QUIDAM: A Framework for Qu ant i zation-aware D NN A ccelerator and M odel Co-Exploration

Abstract

As the machine learning and systems communities strive to achieve higher energy-efficiency through custom deep neural network (DNN) accelerators, varied precision or quantization levels, and model compression techniques, there is a need for design space exploration frameworks that incorporate quantization-aware processing elements into the accelerator design space while having accurate and fast power, performance, and area models. In this work, we present QUIDAM , a highly parameterized quantization-aware DNN accelerator and model co-exploration framework. Our framework can facilitate future research on design space exploration of DNN accelerators for various design choices such as bit precision, processing element type, scratchpad sizes of processing elements, global buffer size, number of total processing elements, and DNN configurations. Our results show that different bit precisions and processing element types lead to significant differences in terms of performance per area and energy. Specifically, our framework identifies a wide range of design points where performance per area and energy varies more than 5 × and 35 ×, respectively. With the proposed framework, we show that lightweight processing elements achieve on par accuracy results and up to 5.7 × more performance per area and energy improvement when compared to the best INT16 based implementation. Finally, due to the efficiency of the pre-characterized power, performance, and area models, QUIDAM can speed up the design exploration process by 3-4 orders of magnitude as it removes the need for expensive synthesis and characterization of each design.

Demeter: A Fast and Energy-Efficient Food Profiler Using Hyperdimensional Computing in Memory

“…Several prior works(e.g., [22,51,55,128]) investigate the new requirements, tradeoffs, and challenges that arise from using the CIM paradigm (e.g., dealing with non-idealities in the analog operations). To our knowledge, no work has proposed a complete solution or framework for these challenges; thus, this area requires further investigation.…”

EfficientRainNet: Smaller Neural Networks Based on EfficientNetV2 for Rainfall Nowcasting

“…Workloads worsen the memory bottleneck that lowers the overall performance of systems where deep learning models are run since neural network workloads continue to have big memory footprints and significant computational requirements to attain improved accuracy (Inci 2022;Inci et al, 2022a). As deep learning models have become more proficient in many tasks, developing smaller neural network models in terms of trainable parameters has attracted interest from many researchers in the field (Inci et al, 2022b) to support operational needs in flood forecasting (Krajewski et al, 2021;Xiang and Demir, 2022) and inundation mapping (Hu and Demir, 2021;Li et al, 2022).…”