WAGONN: Weight Bit Agglomeration in Crossbar Arrays for Reduced Impact of Interconnect Resistance on DNN Inference Accuracy
This work addresses the interconnect resistance problem in IMC-based DNN accelerators, which is critical for maintaining inference accuracy in deeply scaled technologies.
The paper proposes SWANN, a technique that shuffles weights in crossbar arrays to mitigate the accuracy drop caused by interconnect resistance in DNN accelerators. For ResNet-20/CIFAR-10 on 128x128 arrays in 7nm technology, SWANN improves accuracy from 47.78% to 83.5% with minimal hardware overhead.
Deep neural network (DNN) accelerators employing crossbar arrays capable of in-memory computing (IMC) are highly promising for neural computing platforms. However, in deeply scaled technologies, interconnect resistance severely impairs IMC robustness, leading to a drop in the system accuracy. To address this problem, we propose SWANN - a technique based on shuffling weights in crossbar arrays which alleviates the detrimental effect of wire resistance on IMC. For 8T-SRAM-based 128x128 crossbar arrays in 7nm technology, SWANN enhances the accuracy from 47.78% to 83.5% for ResNet-20/CIFAR-10. We also show that SWANN can be used synergistically with Partial-Word-LineActivation, further boosting the accuracy. Moreover, we evaluate the implications of SWANN for compact ferroelectric-transistorbased crossbar arrays. SWANN incurs minimal hardware overhead, with less than a 1% increase in energy consumption. Additionally, the latency and area overheads of SWANN are ~1% and ~16%, respectively when 1 ADC is utilized per crossbar array.