[02.07.2020] Report published on mapping common micro-kernels - enhanced with in-memory computing accelerators - to realistic digital applications
Prepared by MNEMOSENE consortium partner ETH Zurich, the report explains the process and initial experimental results of mapping common micro-kernels to realistic digital applications (e.g. database, image processing and IoT sensory), which have been enhanced using in-memory computing accelerators.
Compared to service-oriented architectures (SoA), the experimental results demonstrated improvements of an order-of-magnitude, even for smaller digital systems not designed to fully exploit the advantages of in-memory computing systems. Notably, under realistic conditions, a 38.7x improvement in energy efficiency was observed.
To read the related project report, please click on the following link.
Compared to service-oriented architectures (SoA), the experimental results demonstrated improvements of an order-of-magnitude, even for smaller digital systems not designed to fully exploit the advantages of in-memory computing systems. Notably, under realistic conditions, a 38.7x improvement in energy efficiency was observed.
To read the related project report, please click on the following link.