Engineering topics

Luo, Tao

Publications and source records attributed to Luo, Tao.

5 records

DOE OSTI2024

autoGEMM: Pushing the Limits of Irregular Matrix Multiplication on Arm Architectures

This paper presents an open-source library that pushes the limits of performance portability for irregular General Matrix Multiplication (GEMM) on the widely-used Arm architectures. Our library, autoGEMM, is designed to support a wide range of Arm processors: from edge devices to HPC-grade CPUs. autoGEMM generates optimized kernels for various hardware configurations by auto-combining fragments of autogenerated micro-kernels that employ hand-written optimizations to maximize computational efficiency. We optimize the kernel pipeline by tuning the register reuse and the data load/store overlapping. In addition, we use a dynamic tiling scheme to generate balanced tile shapes. Finally, we position autoGEMM on top of the TVM framework where our dynamic tiling scheme prunes the search space for TVM to identify the optimal combination of parameters for code optimization. Evaluations on five different classes of Arm chips demonstrate the advantages of autoGEMM. For small matrices, autoGEMM achieves 98% of peak and up to 2.0x speedup over state-of-the-art libraries such as LIBXSMM and LibShalom. For irregular matrices (i.e. tall skinny and long rectangles), autoGEMM is 1.3-2.0x faster than widely-used libraries such as OpenBLAS and Eigen. autoGEMM is publicly available at: https://github.com/wudu98/autoGEMM.

Wu, Du↗

DOE OSTI2024

Unlocking the potential of biogas systems for energy production and climate solutions in rural communities

On-site conversion of organic waste into biogas to satisfy consumer energy demand has the potential to realize energy equality and mitigate climate change reliably. However, existing methods ignore either real-time full supply or methane escape when supply and demand are mismatched. Here, we show an improved design of community biogas production and distribution system to overcome these and achieve full co-benefits in developing economies. We take five existing systems as empirical examples. Mechanisms of synergistic adjusting out-of-step biogas flow rates on both the plant-side and user-side are defined to obtain consumption-to-production ratios of close to 1, such that biogas demand of rural inhabitants can be met. Furthermore, carbon mitigation and its viability under universal prevailing climates are illustrated. Coupled with manure management optimization, Chinese national deployment of the proposed system would contribute a 3.77% reduction towards meeting its global 1.5 °C target. Additionally, fulfilling others’ energy demands has considerable decarbonization potential.

09 BIOMASS FUELS↗

DOE OSTI2024

MedNAS: Multiscale Training-Free Neural Architecture Search for Medical Image Analysis

Deep neural networks have demonstrated impressive results in medical image analysis, but designing suitable architectures for each specific task is expertise dependent and time consuming. Neural architecture search (NAS) offers an effective means of discovering architectures. It has been highly successful in numerous applications, particularly in natural image classification. Yet, medical images possess unique characteristics, such as small regions and a wide variety of lesion sizes, that differentiate them from natural images. Furthermore, most current NAS methods struggle with high computational costs, especially when dealing with high-resolution image datasets. In this article, we present a novel evolutionary NAS method called multiscale training-free neural architecture search (MSTF-NAS) to address these challenges. Specifically, to accommodate the broad range of lesion region sizes in disease diagnosis, we develop a new reduction cell search space that enables the search algorithm to explicitly identify the optimal scale combination for multiscale feature extraction. Further, to overcome the issue of high computational costs, we utilize training-free indicators as performance measures for candidate architectures, which allows us to search for the optimal architecture more efficiently. More specifically, by considering the capability and simplicity of various networks, we formulate a multiobjective optimization problem that involves two training-free indicators and model complexity for candidate architectures. Extensive experiments on a large medical image benchmark and a publicly available breast cancer detection dataset are conducted. The empirical results demonstrate that our MSTF-NAS outperforms both human-designed architectures and current state-of-the-art NAS algorithms on both datasets, indicating the effectiveness of our proposed method.

60 APPLIED LIFE SCIENCES↗

DOE OSTI2023

Evolutionary Architecture Search for Generative Adversarial Networks Based on Weight Sharing

Generative adversarial networks (GANs) are a powerful generative technique but frequently face challenges with training stability. Network architecture plays a significant role in determining the final output of GANs, but designing a fine architecture demands extensive domain expertise. This article aims to address this issue by searching for high-performance generator’s architectures through neural architecture search (NAS). The proposed approach, called evolutionary weight sharing GANs (EWSGAN), is based on weight sharing and comprises two steps. First, a supernet of the generator is trained using weight sharing. Second, a multiobjective evolutionary algorithm (MOEA) is employed to identify optimal subnets from the supernet. These subnets inherit weights directly from the supernet for fitness assessment. Two strategies are used to stabilize the training of the generator supernet: 1) a fair single-path sampling strategy and 2) a discarding strategy. Experimental results indicate that the architecture searched by our method achieved a new state-of-the-art among NAS–GAN methods with a Fréchet inception distance (FID) of 9.09 and an inception score (IS) of 8.99 on the CIFAR-10 dataset. Finally, it also demonstrates competitive performance on the STL-10 dataset, achieving FID of 21.89 and IS of 10.51.

97 MATHEMATICS AND COMPUTING↗

NASA NTRS2006

Single-Frequency Narrow Linewidth 2 Micron Fiber Laser

A compact single frequency, single-mode 2 .mu.m fiber laser with narrow linewidth, <100 kHz and preferably <100 kHz, is formed with a low phonon energy glass doped with triply ionized rare-earth thulium and/or holmium oxide and fiber gratings formed in sections of passive silica fiber and fused thereto. Formation of the gratings in passive silica fiber both facilitates splicing to other optical components and reduces noise thus improving linewidth. An increased doping concentration of 0.5 to 15 wt. % for thulium, holmium or mixtures thereof produces adequate gain, hence output power levels for fiber lengths less than 5 cm and preferably less than 3 cm to enable single-frequency operation.

Jiang, Shibin↗