LV: Latency-Versatile Floating-Point Engine for High-Performance Deep Neural Networks

Computing latency is an important system metric for Deep Neural Networks (DNNs) accelerators. To reduce latency, this work proposes LV, a latency-versatile floating-point engine (FP-PE), which contains the following key contributions: 1) an approximate bit-versatile multiplier-and-accumulate (BV-MAC) unit with early shifter and 2) an on-demand fixed-point-to-floating-point conversion (FXP2FP) unit. The extensive experimental results show that LV outperforms baseline FP-PE and redundancy-aware FP-PE by up to 2.12x and 1.3x speedup using TSMC 40-nm technology, achieving comparable accuracy on the ImageNet classification tasks.