Projects
The content on this page was translated automatically.
ELYSIA
Named after a sea snail that ingests algae and uses it for photosynthesis, Elysia is the name of our sythesis tool for implementing neural networks.
The tool enables the experimental investigation of different optimization approaches for pure hardware implementations of pre-trained networks. The core topic is the quantization of all occurring values (parameters and data) of the original network, which usually works with floating point numbers, and the generation and adaptation of the associated hardware operators.
ORIGAMI HLS
Originally from the Japanese for the art of paper folding, this is the name of our High-Level Synthesis (HLS) tool that we developed for circuit optimization experiments.
The optimization ultimately consists of "folding" a functionality to be implemented around the available operators in order to (re)use them as efficiently as possible.
The tool allows design space exploration (DSE), integrates various optimization algorithms and is currently used by us to optimize hardware structures for modulo scheduling and loop pipelining.
Pipelined Multiple Constant Multiplication using RPAG
RPAG is an optimization algorithm for the pipelined multiple constant multiplication (PMCM) problem. Its sources are available as open-source.
More information can be found here.
m-learning project ESkaLATE
Development ofa software environmentfor course-integrated, adaptivelearningopportunitieson mathematical topicsin electrical engineering
Apps and software in teaching and learning
We develop apps for mobile devices and other software to support teaching and learning.
Further information can be found here.