Chen, Peng | Search NASA

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

Adaptive Patching for High-resolution Image Segmentation with Transformers

Attention-based models are proliferating in the space of image analytics, including segmentation. The standard method of feeding images to transformer encoders is to divide the images into patches and then feed the patches to the model as a linear sequence of tokens. For high-resolution images, e.g. microscopic pathology images, the quadratic compute and memory cost prohibits the use of an attention-based model, if we are to use smaller patch sizes that are favorable in segmentation. The solution is to either use custom complex multi-resolution models or approximate attention schemes. We take inspiration from Adapative Mesh Refinement (AMR) methods in HPC by adaptively patching the images, as a pre-processing step, based on the image details to reduce the number of patches being fed to the model, by orders of magnitude. This method has a negligible overhead, and works seamlessly with any attention-based model, i.e. it is a pre-processing step that can be adopted by any attention-based model without friction. We demonstrate superior segmentation quality over SoTA segmentation models for realworld pathology datasets while gaining a geomean speedup of 6.9× for resolutions up to 64K2, on up to 2, 048 GPUs.

Zhang, Enzhi↗

NASA NTRS2019

Direct-Write Printing on Three-Dimensional Geometries for Miniaturized Detector and Electronic Assemblies

Aerosol Jet Printing (AJP) and other additive electronics technologies will enable new detector assemblies that were not previously possible with traditional assembly processes. This technology has the potential to reduce mass and volume for fabrication and assembly of detector, instrument and electronics concepts by reducing trace widths down to 10 microns, printing on complex geometries, and allowing new concept production. This poster summarizes development efforts of printed interconnects for space applications.

Printed Hybrid Electronics↗

NASA NTRS2018

Direct-Write Printing on Three-Dimensional Geometries for Miniaturized Detector and Electronic Assemblies

Direct-write printing techniques will enable new detector assemblies that were not previously possible with traditional assembly processes. Detector concepts were manufactured using this technology to validate repeatability. Additional detector applications and printed wires on a 3-dimensional magnetometer bobbin will be designed for print. This effort focuses on evaluating performance for direct-write manufacturing techniques on 3-dimensional surfaces. Direct-write manufacturing has the potential to reduce mass and volume for fabrication and assembly of advanced detector concepts by reducing trace widths down to 10 microns, printing on complex geometries, allowing new electronic concept production, and reduced production times of complex those electronics.

Direct-write printing↗

NASA NTRS2017

Direct-Write Printing on Three-Dimensional Geometries for Miniaturized Detector and Electronic Assemblies

Direct-write printing techniques will enable new detector assemblies that were not previously possible with traditional assembly processes. Detector concepts were manufactured using this technology to validate repeatability. Additional detector applications and printed wires on a 3-dimensional magnetometer bobbin will be designed for print. This effort focuses on evaluating performance for direct-write manufacturing techniques on 3-dimensional surfaces. Direct-write manufacturing has the potential to reduce mass and volume for fabrication and assembly of advanced detector concepts by reducing trace widths down to 10 microns, printing on complex geometries, allowing new electronic concept production, and reduced production times of complex those electronics.

Additive Manufacturing↗

NASA NTRS2010

Fiber optic gas sensor

A gas sensor includes an in-fiber resonant wavelength device provided in a fiber core at a first location. The fiber propagates a sensing light and a power light. A layer of a material is attached to the fiber at the first location. The material is able to absorb the gas at a temperature dependent gas absorption rate. The power light is used to heat the material and increases the gas absorption rate, thereby increasing sensor performance, especially at low temperatures. Further, a method is described of flash heating the gas sensor to absorb more of the gas, allowing the sensor to cool, thereby locking in the gas content of the sensor material, and taking the difference between the starting and ending resonant wavelengths as an indication of the concentration of the gas in the ambient atmosphere.

Chen, Peng↗