Improving Energy Saving of One-sided Matrix Decompositions on CPU-GPU Heterogeneous Systems

Abstract

One-sided dense matrix decompositions (e.g., Cholesky, LU, and QR) are the key components in scientific computing in many different fields. Although their design has been highly optimized for modern processors, they still consume a considerable amount of energy. As CPU-GPU heterogeneous systems are commonly used for matrix decompositions, in this work, we aim to further improve the energy saving of onesided matrix decompositions on CPU-GPU heterogeneous systems. We first build an Algorithm-Based Fault Tolerance protected overclocking technique (ABFT-OC) to enable us to exploit reliable overclocking for key matrix decomposition operations. Then, we design an energy-saving matrix decomposition framework, Bi-directional Slack Reclamation (BSR), that can intelligently combine the capability provided by ABFT-OC and DVFS to maximize energy saving and maintain performance and reliability. Experiments show that BSR is able to save up to 11.7% more energy compared with the current best energy saving optimization approach with no performance degradation and up to 14.1% 𝐸𝑛𝑒𝑟𝑔𝑦×𝐷𝑒𝑙𝑎𝑦 2 reduction. Also, BSR enables the Pareto efficient performanceenergy trade-off, which is able to provide up to 1.43× performance improvement without costing extra energy.CCS Concepts: • Hardware → Power and energy; • Computer systems organization → Dependable and fault-tolerant systems and networks.

Exploring Winograd Convolution for Cost-Effective Neural Network Fault Tolerance

“…In addition, many prior works [9]- [12] demonstrate that the fault tolerance of DNNs can also be utilized to relax the requirements of 100% correctness of the execution, and leveraged for higher performance and energy efficiency through techniques such as voltage scaling [13] [14], overclocking [15] [16] and model pruning [17].…”