Using HPC for AI/ML
A presentation providing information on why using HPC for AI/ML is valuable as well as how INL HPC can be used to perform AI/ML at scale.
SEARCH · Engineering Papers
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.
A presentation providing information on why using HPC for AI/ML is valuable as well as how INL HPC can be used to perform AI/ML at scale.
Nuclear magnetic resonance (NMR) spectroscopy and infrared (IR) spectroscopy are powerful chemical characterization techniques with broad general usage. However, the manual evaluation of the resulting spectra is time-consuming and requires significant expertise, preventing insights from being used in real-time applications. With recent advances in computation and artificial intelligence (AI), new tools are available for automating spectral interpretation. In this work, machine learning (ML) algorithms using 1-dimensional convolutional neural networks (CNNs) were applied to identify common functional groups from spectral information. Raw spectra were collected virtually from the Human Metabolome Database (HMDB) and National Institute of Standards and Technology (NIST) Chemistry WebBook and processed into a suitable standard. Algorithm design was tailored to best fit the nature of the problem, with built-in flexibility to accommodate relevant parameters beyond the raw spectral input, specifically solvent identity and magnetic frequency for NMR. The predictive capability of the algorithm in identifying functional groups is displayed in several examples. This methodology has been compiled into a code repository and could easily be modified to adapt alternative data sources, including other spectrum types. To mitigate overfitting, a common problem in mathematical modeling where overfamiliarity with training data produces trends that are not representative of the general data, a novel metric was developed, referred to as Accufit. Accufit includes a parameter that penalizes substantial differences in the training accuracy and the accuracy of an independent validation set. Examples are presented showing the effectiveness of Accufit in maintaining the model’s predictive capability while controlling the overfitting when used as a custom metric for hyperparameter tuning.
he exponential growth of data centers—driven by artificial intelligence and cloud computing—is reshaping the U.S. energy landscape, presenting urgent challenges and transformative opportunities for data center developers, nuclear energy providers, and utility operators. As data centers are projected to consume up to 12% of U.S. electricity by 2028, stakeholders must address rapid deployment needs, grid congestion, and the demand for reliable, high-quality power. This presentation explores the multifaceted barriers to integrating data centers with nuclear and utility infrastructure, including land use constraints, public perception, regulatory complexity, and workforce alignment. It highlights the distinct priorities and operational cultures of each sector, and the friction that arises from misaligned planning horizons and risk tolerances. We examine collaborative strategies such as co-siting, hybrid power-purchase agreements, unified community engagement, and innovative financing models.
The present invention provides methods and devices for the knowledge-based discovery and optimization of differences between cell types. In particular, the present invention provides visual servoing optical microscopy, as well as analysis methods. The present invention provides means for the close monitoring of hundreds of individual, living cells over time: quantification of dynamic physiological responses in multiple channels; real-time digital image segmentation and analysis; intelligent, repetitive computer-applied cell stress and cell stimulation; and the ability to return to the same field of cells for long-term studies and observation. The present invention further provides means to optimize culture conditions for specific subpopulations of cells.
The four priority research directions outlined in this brochure represent a cohesive vision for advancing the science of inverse problems for complex systems under uncertainty. Together, they address the critical challenges of: discovering, exploiting, and preserving physical and problem structure; overcoming model limitations; integrating disparate, multimodal, and/or dynamic data; and tailoring the solution of inverse problems to downstream tasks. While each PRD focuses on a distinct aspect of inverse-problem research, their interconnected nature highlights the importance of a holistic approach that leverages progress across all areas to achieve transformative solutions. This agenda calls for research across mathematics, statistics, and computer science disciplines, which are guided and complemented by rapid advances in artificial intelligence, high-performance computing, and experimental facilities, to unlock new capabilities, maximize scientific impact, and meet the growing demands of inverse problems that arise across applications that are critical to DOE's mission.
This perspective sheds light on the transformative impact of recent computational advancements in the field of protein therapeutics, with a particular focus on the design and development of antibodies. Cutting-edge computational methods have revolutionized our understanding of protein–protein interactions (PPIs), enhancing the efficacy of protein therapeutics in preclinical and clinical settings. Central to these advancements is the application of machine learning and deep learning, which offers unprecedented insights into the intricate mechanisms of PPIs and facilitates precise control over protein functions. Despite these advancements, the complex structural nuances of antibodies pose ongoing challenges in their design and optimization. Our review provides a comprehensive exploration of the latest deep learning approaches, including language models and diffusion techniques, and their role in surmounting these challenges. We also present a critical analysis of these methods, offering insights to drive further progress in this rapidly evolving field. The paper includes practical recommendations for the application of these computational techniques, supplemented with independent benchmark studies. These studies focus on key performance metrics such as accuracy and the ease of program execution, providing a valuable resource for researchers engaged in antibody design and development. Through this detailed perspective, we aim to contribute to the advancement of antibody design, equipping researchers with the tools and knowledge to navigate the complexities of this field.
We live in momentous times. The science community is empowered with an arsenal of cosmic messengers to study the universe in unprecedented detail. Gravitational waves, electromagnetic waves, neutrinos, and cosmic rays cover a wide range of wavelengths and timescales. Combining and processing these datasets that vary in volume, speed, and dimensionality requires new modes of instrument coordination, funding, and international collaboration with a specialized human and technological infrastructure. In tandem with the advent of large-scale scientific facilities, the last decade has experienced an unprecedented transformation in computing and signal-processing algorithms. The combination of graphics processing units, deep learning, and the availability of open source, high-quality datasets has powered the rise of artificial intelligence. This digital revolution now powers a multibillion dollar industry, with far-reaching implications in technology and society. In this chapter, we describe pioneering efforts to adapt artificial intelligence algorithms to address computational grand challenges in multi-messenger astrophysics. We review the rapid evolution of these disruptive algorithms, from the first class of algorithms introduced in early 2017 to the sophisticated algorithms that now incorporate domain expertise in their architectural design and optimization schemes. We discuss the importance of scientific visualization and extreme-scale computing in reducing time-to-insight and obtaining new knowledge from the interplay between models and data.
This paper proposes an efficient data memory management approach for the Intelligent RuntIme System (IRIS) heterogeneous computing framework along with new data transfer policies. IRIS provides a task-based programming model for extreme heterogeneous computing (e.g., CPU, GPU, DSP, FPGA) with support for today's most important programming languages (e.g., OpenMP, OpenCL, CUDA, HIP, OpenACC). However, the IRIS framework either forces the programmer to introduce data transfer commands for each task or relies on suboptimal memory management for automatic and transparent data transfers. The work described here extends IRIS with novel heterogeneous memory handling and introduces novel data transfer policies by employing the Distributed data MEMory handler (DMEM) for efficient and optimal movement of data among the various computing resources. The proposed approach achieves performance gains of up to 7× for tiled LU factorization and tiled DGEMM (i.e., matrix multiplication) benchmarks. Moreover, this approach also reduces data transfers by up to 71% when compared to previous IRIS heterogeneous memory management handlers. This work compares the performance results of the IRIS framework's novel DMEM with the StarPU runtime and MAGMA math library for GPUs. Experiments show a performance gain of up to 1.95× over StarPU and 2.1× over MAGMA.
Recent advances in cryo-electron microscopy (cryo-EM) have enabled modeling macromolecular complexes that are essential components of the cellular machinery. The density maps derived from cryo-EM experiments are often integrated with manual, knowledge or artificial intelligence driven, and physics-guided computational methods to build, fit, and refine molecular structures. Going beyond a single stationary- structure determination scheme, it is becoming more common to interpret the experimental data with an ensemble of models, which contributes to an average observation. Hence, there is a need to decide on the quality of an ensemble of protein structures on-the-fly, while refining them against the density maps. Here, we introduce such an adaptive decision making scheme during the molecular dynamics flexible fitting (MDFF) of biomolecules. Using RADICAL-Cybertools, and the new RADICAL augmented MDFF implementation (R-MDFF) is examined in high-performance computing environments for refinement of two protein systems, Adenylate Kinase and Carbon Monoxide Dehydrogenase. For the test cases, use of multiple replicas in flexible fitting with adaptive decision making in R-MDFF improves the overall correlation to the density by 40% relative to the refinements of the brute-force MDFF. The improvements are particularly significant at high, 2 - 3 Å, map resolutions. More importantly, the ensemble model captures key features of biologically relevant molecular dynamics that is inaccessible to a single-model interpretation. Finally, the pipeline is applicable to systems of growing sizes, which is demonstrated using ensemble refinement of capsid proteins from Chimpanzee adenovirus. The overhead for decision making remaining low and robust to computing environments. The software is publicly available on GitHub and includes a short user guide to install the R-MDFF on different computing environments, from local Linux based workstations to High Performance Computing (HPC) environments.
MTTKRP is the bottleneck operation in algorithms used to compute the CP tensor decomposition. For sparse tensors, utilizing the compressed sparse fibers (CSF) storage format and the CSF-oriented MTTKRP algorithms is important for both memory and computational efficiency on distributed-memory architectures. Existing intelligent tensor partitioning models assume the computational cost of MTTKRP to be proportional to the total number of nonzeros in the tensor. However, this is not the case for the CSF-oriented MTTKRP on distributed-memory architectures. We outline two deficiencies of nonzero-based intelligent partitioning models when CSF-oriented MTTKRP operations are performed locally: failure to encode processors' computational loads and increase in total computation due to fiber fragmentation. We focus on existing fine-grain hypergraph model and propose a novel vertex weighting scheme that enables this model encode correct computational loads of processors. We also propose to augment the fine-grain model by fiber nets for reducing the increase in total computational load via minimizing fiber fragmentation. In this way, the proposed model encodes minimizing the load of the bottleneck processor. In conclusion, parallel experiments with real-world sparse tensors on up to 1024 processors prove the validity of the outlined deficiencies and demonstrate the merit of our proposed improvements in terms of parallel runtimes.
Geometric partitioning is fast and effective for load-balancing dynamic applications, particularly those requiring geometric locality of data (particle methods, crash simulations). We present, to our knowledge, the first parallel implementation of a multidimensional-jagged geometric partitioner. In contrast to the traditional recursive coordinate bisection algorithm (RCB), which recursively bisects subdomains perpendicular to their longest dimension until the desired number of parts is obtained, our algorithm does recursive multi-section with a given number of parts in each dimension. By computing multiple cut lines concurrently and intelligently deciding when to migrate data while computing the partition, we minimize data movement compared to efficient implementations of recursive bisection. We demonstrate the algorithm's scalability and quality relative to the RCB implementation in Zoltan on both real and synthetic datasets. Our experiments show that the proposed algorithm performs and scales better than RCB in terms of run-time without degrading the load balance. Lastly, our implementation partitions 24 billion points into 65,536 parts within a few seconds and exhibits near perfect weak scaling up to 6K cores.
As we move towards the exascale era, the new architectures must be capable of running the massive computational problems efficiently. Scientists and researchers are continuously investing in tuning the performance of extreme-scale computational problems. These problems arise in almost all areas of computing, ranging from big data analytics, artificial intelligence, search, machine learning, virtual/augmented reality, computer vision, image/signal processing to computational science and bioinformatics. With Moore’s law driving the evolution of hardware platforms towards exascale, the dominant performance metric (time efficiency) has now expanded to also incorporate power/energy efficiency. Therefore the major challenge that we face in computing systems research is: “how to solve massive-scale computational problems in the most time/power/energy efficient manner?”
The ability to establish accurate occupancy is a key enabler for improved energy efficiency in human-occupied indoor spaces. The accurate counts can be used to adjust, in real-time, air exchange, heating, and cooling, to closely match the environmental needs of the occupants rather than to use a single setting based on maximum room occupancy. This project focuses on the use of overhead fisheye cameras and artificial intelligence (AI) algorithms as the technology to realize accurate people counting in large commercial spaces. We developed Computational Occupancy Sensing SYstem ("COSSY''), which supports large-scale deployment (thousands of square feet) and straightforward installation (using conventional power over ethernet), while providing high accuracy and low cost. COSSY has been validated at Boston University and by a third party, demonstrating people-counting accuracy of at least 89.5% in large spaces (2,000 sqft). A cost/benefit analysis performed for Boston commercial market indicates that a full payback of COSSY deployment costs due to energy savings would occur within 3-6 months from installation. Applications of COSSY reach far beyond saving energy. The precise people counting in real time could also be useful for emergency response (e.g., fire), in spatial analytics (e.g., office space management) and in some health applications (e.g., social distance assessment during pandemics).
Resource constraints on the computing continuum require that we make smart decisions for serving AI-based services at the network edge. AI-based services typically have multiple implementations (e.g., image classification implementations include SqueezeNet, DenseNet, and others) with varying trade-offs (e.g., latency and accuracy). The question then is how should AI-based services be placed across Function-as-a-Service (FaaS) based edge computing systems in order to maximize total Quality-of-Service (QoS). To address this question, we propose a problem that jointly aims to solve (i) edge AI service placement and (ii) request scheduling. These are done across two time-scales (one for placement and one for scheduling). Here we first cast the problem as an integer linear program. We then decompose the problem into separate placement and scheduling subproblems and prove that both are NP-hard. We then propose a novel placement algorithm that places services while considering device-to-device communication across edge clouds to offload requests to one another. Our results show that the proposed placement algorithm is able to outperform a state-of-the-art placement algorithm for AI-based services, and other baseline heuristics, with regard to maximizing total QoS. Additionally, we present a federated learning-based framework, FLIES, to predict the future incoming service requests and their QoS requirements. Our results also show that our FLIES algorithm is able to outperform a standard decentralized learning baseline for predicting incoming requests and show comparable predictive performance when compared to centralized training.
Heterogeneous integration (HI) of electronics components is broadly recognized as a powerful and crucial enabler for the continued growth of computing and communication. From 2010 onwards, the value of HI is increasingly visible in the advanced packaging used in artificial intelligence, high-performance computing, smartphones and communications product implementations. In this Perspective, we argue that HI is crucial to semiconductors and more broadly to the continued evolution of computing and communications. We use leading-edge advanced packaging examples to represent the value, advancements and opportunities for HI. To succeed, it is critical to develop comprehensive HI roadmaps that inform collaborations across the design, manufacturing and reliability spectrum between systems architects, packaging and semiconductor technologists to common goals. Although this article does not provide a full roadmap, we instead detail additional parameters for artificial intelligence, smartphone and other cellular communication devices, and their constituent building blocks including interconnects, power electronics, photonics, thermal management, reliability, modelling and co-design, to foster greater collaboration opportunities among academia, research laboratories and industry.
The exponential growth in computing demands driven by scientific computing, data analytics, and artificial intelligence is pushing conventional CMOS-based high-performance computing systems to their physical and energy efficiency limits. As we approach the era of post-exascale computing, disruptive approaches are necessary to overcome these barriers and achieve substantial gains in energy efficiency. Analog and hybrid digital-analog computing systems have emerged as promising alternatives, offering the potential for orders-of-magnitude improvements in efficiency. Among these, biochemical computing stands out as a novel paradigm capable of leveraging the natural efficiency of chemical reactions, which have shown promise in solving optimization problems by converging to steady states. By scaling up reaction networks or reaction vessel sizes, biochemical systems present an opportunity to meet the high-performance demands of modern computing tasks. Despite their promise, significant theoretical and practical challenges remain, particularly in formulating and mapping computational problems to chemical reaction networks (CRNs) and designing viable biochemical computing devices. This paper addresses these challenges by introducing new ideas to ChemComp, a compilation and emulation framework for chemical computation. This work describes the mechanisms through which solutions to ordinary differential equations (ODEs) that can be represented as CRN systems can be achieved. Furthermore, we explain the design principles of an ODE dialect implemented as a multi-level intermediate representation (MLIR) compiler extension that will be coupled with existing infrastructure. We demonstrate the potential of our framework through a case study emulating a simplified chemical reservoir computing device. This work establishes foundational tools and methodologies necessary to harness the computational power of chemistry, paving the way for the development of energy-efficient, high-performance computing systems tailored to contemporary and future computational needs.
Artificial intelligence (AI) and machine learning (ML) have been shown to be increasingly helpful tools in a growing number of use-cases relevant to scientific research, despite significant software-related obstacles. There exist large technical costs to setting up, using, and maintaining AI/ML models in production. This often prevents researchers from utilizing these models in their work. The growing field of machine learning operations (MLOps) aims to automate much of the AI/ML life cycle while increasing access to these models. This paper presents our initial work in creating a nuclear energy MLOps platform for use by researchers at Idaho National Laboratory (INL) and aims to reduce the barriers of using AI/ML in scientific research. We hope to promote the integration of the latest AI/ML technologies into researchers' workflows and create more opportunity for scientific innovation. In this paper we discuss how our MLOps efforts aim to increase usage and the impact of AI/ML models created by engineers. We also present several use-cases that are currently integrated. Finally, we evaluate the maturity of our project as well as our plans for future functionality.
Computational workflows are a common class of application on supercomputers, yet the loosely coupled and heterogeneous nature of workflows often fails to take full advantage of their capabilities. We created Colmena to leverage the massive parallelism of a supercomputer by using Artificial Intelligence (AI) to learn from and adapt a workflow as it executes. Colmena allows scientists to define how their application should respond to events (e.g., task completion) as a series of cooperative agents. In this paper, we describe the design of Colmena, the challenges we overcame while deploying applications on exascale systems, and the science workflows we have enhanced through interweaving AI. The scaling challenges we discuss include developing steering strategies that maximize node utilization, introducing data fabrics that reduce communication overhead of data-intensive tasks, and implementing workflow tasks that cache costly operations between invocations. These innovations coupled with a variety of application patterns accessible through our agent-based steering model have enabled science advances in chemistry, biophysics, and materials science using different types of AI. In conclusion, our vision is that Colmena will spur creative solutions that harness AI across many domains of scientific computing.