Multiply-Accumulate Enhanced BDD-Based Logic Synthesis on RRAM Crossbars

Saman Fröhlich, Saeideh Shirinzadeh, Rolf Drechsler

In: IEEE International Symposium on Circuits & Systems (ISCAS). IEEE International Symposium on Circuits and Systems (ISCAS-2020) May 17-20 Sevilla Spain 2020.


Resistive random access memory (RRAM) is a non-volatile memory technology which allows to perform computa-tions in both digital and analog circuits. Multiply-Accumulate(MAC) is an analog column-based operation enabled on RRAMcrossbars providing high efficiency to perform complex matrixvector multiplications, which is attractive for neural networkaccelerators. However, the analog computational capability ofRRAM devices has not been yet utilized for logic synthesis.In this paper, we show how a synthesis approach based onbinary decision diagrams (BDD) can efficiently exploit efficientMAC computation enabled by RRAM. The proposed approachhighly benefits from a symmetric structure of Boolean functions.Therefore, a design methodology is presented which optimizesand approximates BDDs under provided error thresholds tomaximize efficiency of synthesized logic circuits under negligibleloss of accuracy. In the experiments, we show that our proposedsynthesis approach allows for an average reduction of up to 47%in the number of operations and up to 66% in the numberof required devices compared to state-of-the art methods, evenwithout approximation. Using approximation, we can furtherreduce the number of required devices.

Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence