Training of Deep Learning Pipelines on Memory-Constrained GPUs via Segmented Fused-Tiled Execution.

Training models with massive inputs is a significant challenge in the development of Deep Learning pipelines to process very large digital image datasets as required by Whole Slide Imaging (WSI) in computational pathology and analysis of brain fMRI images in computational neuroscience. Graphics Processing Units (GPUs) represent the primary workhorse in training and inference of Deep Learning models. In order to use GPUs to run inference or training on a neural network pipeline, state-of-the-art machine learning frameworks like PyTorch and TensorFlow currently require that the collective memory on the GPUs must be larger than the size of the activations at any stage in the pipeline.

NWChem: Past, present, and future.

Specialized computational chemistry packages have permanently reshaped the landscape of chemical and materials science by providing tools to support and guide experimental efforts and for the prediction of atomistic and electronic properties. In this regard, electronic structure packages have played a special role by using first-principle-driven methodologies to model complex chemical and materials processes. Over the past few decades, the rapid development of computing technologies and the tremendous increase in computational power have offered a unique chance to study complex transformations using sophisticated and predictive many-body techniques that describe correlated behavior of electrons in molecular and condensed phase systems at different levels of theory.

Parameterized Specification, Configuration and Execution of Data-Intensive Scientific Workflows.

Data analysis processes in scientific applications can be expressed as coarse-grain workflows of complex data processing operations with data flow dependencies between them. Performance optimization of these workflows can be viewed as a search for a set of optimal values in a multidimensional parameter space consisting of input performance parameters to the applications that are known to affect their execution times. While some performance parameters such as grouping of workflow components and their mapping to machines do not affect the accuracy of the analysis, others may dictate trading the output quality of individual components (and of the whole workflow) for performance.

Performance optimization of tensor contraction expressions for many-body methods in quantum chemistry.

Complex tensor contraction expressions arise in accurate electronic structure models in quantum chemistry, such as the coupled cluster method. This paper addresses two complementary aspects of performance optimization of such tensor contraction expressions. Transformations using algebraic properties of commutativity and associativity can be used to significantly decrease the number of arithmetic operations required for evaluation of these expressions.

An Integrated Framework for Parameter-based Optimization of Scientific Workflows.

Data analysis processes in scientific applications can be expressed as coarse-grain workflows of complex data processing operations with data flow dependencies between them. Performance optimization of these workflows can be viewed as a search for a set of optimal values in a multi-dimensional parameter space. While some performance parameters such as grouping of workflow components and their mapping to machines do not a ect the accuracy of the output, others may dictate trading the output quality of individual components (and of the whole workflow) for performance.

Interactive medical data on demand: a high-performance imaged-based approach across heterogeneous environments.

Medical data in image format continues to increase in both size and complexity. We have integrated advanced techniques in visualization, networked computing, and interface design to improve methods for accessing medical data comprising high-resolution images for reconstructions into three-dimensional volumetric representations. We present two approaches to handle the range of low to high-end client platforms, support visualization functionality, and provide the ability to manipulate very large data over heterogeneous computing and networking environments.