A brand new technical paper titled “Attaining software-equivalent accuracy for hyperdimensional computing with ferroelectric-based in-memory computing” by researchers at College of Notre Dame, Fraunhofer Institute for Photonic Microsystems, College of California Irvine, and Technische Universität Dresden.
“We current a multi-bit IMC system for HDC utilizing ferroelectric field-effect transistors (FeFETs) that concurrently achieves software-equivalent-accuracies, reduces the dimensionality of the HDC system, and improves power consumption by 826x and latency by 30x when in comparison with a GPU baseline. Moreover, for the primary time, we experimentally reveal multi-bit, array-level content-addressable reminiscence (CAM) operations with FeFETs. We additionally current a scalable and environment friendly structure primarily based on CAMs which helps the associative search of huge HVs. Moreover, we research the results of machine, circuit, and architectural-level non-idealities on application-level accuracy with HDC,” states the paper.
Discover the technical paper link here. Revealed November 2022.
Kazemi, A., Müller, F., Sharifi, M.M. et al. Attaining software-equivalent accuracy for hyperdimensional computing with ferroelectric-based in-memory computing. Sci Rep 12, 19201 (2022). https://doi.org/10.1038/s41598-022-23116-w.
FeFETs Bring Promise And Challenges
New expertise may have an effect on NVM, in-memory processing, and neuromorphic computing.
Nonvolatile Capacitive Crossbar Array For In-Memory Computing
Georgia Tech researchers: the nonvolatile capacitive crossbar array was experimentally demonstrated for in-memory computing