 in Chicoutimi — two on quantum computing and one on ReRAM-based graph acceleration. A short tour of what each paper contributes.
How team Qortex used a variational quantum algorithm — a parametrized circuit trained in a hybrid quantum-classical loop and run on IBM's ibm_quebec processor — to forecast a financial time series, and took first place in the Quantum Machine Learning track at Mila's Mil'HaQ Fest 2025.