Energy efficient runtime approximate computing on data flow graphs

Abstract

Approximate computing is an emerging computation paradigm that utilizes many applications' intrinsic error resilience to improve power and energy efficiency. Several approaches have been proposed to identify the non-critical computations by analyzing the output sensitivity to the accuracy of the results, and then perform approximate computing on these computations. However, these static approaches only use the prior knowledge (e.g. input ranges) for analysis and fail to consider the runtime information, which limits the energy saving and incurs large computation error. In this paper, we propose a runtime approximate computing framework to solve this problem. The basic idea is to use a low cost method to estimate the impact of each immediate input value to the accuracy of computation at every node in the data flow graph, and then decide whether we should simply use the estimated value or perform an accurate computation. Our novel runtime estimation method is based on converting data to the logarithmic representation. We propose two algorithms to make the decision at certain nodes whether an accurate computation will be needed to balance energy saving and computation error. Experimental results show that this tradeoff ranges from 40% energy saving with 4.85% error on average to 8% energy saving with 0.18% error. Compared to the static DFG node cutting approach, our approach's estimation accuracy is 32x better to achieve the same amount of energy saving.