TRAM: Training Approximate Multiplier Structures for Low-Power AI Accelerators
This work addresses the need for low-power AI accelerators by co-designing approximate multipliers with model training, offering a practical solution for energy-efficient inference.
TRAM jointly optimizes approximate multiplier structures and AI model parameters to reduce power consumption, achieving up to 25.05% power reduction on CNNs with CIFAR-10 and 27.09% on vision transformers with ImageNet.
Reducing power consumption in AI accelerators is increasingly important. Approximate computing can reduce power consumption while keeping the accuracy loss small. Since multipliers are power-hungry components in AI models, this paper focuses on synthesizing low-power approximate multipliers (AxMs). Unlike prior works that design AxMs separately from AI model training, we present TRAM, which jointly optimizes the AxM structure and AI model parameters to lower power with small accuracy loss. Experiments show that compared to state-of-the-art AxMs, TRAM achieves up to 25.05% AxM power reduction on CNNs with CIFAR-10, and reduces power by up to 27.09% on vision transformers with ImageNet.