Search NASA⌕ Search

SEARCH · Search NASA

Results for “Distributed memory”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 847 records · Page 47

Evolution of CMS Workload Management Towards Multicore Job Support

The successful exploitation of multicore processor architectures is a key element of the LHC distributed computing system in the coming era of the LHC Run 2. High-pileup complex-collision events represent a challenge for the traditional sequential programming in terms of memory and processing time budget. The CMS data production and processing framework is introducing the parallel execution of the reconstruction and simulation algorithms to overcome these limitations. CMS plans to execute multicore jobs while still supporting singlecore processing for other tasks difficult to parallelize, such as user analysis. The CMS strategy for job management thus aims at integrating single and multicore job scheduling across the Grid. This is accomplished by employing multicore pilots with internal dynamic partitioning of the allocated resources, capable of running payloads of various core counts simultaneously. An extensive test programme has been conducted to enable multicore scheduling with the various local batch systems available at CMS sites, with the focus on the Tier-0 and Tier-1s, responsible during 2015 of the prompt data reconstruction. Scale tests have been run to analyse the performance of this scheduling strategy and ensure an efficient use of the distributed resources. This paper presents the evolution of the CMS job management and resource provisioning systems in order to support this hybrid scheduling model, as well as its deployment and performance tests, which will enable CMS to transition to a multicore production model for the second LHC run.

Perez-Calero Yzquierdo, A.↗

Designing Algorithms for the EMU Migrating-threads-based Architecture

The decades-old memory bottleneck problem for data-intensive applications is getting worse as the processor core counts continue to increase. Workloads with sparse memory access characteristics only achieve a fraction of a system’s total memory bandwidth. EMU architecture provides a radical approach to the issue by migrating the computational threads to the location where the data resides. The system enables access to a large PGAS-type memory for hundreds of nodes via a Cilk- based multi-threaded execution scheme.EMU architecture brings brand new challenges in application design and development. Data distribution and thread creation strategies play a crucial role in achieving optimal performance in the EMU platform. In this work, we identify several design considerations that need to be taken care of while developing applications for the new architecture and we evaluate their performance effects on the EMU-chick hardware. We also present a modified BFS algorithm for the EMU system and give experi- mental results for its execution on the platform.

Belviranli, Mehmet E.↗

HDF5 As a Vehicle for in Transit Data Movement

For in transit processing, one of the fundamental challenges is the efficient movement of data from producers to consumers. Exploiting the flexibility offered by the SENSEI generic in situ framework, we have developed a number of different in transit data transport mechanisms. In this work, we focus on the transport mechanism that leverages the HDF5 parallel I/O library, and investigate the performance characteristics of this transport mechanism. For in transit use cases at scale on HPC platforms, one might expect that an in transit data transport mechanism that uses faster layers of the storage hierarchy, such as DRAM memory, would always outperform a transport that uses slower layers of the storage hierarchy, such as an NVRAM-based persistent storage presented as a distributed file system. However, our test results show that the performance of the transport using NVRAM is competitive with the transport that uses socket-based data movement across varying levels of producer and consumer concurrency.}, booktitle = {Proceedings of the Workshop on In Situ Infrastructures for Enabling Extreme-Scale Analysis and Visualization

Gu, Junmin↗

Hybrid and Parallel Domain-Decomposition Methods Development to Enable Monte Carlo for Reactor Analyses

This paper describes code and methods development at the Oak Ridge National Laboratory focused on enabling high-fidelity, large-scale reactor analyses with Monte Carlo (MC). Current state-of-the-art tools and methods used to perform real commercial reactor analyses have several undesirable features, the most significant of which is the non-rigorous spatial decomposition scheme. Monte Carlo methods, which allow detailed and accurate modeling of the full geometry and are considered the gold standard for radiation transport solutions, are playing an ever-increasing role in correcting and/or verifying the deterministic, multi-level spatial decomposition methodology in current practice. However, the prohibitive computational requirements associated with obtaining fully converged, system-wide solutions restrict the role of MC to benchmarking deterministic results at a limited number of state-points for a limited number of relevant quantities. The goal of this research is to change this paradigm by enabling direct use of MC for full-core reactor analyses. The most significant of the many technical challenges that must be overcome are the slow, non-uniform convergence of system-wide MC estimates and the memory requirements associated with detailed solutions throughout a reactor (problems involving hundreds of millions of different material and tally regions due to fuel irradiation, temperature distributions, and the needs associated with multi-physics code coupling). To address these challenges, our research has focused on the development and implementation of (1) a novel hybrid deterministic/MC method for determining high-precision fluxes throughout the problem space in k-eigenvalue problems and (2) an efficient MC domain-decomposition (DD) algorithm that partitions the problem phase space onto multiple processors for massively parallel systems, with statistical uncertainty estimation. The hybrid method development is based on an extension of the FW-CADIS method, which attempts to achieve uniform statistical uncertainty throughout a designated problem space. The MC DD development is being implemented in conjunction with the Denovo deterministic radiation transport package to have direct access to the 3-D, massively parallel discrete-ordinates solver (to support the hybrid method) and the associated parallel routines and structure. This paper describes the hybrid method, its implementation, and initial testing results for a realistic 2-D quarter core pressurized-water reactor model and also describes the MC DD algorithm and its implementation.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Addressing GPU memory limitations for Graph Neural Networks in High-Energy Physics applications

Introduction Reconstructing low-level particle tracks in neutrino physics can address some of the most fundamental questions about the universe. However, processing petabytes of raw data using deep learning techniques poses a challenging problem in the field of High Energy Physics (HEP). In the Exa.TrkX Project, an illustrative HEP application, preprocessed simulation data is fed into a state-of-art Graph Neural Network (GNN) model, accelerated by GPUs. However, limited GPU memory often leads to Out-of-Memory (OOM) exceptions during training, due to the large size of models and datasets. This problem is exacerbated when deploying models on High-Performance Computing (HPC) systems designed for large-scale applications. Methods We observe a high workload imbalance issue during GNN model training caused by the irregular sizes of input graph samples in HEP datasets, contributing to OOM exceptions. We aim to scale GNNs on HPC systems, by prioritizing workload balance in graph inputs while maintaining model accuracy. Our paper introduces diverse balancing strategies aimed at decreasing the maximum GPU memory footprint and avoiding the OOM exception, across various datasets. Results Our experiments showcase memory reduction of up to 32.14% compared to the baseline. We also demonstrate the proposed strategies can avoid OOM in application. Additionally, we create a distributed multi-GPU implementation using these samplers to demonstrate the scalability of these techniques on the HEP dataset. Discussion By assessing the performance of these strategies as data loading samplers across multiple datasets, we can gauge their effectiveness in both single-GPU and distributed environments. Our experiments, conducted on datasets of varying sizes and across multiple GPUs, broaden the applicability of our work to various GNN applications that handle input datasets with irregular graph sizes.

Lee, Claire Songhyun↗

Hybrid and Parallel Domain-Decomposition Methods Development to Enable Monte Carlo for Reactor Analyses

This paper describes code and methods development at the Oak Ridge National Laboratory focused on enabling high-fidelity, large-scale reactor analyses with Monte Carlo (MC). Current state-of-the-art tools and methods used to perform ''real'' commercial reactor analyses have several undesirable features, the most significant of which is the non-rigorous spatial decomposition scheme. Monte Carlo methods, which allow detailed and accurate modeling of the full geometry and are considered the ''gold standard'' for radiation transport solutions, are playing an ever-increasing role in correcting and/or verifying the deterministic, multi-level spatial decomposition methodology in current practice. However, the prohibitive computational requirements associated with obtaining fully converged, system-wide solutions restrict the role of MC to benchmarking deterministic results at a limited number of state-points for a limited number of relevant quantities. The goal of this research is to change this paradigm by enabling direct use of MC for full-core reactor analyses. The most significant of the many technical challenges that must be overcome are the slow, non-uniform convergence of system-wide MC estimates and the memory requirements associated with detailed solutions throughout a reactor (problems involving hundreds of millions of different material and tally regions due to fuel irradiation, temperature distributions, and the needs associated with multi-physics code coupling). To address these challenges, our research has focused on the development and implementation of (1) a novel hybrid deterministic/MC method for determining high-precision fluxes throughout the problem space in k-eigenvalue problems and (2) an efficient MC domain-decomposition (DD) algorithm that partitions the problem phase space onto multiple processors for massively parallel systems, with statistical uncertainty estimation. The hybrid method development is based on an extension of the FW-CADIS method, which attempts to achieve uniform statistical uncertainty throughout a designated problem space. The MC DD development is being implemented in conjunction with the Denovo deterministic radiation transport package to have direct access to the 3-D, massively parallel discrete-ordinates solver (to support the hybrid method) and the associated parallel routines and structure. This paper describes the hybrid method, its implementation, and initial testing results for a realistic 2-D quarter core pressurized-water reactor model and also describes the MC DD algorithm and its implementation.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Forward-Weighted CADIS Method for Variance Reduction of Monte Carlo Reactor Analyses

Current state-of-the-art tools and methods used to perform 'real' commercial reactor analyses use high-fidelity transport codes to produce few-group parameters at the assembly level for use in low-order methods applied at the core level. Monte Carlo (MC) methods, which allow detailed and accurate modeling of the full geometry and energy details and are considered the 'gold standard' for radiation transport solutions, are playing an ever-increasing role in correcting and/or verifying the several-decade-old methodology used in current practice. However, the prohibitive computational requirements associated with obtaining fully converged system-wide solutions restrict the role of MC to benchmarking deterministic results at a limited number of state-points for a limited number of relevant quantities. A goal of current research at Oak Ridge National Laboratory (ORNL) is to change this paradigm by enabling the direct use of MC for full-core reactor analyses. The most significant of the many technical challenges that must be overcome is the slow non-uniform convergence of system-wide MC estimates and the memory requirements associated with detailed solutions throughout a reactor (problems involving hundreds of millions of different material and tally regions due to fuel irradiation, temperature distributions, and the needs associated with multi-physics code coupling). To address these challenges, research has focused on development in the following two areas: (1) a hybrid deterministic/MC method for determining high-precision fluxes throughout the problem space in k-eigenvalue problems and (2) an efficient MC domain-decomposition algorithm that partitions the problem phase space onto multiple processors for massively parallel systems, with statistical uncertainty estimation. The focus of this paper is limited to the first area mentioned above. It describes the FW-CADIS method applied to variance reduction of MC reactor analyses and provides initial results for calculating group-wise fluxes throughout a generic 2-D pressurized water reactor (PWR) quarter core model.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Scaling Deep Learning on GPU and Knights Landing clusters

The speed of deep neural networks training has become a big bottleneck of deep learning research and development. For example, training GoogleNet by ImageNet dataset on one Nvidia K20 GPU needs 21 days. To speed up the training process, the current deep learning systems heavily rely on the hardware accelerators. However, these accelerators have limited on-chip memory compared with CPUs. To handle large datasets, they need to fetch data from either CPU memory or remote processors. We use both self-hosted Intel Knights Landing (KNL) clusters and multi-GPU clusters as our target platforms. From an algorithm aspect, current distributed machine learning systems are mainly designed for cloud systems. These methods are asynchronous because of the slow network and high fault-tolerance requirement on cloud systems. We focus on Elastic Averaging SGD (EASGD) to design algorithms for HPC clusters. Original EASGD used round-robin method for communication and updating. The communication is ordered by the machine rank ID, which is inefficient on HPC clusters. First, we redesign four efficient algorithms for HPC systems to improve EASGD's poor scaling on clusters. Async EASGD, Async MEASGD, and Hogwild EASGD are faster \textcolor{black}{than} their existing counterparts (Async SGD, Async MSGD, and Hogwild SGD, resp.) in all the comparisons. Finally, we design Sync EASGD, which ties for the best performance among all the methods while being deterministic. In addition to the algorithmic improvements, we use some system-algorithm codesign techniques to scale up the algorithms. By reducing the percentage of communication from 87% to 14%, our Sync EASGD achieves 5.3x speedup over original EASGD on the same platform. We get 91.5% weak scaling efficiency on 4253 KNL cores, which is higher than the state-of-the-art implementation.

60 APPLIED LIFE SCIENCES↗

Charge trapping of Ge-nanocrystals embedded in TaZrO{sub x} dielectric films

Ge-nanocrystals (NCs) were synthesized in amorphous TaZrO{sub x} by thermal annealing of co-sputtered Ge-TaZrO{sub x} layers. Formation of spherical shaped Ge-NCs with small variation of size, areal density, and depth distribution was confirmed by high-resolution transmission electron microscopy. The charge storage characteristics of the Ge-NCs were investigated by capacitance-voltage and constant-capacity measurements using metal-insulator-semiconductor structures. Samples with Ge-NCs exhibit a maximum memory window of 5 V by sweeping the bias voltage from −7 V to 7 V and back. Below this maximum, the width of the memory window can be controlled by the bias voltage. The fitted slope of the memory window versus bias voltage characteristics is very close to 1 for samples with one layer Ge-NCs. A second layer Ge-NCs does not result in a second flat stair in the memory window characteristics. Constant-capacity measurements indicate charge storage in trapping centers at the interfaces between the Ge-NCs and the surrounding materials (amorphous matrix/tunneling oxide). Charge loss occurs by thermal detrapping and subsequent band-to-band tunneling. Reference samples without Ge-NCs do not show any memory window.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

An adaptive Hessian approximated stochastic gradient MCMC method

Bayesian approaches have been successfully integrated into training deep neural networks. One popular family is stochastic gradient Markov chain Monte Carlo methods (SG-MCMC), which have gained increasing interest due to their ability to handle large datasets and the potential to avoid overfitting. Although standard SG-MCMC methods have shown great performance in a variety of problems, they may be inefficient when the random variables in the target posterior densities have scale differences or are highly correlated. Here, we present an adaptive Hessian approximated stochastic gradient MCMC method to incorporate local geometric information while sampling from the posterior. The idea is to apply stochastic approximation (SA) to sequentially update a preconditioning matrix at each iteration. The preconditioner possesses second-order information and can guide the random walk of a sampler efficiently. Instead of computing and saving the full Hessian of the log posterior, we use limited memory of the samples and their stochastic gradients to approximate the inverse Hessian-vector multiplication in the updating formula. Moreover, by smoothly optimizing the preconditioning matrix via SA, our proposed algorithm can asymptotically converge to the target distribution with a controllable bias under mild conditions. To reduce the training and testing computational burden, we adopt a magnitude-based weight pruning method to enforce the sparsity of the network. Our method is user-friendly and demonstrates better learning results compared to standard SG-MCMC updating rules. The approximation of inverse Hessian alleviates storage and computational complexities for large dimensional models. Numerical experiments are performed on several problems, including sampling from 2D correlated distribution, synthetic regression problems, and learning the numerical solutions of heterogeneous elliptic PDE. The numerical results demonstrate great improvement in both the convergence rate and accuracy.

97 MATHEMATICS AND COMPUTING↗

Planning and processing of new or upgraded electric transmission systems in New Mexico

RETA has been requested to identify and prioritize renewable energy resource zones in New Mexico that have a potential to support industry development among renewable energy developers for renewable resource generation projects. Moreover, Senate Memorial 44 requests that RETA identify and prioritize the best viable options for potential transmission corridors to accommodate renewable energy export from New Mexico in accordance with a defined time-line and to convene a working group to submit recommendations to the legislature for establishing a process to streamline procedures for establishing renewable energy transmission projects in New Mexico. RETA's July 9, 2009 request for comments outlined seven topical areas of specific value to Senate Memorial 44. This document addresses Topics 1, 3 and 4.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Memory operation mechanism of fullerene-containing polymer memory

The memory operation mechanism in fullerene-containing nanocomposite gate insulators was investigated while varying the kind of fullerene in a polymer gate insulator. It was cleared what kind of traps and which positions in the nanocomposite the injected electrons or holes are stored in. The reason for the difference in the easiness of programming was clarified taking the role of the charging energy of an injected electron into account. The dependence of the carrier dynamics on the kind of fullerene molecule was investigated. A nonuniform distribution of injected carriers occurred after application of a large magnitude programming voltage due to the width distribution of the polystyrene barrier between adjacent fullerene molecules. Through the investigations, we demonstrated a nanocomposite gate with fullerene molecules having excellent retention characteristics and a programming capability. This will lead to the realization of practical organic memories with fullerene-containing polymer nanocomposites.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Efficient Distributed Sequence Parallelism for Transformer-Based Image Segmentation

We introduce an efficient distributed sequence parallel approach for training transformer-based deep learning image segmentation models. The neural network models are comprised of a combination of a Vision Transformer encoder with a convolutional decoder to provide image segmentation mappings. The utility of the distributed sequence parallel approach is especially useful in cases where the tokenized embedding representation of image data are too large to fit into standard computing hardware memory. To demonstrate the performance and characteristics of our models trained in sequence parallel fashion compared to standard models, we evaluate our approach using a 3D MRI brain tumor segmentation dataset. We show that training with a sequence parallel approach can match standard sequential model training in terms of convergence. Furthermore, we show that our sequence parallel approach has the capability to support training of models that would not be possible on standard computing resources.

Lyngaas, Isaac↗

Targeted deletion of HYBID (hyaluronan binding protein involved in hyaluronan depolymerization/ KIAA1199/CEMIP) decreases dendritic spine density in the dentate gyrus through hyaluronan accumulation

Highlights: • Hyaluronan was accumulated in the dentate gyrus of Hybid KO mice. • Newly born neurons were decreased in the dentate gyrus of Hybid KO mice. • Dendritic spine density were decreased in the dentate gyrus neuron of Hybid KO mice. HYBID (hyaluronan binding protein involved in hyaluronan [HA] depolymerization, KIAA1199/CEMIP) is a key player in HA depolymerization of the skin fibroblasts, arthritic synovial fibroblasts, and brain. Our previous study demonstrated that Hybid knock-out (KO) mice showed spatial memorial impairment, which is accompanied by the accumulation of high molecular weight HA in the hippocampus. However, the mechanism underlying cognitive impairment by Hybid deficiency remains unclear. In the present study, we examined the HA distribution patterns in the brains of wild-type (WT) and Hybid KO mice by HA staining using HA binding protein, and found that in Hybid KO mice, HA is accumulated and doublecortin-positive immature neurons are significantly decreased in the dentate gyrus of the hippocampus, where Hybid mRNA is highly expressed in WT mice. The Golgi-Cox staining demonstrated that the dendritic spine density is significantly decreased in the dentate gyrus granule cells in Hybid KO mice. These results suggest that Hybid-mediated HA degradation is critical for the synaptic formation process by contributing to cognitive functions, such as learning and memory, in the mouse brain.

60 APPLIED LIFE SCIENCES↗

Effect of Annealing on Perpendicular Magnetic Anisotropy and Low Saturation Magnetization of MgO/Co{sub 2}FeAl/Mo Trilayer Films

In this study, we report our approach to manipulate the strength of perpendicular magnetic anisotropy (PMA) at oxide/ferromagnetic metal of Heusler alloy (HA) interface by tuning the parameter of annealing temperature. Robust PMA energy density (K{sub eff}) of ∼ 0.43 × 10{sup 6} erg/cm{sup 3} and small saturation magnetization (M{sub s}) as low as ∼ 248 emu/cm{sup 3} have been achieved for MgO/Co{sub 2}FeAl (CFA)/Mo stacks at 400 {sup ∘}C. We identify that PMA energy is thermally stable up to the maximum temperature of 400 {sup ∘}C. Effect of annealing temperature on surface morphology, grain distribution, and surface roughness of the stacks are investigated by atomic force microscopy (AFM). Further, X-ray photoelectron spectroscopy (XPS) study demonstrates that the interfacial oxidation decides the strong PMA property at MgO/Co{sub 2}FeAl interface during the annealing process. This MgO/CFA/Mo trilayer system can be a suitable candidate for the development of low-power consumption, huge-density data storage, high-thermal stability against the annealing temperature, and nonvolatile magnetic memories.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Dynamic Simulation with on the fly Doppler Broadening treatment in RMC

In conventional transient analysis, thermal hydraulics calculations are generally coupled to deterministic neutronics solvers. Due to the limited accuracy of deterministic methods and the rapid development of computer power, Monte Carlo neutronics calculations have received increasing attention nowadays. One of challenges for Monte Carlo methods coupled to thermal hydraulics calculations is the temperature dependent cross sections. Taking the detailed temperature distribution feedback into account, each nuclide may be in a large range of temperatures, pre-generated libraries at every temperature point are no longer applicable owing to unaffordable memory footprint. Different solutions have been proposed to deal with this problem. TMS, accounting for Doppler broadening effect by the explicit treatment of thermal motion, was used to model fast transients in Serpent 2. Stochastic mixing, using two sets of libraries for each nuclide and mixing them to get the correct temperature, was used in transient analysis by Sjenitzer and Hoogenboom. Besides, there are other on the fly Doppler broadening methods being developed, including series expansion of cross sections, multipole representation, different integration algorithms based on Doppler broadening equation. In this paper, on the fly Doppler broadening treatment based on integration algorithms was applied to transient calculation in Reactor Monte Carlo code RMC . In the following section, the integration algorithms implemented and dynamic simulation method in RMC is briefly introduced. Finally, a pure physical transient fuel assembly model was chosen for numerical validation. (authors)

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Technical note: Using long short-term memory models to fill data gaps in hydrological monitoring networks

Abstract. Quantifying the spatiotemporal dynamics in subsurface hydrological flows over a long time window usually employs a network of monitoring wells. However, such observations are often spatially sparse with potential temporal gaps due to poor quality or instrument failure. In this study, we explore the ability of recurrent neural networks to fill gaps in a spatially distributed time-series dataset. We use a well network that monitors the dynamic and heterogeneous hydrologic exchanges between the Columbia River and its adjacent groundwater aquifer at the U.S. Department of Energy's Hanford site. This 10-year-long dataset contains hourly temperature, specific conductance, and groundwater table elevation measurements from 42 wells with gaps of various lengths. We employ a long short-term memory (LSTM) model to capture the temporal variations in the observed system behaviors needed for gap filling. The performance of the LSTM-based gap-filling method was evaluated against a traditional autoregressive integrated moving average (ARIMA) method in terms of error statistics and accuracy in capturing the temporal patterns of river corridor wells with various dynamics signatures. Our study demonstrates that the ARIMA models yield better average error statistics, although they tend to have larger errors during time windows with abrupt changes or high-frequency (daily and subdaily) variations. The LSTM-based models excel in capturing both high-frequency and low-frequency (monthly and seasonal) dynamics. However, the inclusion of high-frequency fluctuations may also lead to overly dynamic predictions in time windows that lack such fluctuations. The LSTM can take advantage of the spatial information from neighboring wells to improve the gap-filling accuracy, especially for long gaps in system states that vary at subdaily scales. While LSTM models require substantial training data and have limited extrapolation power beyond the conditions represented in the training data, they afford great flexibility to account for the spatial correlations, temporal correlations, and nonlinearity in data without a priori assumptions. Thus, LSTMs provide effective alternatives to fill in data gaps in spatially distributed time-series observations characterized by multiple dominant frequencies of variability, which are essential for advancing our understanding of dynamic complex systems.

54 ENVIRONMENTAL SCIENCES↗

A Scalable Pipeline for Gigapixel Whole Slide Imaging Analysis on Leadership Class HPC Systems

Whole Slide Imaging (WSI) captures microscopic details of a patient's histopathological features at multiple res-olutions organized across different levels. Images produced by WSI are gigapixel-sized, and saving a single image in memory requires a few gigabytes which is scarce since a complicated model occupies tens of gigabytes. Performing a simple met-ric operation on these large images is also expensive. High-performance computing (HPC) can help us quickly analyze such large images using distributed training of complex deep learning models. One popular approach in analyzing these images is to divide a WSI image into smaller tiles (patches) and then train a simpler model with these reduced-sized but large numbers of patches. However, we need to solve three pre-processing challenges efficiently for pursuing this patch-based approach. 1) Creating small patches from a high-resolution image can result in a high number (hundreds of thousands per image) of patches. Storing and processing these images can be challenging due to a large number of I/O and arithmetic operations. To reduce I/Oand memory accesses, an optimal balance between the size and number of patches must exist to reduce I/O and memory accesses. 2) WSI images may have tiny annotated regions for cancer tissue and a significant portion with normal and fatty tissues; correct patch sampling should avoid dataset imbalance. 3) storing and retrieving many patches to and from disk storage might incur I/O latency while training a deep learning model. An efficient distributed data loader should reduce I/O latency during the training and inference steps. This paper explores these three challenges and provides empirical and algorithmic solutions deployed on the Summit supercomputer hosted at the Oak Ridge Leadership Computing Facility.

Dash, Sajal↗