Search

"Derecho" is the most powerful supercomputer at the National Center for Atmospheric Research (NCAR), running at a peak speed of 19.87 petaflops and ranking as the 59th fastest supercomputer in the world. It supports a wide range of atmospheric science research, including climate modeling, weather forecasting, and space weather analysis. Derecho's heterogenous architecture, combining CPUs and GPUs, allows it to efficiently handle physics-based simulations and machine learning applications. The supercomputer is housed in the NCAR-Wyoming Supercomputing Center in Cheyenne, Wyoming, and requires continuous maintenance and monitoring to ensure optimal performance for researchers worldwide. [Extracted from the article]

Summary: A memory cube network is an interconnection network composed of 3D stacked memories called memory cubes. By exploiting a packet switching, it can provide fast memory accesses to a large number of memory cubes. Although interconnection networks have been studied in many years for supercomputers and data centers, existing technologies are difficult to apply to memory cube networks. This is because the link length and the number of ports are limited, and hence the hop count increases. In this article, we propose a dual diagonal mesh (DDM), a layout‐oriented memory‐cube network. Furthermore, we propose the routing algorithm and the router architecture with bypassed pipelines for DDM. Our experimental results demonstrate that our routing and router architecture with bypassed pipelines reduces the memory access latency. We implement four router architectures and evaluate them with the traffic patterns derived from the NAS parallel benchmark. [ABSTRACT FROM AUTHOR]

Networks are designed to communicate, operate, and allocate tasks to respective commodities. Operating supercomputers became challenging, which was handled by the network design commonly known as hypercube, denoted by Q n . In a recent study, the hypercube networks were insufficient to hold supercomputers' parallel processors. Thus, variants of hypercubes were discovered to produce an alternative to the hypercube. A new variant of the hypercube, the fractal cubic network, can be used as the best alternative in the case of hypercubes. Our research investigates that the fractal cubic network is a rooted product of two graphs. We try to determine its domination and resolving domination parameters, which could be applied to resource location and broadcasting-related problems. [ABSTRACT FROM AUTHOR]

Recent advances in cloud computing, large language models, and deep learning have started a race to create massive High-Performance Computing (HPC) centers worldwide. These centers increase in energy consumption proportionally to their computing capabilities; for example, according to the top 500 organization, the HPC centers Frontier, Aurora, and Super Computer Fugaku report energy consumptions of 22,786 kW, 38,698 kW, and 29,899 kW, respectively. Currently, energy-aware scheduling is a topic of interest to many researchers. However, as far as we know, this work is the first approach considering the idle energy consumption by the HPC units and the possibility of turning off unused units entirely, driven by a quantitative objective function. We found that even when turning off unused machines, the objectives of makespan and energy consumption still conflict and, therefore, their multi-objective optimization nature. This work presents empirical results for AGEMOEA, AGEMOEA2, GWASFGA, MOCell, MOMBI, MOMBI2, NSGA2, and SMS-EMOA. The best-performing algorithm is MOCell for the 400 real scheduling problem tests. In contrast, the best-performing algorithm is GWASFGA for a small-instance synthetic testbed. [ABSTRACT FROM AUTHOR]

This project introduces a metric-driven strategy for enhancing real-time task planning within a mobile distributed system. The core design revolves around a realtime task planning algorithm that exploits consensus mechanisms among multiple nodes. By incorporating innovative real-time scheduling algorithms, validated on a supercomputer, the project establishes frameworks for simulating distributed mobile environments in real-time. This enables the execution of tasks in unpredictable contexts, ensuring adherence to stringent time constraints and empowering decision-making capabilities. The primary objective is to optimize task allocation in mobile distributed systems, including aerial nodes, thereby facilitating seamless data transfer while maintaining data integrity. The project builds upon the Fan task scheduler, integrating compensatory measures for online load distribution and optimization of message routes, effectively reducing communication latency in dynamic mobile environments. The contributions made aim to enhance real-time task allocation across diverse nodes, enabling the transfer of location and search data across networks without any data loss. This approach bears significant importance for the efficient management of tasks within dynamic scenarios. [ABSTRACT FROM AUTHOR]

Electronic structure calculations have the potential to predict key matter transformations for applications of strategic technological importance, from drug discovery to material science and catalysis. However, a predictive physicochemical characterization of these processes often requires accurate quantum chemical modeling of complex molecular systems with hundreds to thousands of atoms. Due to the computationally demanding nature of electronic structure calculations and the complexity of modern high-performance computing hardware, quantum chemistry software has historically failed to operate at such large molecular scales with accuracy and speed that are useful in practice. In this paper, novel algorithms and software are presented that enable extreme-scale quantum chemistry capabilities with particular emphasis on exascale calculations. This includes the development and application of the multi-Graphics Processing Unit (GPU) library LibCChem 2.0 as part of the General Atomic and Molecular Electronic Structure System package and of the standalone Extreme-scale Electronic Structure System (EXESS), designed from the ground up for scaling on thousands of GPUs to perform high-performance accurate quantum chemistry calculations at unprecedented speed and molecular scales. Among various results, we report that the EXESS implementation enables Hartree–Fock/cc-pVDZ plus RI-MP2/cc-pVDZ/cc-pVDZ-RIFIT calculations on an ionic liquid system with 623 016 electrons and 146 592 atoms in less than 45 min using 27 600 GPUs on the Summit supercomputer with a 94.6% parallel efficiency. [ABSTRACT FROM AUTHOR]

With the growing reliance of modern supercomputers on accelerator-based architecture such a graphics processing units (GPUs), the development and optimization of electronic structure methods to exploit these massively parallel resources has become a recent priority. While significant strides have been made in the development GPU accelerated, distributed memory algorithms for many modern electronic structure methods, the primary focus of GPU development for Gaussian basis atomic orbital methods has been for shared memory systems with only a handful of examples pursing massive parallelism. In the present work, we present a set of distributed memory algorithms for the evaluation of the Coulomb and exact exchange matrices for hybrid Kohn–Sham DFT with Gaussian basis sets via direct density-fitted (DF-J-Engine) and seminumerical (sn-K) methods, respectively. The absolute performance and strong scalability of the developed methods are demonstrated on systems ranging from a few hundred to over one thousand atoms using up to 128 NVIDIA A100 GPUs on the Perlmutter supercomputer. [ABSTRACT FROM AUTHOR]

As part of the US Department of Energy's Exascale Computing Project (ECP), Rice University has been refining its HPCToolkit performance tools to better support measurement and analysis of applications executing on exascale supercomputers. To efficiently collect performance measurements of GPU-accelerated applications, HPCToolkit employs novel non-blocking data structures to communicate performance measurements between tool threads and application threads. To attribute performance information in detail to source lines, loop nests, and inlined call chains, HPCToolkit performs parallel analysis of large CPU and GPU binaries involved in the execution of an exascale application to rapidly recover mappings between machine instructions and source code. To analyze terabytes of performance measurements gathered during executions at exascale, HPCToolkit employs distributed-memory parallelism, multithreading, sparse data structures, and out-of-core streaming analysis algorithms. To support interactive exploration of profiles up to terabytes in size, HPCToolkit's hpcviewer graphical user interface uses out-of-core methods to visualize performance data. The result of these efforts is that HPCToolkit now supports collection, analysis, and presentation of profiles and traces of GPU-accelerated applications at exascale. These improvements have enabled HPCToolkit to efficiently measure, analyze and explore terabytes of performance data for executions using as many as 64K MPI ranks and 64K GPU tiles on ORNL's Frontier supercomputer. HPCToolkit's support for measurement and analysis of GPU-accelerated applications has been employed to study a collection of open-science applications developed as part of ECP. This paper reports on these experiences, which provided insight into opportunities for tuning applications, strengths and weaknesses of HPCToolkit itself, as well as unexpected behaviors in executions at exascale. [ABSTRACT FROM AUTHOR]

The achievement of quantum computational advantage, also known as quantum supremacy, is a major milestone at which a quantum computer can solve a problem significantly faster than the world's most powerful classical computers. Two tasks, boson sampling and random quantum circuit sampling, have experimentally exhibited quantum advantages on photonic and superconducting platforms respectively. Classical benchmarking is essential, yet challenging, because these tasks are intractable for classical computers. This study reviews models, algorithms and large‐scale simulations of these two sampling tasks. These approaches continue to hold substantial significance for research in both current noisy intermediate‐scale quantum (NISQ) systems and future fault‐tolerant quantum computing. [ABSTRACT FROM AUTHOR]

Summary: NERSC's newest system, Perlmutter, features a 35 PB all‐flash Lustre file system built on HPE Cray ClusterStor E1000. We present its architecture, early performance figures, and performance considerations unique to this architecture. We demonstrate the performance of E1000 OSSes through low‐level Lustre tests that achieve over 90% of the theoretical bandwidth of the SSDs at the OST and LNet levels. We also show end‐to‐end performance for both traditional dimensions of I/O performance (peak bulk‐synchronous bandwidth) and nonoptimal workloads endemic to production computing (small, incoherent I/Os at random offsets) and compare them to NERSC's previous system, Cori, to illustrate that Perlmutter achieves the performance of a burst buffer and the resilience of a scratch file system. Finally, we discuss performance considerations unique to all‐flash Lustre and present ways in which users and HPC facilities can adjust their I/O patterns and operations to make optimal use of such architectures. [ABSTRACT FROM AUTHOR]

In this study, we develop new reverse engineering (RE) techniques to identify the organization of the synaptic inputs generating firing patterns of populations of neurons. We tested these techniques in silico to allow rigorous evaluation of their effectiveness, using remarkably extensive parameter searches enabled by massively-parallel computation on supercomputers. We chose spinal motoneurons as our target neural system, since motoneurons process all motor commands and have well-established input-output properties. One set of simulated motoneurons was driven by 300,000+ simulated combinations of excitatory, inhibitory, and neuromodulatory inputs. Our goal was to determine if these firing patterns had sufficient information to allow RE identification of the input combinations. Like other neural systems, the motoneuron input-output system is likely non-unique. This non-uniqueness could potentially limit this RE approach, as many input combinations can produce similar outputs. However, our simulations revealed that firing patterns contained sufficient information to sharply restrict the solution space. Thus, our RE approach successfully generated estimates of the actual simulated patterns of excitation, inhibition, and neuromodulation, with variances accounted for ranging from 75-90%. It was striking that nonlinearities induced in firing patterns by the neuromodulation inputs did not impede RE, but instead generated distinctive features in firing patterns that aided RE. These simulations demonstrate the potential of this form of RE analysis. It is likely that the ever-increasing capacity of supercomputers will allow increasingly accurate RE of neuron inputs from their firing patterns from many neural systems. [ABSTRACT FROM AUTHOR]

The InfiniBand (IB) interconnection technology is widely used in the networks of modern supercomputers and data centers. Among other advantages, the IB-based network devices allow for building multiple network topologies, and the IB control software (subnet manager) supports several routing engines suitable for the most common topologies. However, the implementation of some novel topologies in IB-based networks may be difficult if suitable routing algorithms are not supported, or if the IB switch or NIC architectures are not directly applicable for that topology. This work describes the implementation of the network topology known as KNS in a real HPC cluster using an IB network. As far as we know, this is the first implementation of this topology in an IB-based system. In more detail, we have implemented the KNS routing algorithm in the OpenSM software distribution of the subnet manager, and we have adapted the available IB-based switches to the particular structure of this topology. We have evaluated the correctness of our implementation through experiments in the real cluster, using well-known benchmarks. The obtained results, which match the expected performance for the KNS topology, show that this topology can be implemented in IB-based clusters as an alternative to other interconnection patterns. [ABSTRACT FROM AUTHOR]

The increasing trend of manycore processors makes multithreaded communication more important to avoid costly global synchronization among cores. One of the representative approaches that require multithreaded communication is the global task-based programming model. In the model, a program is divided into tasks, and tasks are asynchronously executed by each node, and independent thread-to-thread communications are expected. However, the Message passing interface (MPI) based approach is not efficient because of design issues. In this research, we design and implement the utofu transport layer in an abstracted communication library called Unified communication-X (UCX) for efficient remote direct memory access (RDMA) based multithreaded communication on Tofu Interconnect D. The evaluation results on Fugaku show that UCX can significantly improve the multithreaded performance over MPI, while maintaining portability between systems thanks to UCX. UCX shows about 32.8 times lower latency than Fujitsu MPI with 24 threads in the multithreaded pingpong benchmark and about 37.8 times higher update rate than Fujitsu MPI with 24 threads on 256 nodes in multithreaded GUPs benchmark. [ABSTRACT FROM AUTHOR]

The MFEM (Modular Finite Element Methods) library is a high-performance C++ library for finite element discretizations. MFEM supports numerous types of finite element methods and is the discretization engine powering many computational physics and engineering applications across a number of domains. This paper describes some of the recent research and development in MFEM, focusing on performance portability across leadership-class supercomputing facilities, including exascale supercomputers, as well as new capabilities and functionality, enabling a wider range of applications. Much of this work was undertaken as part of the Department of Energy's Exascale Computing Project (ECP) in collaboration with the Center for Efficient Exascale Discretizations (CEED). [ABSTRACT FROM AUTHOR]

Radiative transfer (RT) is a crucial ingredient for self-consistent modelling of numerous astrophysical phenomena across cosmic history. However, on-the-fly integration into radiation hydrodynamics (RHD) simulations is computationally demanding, particularly due to the stringent time-stepping conditions and increased dimensionality inherent in multifrequency collisionless Boltzmann physics. The emergence of exascale supercomputers, equipped with extensive CPU cores and GPU accelerators, offers new opportunities for enhancing RHD simulations. We present the first steps towards optimizing arepo - rt for such high-performance computing environments. We implement a novel node-to-node (n-to-n) communication strategy that utilizes shared memory to substitute intranode communication with direct memory access. Furthermore, combining multiple internode messages into a single message substantially enhances network bandwidth utilization and performance for large-scale simulations on modern supercomputers. The single-message n-to-n approach also improves performance on smaller scale machines with less optimized networks. Furthermore, by transitioning all RT-related calculations to GPUs, we achieve a significant computational speedup of around 15 for standard benchmarks compared to the original CPU implementation. As a case study, we perform cosmological RHD simulations of the Epoch of Reionization, employing a similar setup as the THESAN project. In this context, RT becomes sub-dominant such that even without modifying the core arepo codebase, there is an overall threefold improvement in efficiency. The advancements presented here have broad implications, potentially transforming the complexity and scalability of future simulations for a wide variety of astrophysical studies. Our work serves as a blueprint for porting similar simulation codes based on unstructured resolution elements to GPU-centric architectures. [ABSTRACT FROM AUTHOR]

Quantum computers promise to surpass the most powerful classical supercomputers in tackling critical practical problems, such as designing pharmaceuticals and fertilizers, optimizing supply chains and traffic, and enhancing machine learning. Since quantum computers operate fundamentally differently from classical ones, their emergence will give rise to a new evolutionary branch of statistical and data analytics methodologies. This review aims to provide an introduction to quantum computing accessible to statisticians and data scientists, equipping them with a comprehensive framework, the basic language, and building blocks of quantum algorithms, as well as an overview of existing quantum applications in statistics and data analysis. Our objective is to empower statisticians and data scientists to follow quantum computing literature relevant to their fields, collaborate with quantum algorithm designers, and ultimately drive the development of the next generation of statistical and data analytics tools. [ABSTRACT FROM AUTHOR]

Acceleration of the density-functional tight-binding (DFTB) method on single and multiple graphical processing units (GPUs) was accomplished using the MAGMA linear algebra library. Two major computational bottlenecks of DFTB ground-state calculations were addressed in our implementation: the Hamiltonian matrix diagonalization and the density matrix construction. The code was implemented and benchmarked on two different computer systems: (1) the SUMMIT IBM Power9 supercomputer at the Oak Ridge National Laboratory Leadership Computing Facility with 1–6 NVIDIA Volta V100 GPUs per computer node and (2) an in-house Intel Xeon computer with 1–2 NVIDIA Tesla P100 GPUs. The performance and parallel scalability were measured for three molecular models of 1-, 2-, and 3-dimensional chemical systems, represented by carbon nanotubes, covalent organic frameworks, and water clusters. [ABSTRACT FROM AUTHOR]

It is a challenging but key task to reduce the tritium inventory in EU-DEMO to levels that are acceptable for a nuclear regulator. As solution to this issue, a smart fuel cycle architecture is proposed based on the concept of Direct Internal Recycling (DIR), in which the Metal Foil Pump (MFP) will play an important role to separate the unburnt hydrogen isotopes coming from the divertor by exploiting the superpermeation phenomenon. In this study, we will present the assessment of the performance of the lower port of EU-DEMO after the integration of the MFP. For the first time, a thorough comparison of three different MFP (parallel long tubes, sandwich and halo) designs is performed regarding conductance for helium molecules, the pumping speed and the separation factor for deuterium molecules under different physical and geometric parameters. All simulations were carried out in supercomputer Marconi-Fusion with our in-house Test Particle Monte Carlo (TPMC) simulation code ProVac3D because the code had been parallelized with high efficiency. These results are essential for the development of a suitable MFP design in the vacuum-pumping train of EU-DEMO. [ABSTRACT FROM AUTHOR]

To address the challenge of finding new combination therapies against castration-sensitive prostate cancer, we introduce Vini, a computational tool that predicts the efficacy of drug combinations at the intracellular level by integrating data from the KEGG, DrugBank, Pubchem, Protein Data Bank, Uniprot, NCI-60 and COSMIC databases. Vini is a computational tool that predicts the efficacy of drugs and their combinations at the intracellular level. It addresses the problem comprehensively by considering all known target genes, proteins and small molecules and their mutual interactions involved in the onset and development of cancer. The results obtained point to new, previously unexplored combination therapies that could theoretically be promising candidates for the treatment of castration-sensitive prostate cancer and could prevent the inevitable progression of the cancer to the incurable castration-resistant stage. Furthermore, after analyzing the obtained triple combinations of drugs and their targets, the most common targets became clear: ALK, BCL-2, mTOR, DNA and androgen axis. These results may help to define future therapies against castration-sensitive prostate cancer. The use of the Vini computer model to explore therapeutic combinations represents an innovative approach in the search for effective treatments for castration-sensitive prostate cancer, which, if clinically validated, could potentially lead to new breakthrough therapies. [ABSTRACT FROM AUTHOR]

The increasing availability of machines relying on non-GPU architectures, such as ARM A64FX in high-performance computing, provides a set of interesting challenges to application developers. In addition to requiring code portability across different parallelization schemes, programs targeting these architectures have to be highly adaptable in terms of compute kernel sizes to accommodate different execution characteristics for various heterogeneous workloads. In this paper, we demonstrate an approach to write compute kernels using Kokko's abstraction layer to be executed on x86 and A64FX CPUs and NVIDIA GPUs. In addition to applying Kokkos as an abstraction over the execution of compute kernels on different heterogeneous execution environments, we show that the use of standard C++ constructs, as exposed by the HPX runtime system, enables platform portability based on the real-world Octo-Tiger astrophysics application. We report our experience with porting Octo-Tiger to the ARM A64FX architecture provided by Stony Brook's Ookami and Riken's Supercomputer Fugaku and compare the resulting performance with that achieved on well-established GPU-oriented HPC machines such as ORNL's Summit, NERSC's Perlmutter, and CSCS's Piz Daint systems. Octo-Tiger scaled well on Supercomputer Fugaku without any major code changes due to the abstraction levels provided by HPX and Kokkos. Adding vectorization support for ARM's SVE to Octo-Tiger was trivial thanks to using standard C++ interfaces. [ABSTRACT FROM AUTHOR]

Summary: Parallel I/O is a critical technique for moving data between compute and storage subsystems of supercomputers. With massive amounts of data produced or consumed by compute nodes, high‐performant parallel I/O is essential. I/O benchmarks play an important role in this process; however, there is a scarcity of I/O benchmarks representative of current workloads on HPC systems. Toward creating representative I/O kernels from real‐world applications, we have created h5bench , a set of I/O kernels that exercise hierarchical data format version 5 (HDF5) I/O on parallel file systems in numerous dimensions. Our focus on HDF5 is due to the parallel I/O library's heavy usage in various scientific applications running on supercomputing systems. The various tests benchmarked in the h5bench suite include I/O operations (read and write), data locality (arrays of basic data types and arrays of structures), array dimensionality (one‐dimensional arrays, two‐dimensional meshes, three‐dimensional cubes), I/O modes (synchronous and asynchronous). In this paper, we present the observed performance of h5bench executed along several of these dimensions on existing supercomputers (Cori and Summit) and pre‐exascale platforms (Perlmutter, Theta, and Polaris). h5bench measurements can be used to identify performance bottlenecks and their root causes and evaluate I/O optimizations. As the I/O patterns of h5bench are diverse and capture the I/O behaviors of various HPC applications, this study will be helpful to the broader supercomputing and I/O community. [ABSTRACT FROM AUTHOR]

Quantum computing as emerging technology raises high expectations to accelerate complicated calculations, superseding capabilities of presently available supercomputers and to provide a solution to the ever-increasing challenge of processing big amount of data. Besides theoretical promises, quantum computers have successfully been realized in experimental settings during the past few years, and can now also be accessed commercially. Given this availability of quantum computers, their application potential can be scrutinized. Promised to lead to disruptive changes in addressing and solving problems from the domains of simulation, optimization and machine learning, quantum computers have been researched in context of many application scenarios in industrial and academic setting in the recent years. However, it turns out that currently available quantum computers are still to small in size and limited in quality to fulfill the high hopes set in them at the moment. This review reflects on the current state of the technology coming from the perspective of the application, and indicates the developments required on hardware, software and algorithmic side to eventually make quantum computing beneficial for applications. [ABSTRACT FROM AUTHOR]

Ultra-sensitive sensors based on the resonant properties of whispering gallery modes (WGMs) can detect fractional changes in nanoscale environments down to the length and time scales of single molecules. However, it is challenging to isolate single-molecule signals from competing noise sources in experiments, such as thermal and mechanical sources of noise, and—at the most fundamental level—the shot noise limit of classical light. Additionally, in contrast to traditional bulk refractive index measurements, analyzing single-molecule signals is complicated by the localized nature of their interactions with nanoscale field gradients. This perspective discusses multifaceted solutions to these challenges, including the use of quantum light sources to boost the signal-to-noise ratio in experiments and leveraging the power of supercomputers to predict the electronic response of molecules to WGM optoplasmonic fields. We further discuss the role of machine learning in WGM sensing, including several advanced models that can predict molecular polarizability and solvent effects. These advancements in WGM spectroscopy and computational modeling can help to decipher the molecular mechanics of enzymes, enable studies of catalysis on the nanoscale, and probe the quantum nature of molecules. [ABSTRACT FROM AUTHOR]

Currently, research on the lattice Boltzmann method mainly focuses on its numerical simulation and applications, and there is an increasing demand for large-scale simulations in practical scenarios. In response to this situation, this study successfully implemented a large-scale heterogeneous parallel algorithm for the lattice Boltzmann method using OpenMP, MPI, Pthread, and OpenCL parallel technologies on the "Dongfang" supercomputer system. The accuracy and effectiveness of this algorithm were verified through the lid-driven cavity flow simulation. The paper focused on optimizing the algorithm in four aspects: Firstly, non-blocking communication was employed to overlap communication and computation, thereby improving parallel efficiency. Secondly, high-speed shared memory was utilized to enhance memory access performance and reduce latency. Thirdly, a balanced computation between the central processing unit and the accelerator was achieved through proper task partitioning and load-balancing strategies. Lastly, memory access efficiency was improved by adjusting the memory layout. Performance testing demonstrated that the optimized algorithm exhibited improved parallel efficiency and scalability, with computational performance that is 4 times greater than before optimization and 20 times that of a 32-core CPU. [ABSTRACT FROM AUTHOR]

Summary: The Oak Ridge Leadership Computing Facility (OLCF) has a long history of supporting and promoting GPU‐accelerated computing starting with the deployment of the Titan supercomputer in 2021 and continuing with the Summit supercomputer which has a theoretical peak performance of approximately 200 petaflops. Because the majority of Summit's computational power comes from its 27,972 GPUs, users must port their applications to one of the supported programming models in order to make efficient use of the system. To prepare the transition to Frontier, the OLCF's exascale supercomputer, users will need to adapt to an entirely new ecosystem which will include new hardware and software technologies. First, users will need to familiarize themselves with the AMD Radeon GPU architecture. Furthermore, users who have been previously relying on CUDA will need to transition to the Heterogeneous‐Computing Interface for Portability (HIP) or one of the other supported programming models (e.g., OpenMP, OpenACC). In this work, we describe our initial experiences and lessons learned in porting three applications or proxy apps currently running on Summit to the HPE/Cray ecosystem to leverage the compute power from AMD GPUs: minisweep, GenASiS, and Sparkler. Each one is representative of current production workloads utilized at the OLCF, different programming languages, and different programming models. [ABSTRACT FROM AUTHOR]

The ability to accurately predict when a job on a supercomputer will leave the queue and start to run is not only beneficial for providing insights to users, but can also help enable non‐traditional HPC workloads that are not necessarily suited to the batch queue style‐approach that is ubiquitous on production HPC machines. However there are numerous challenges in achieving such a prediction with high accuracy, not least because the queue's state can change rapidly and depend upon many factors. In this work, we explore a novel machine learning approach for predicting queue wait times, hypothesising that such a model can capture the complex behavior resulting from the queue policy and other interactions to generate accurate job start times. For ARCHER2 (HPE Cray EX), Cirrus (HPE 8600), and 4‐cabinet (HPE Cray EX) we explore how different machine learning approaches and techniques improve the accuracy of our predictions, comparing against the estimation generated by Slurm. By combining categorization and regression models, we demonstrate that our approach delivers the most accurate predictions across our machines of interest, with the result of this work being the ability to predict job start times within 1 min of the actual start time for around 65% of jobs on ARCHER2 and 4‐cabinet, and 76% of jobs on Cirrus. When compared against what Slurm can deliver, via the backfill plugin, this represents around 3.8 times better accuracy on ARCHER2 and 18 times better for Cirrus. Furthermore our approach can accurately predicting the start time for three quarters of all job within 10 min of the actual start time on ARCHER2 and 4‐cabinet, and for 90% of jobs on Cirrus. Whilst the initial driver of this work was to better facilitate non‐traditional, interactive and urgent, workloads on HPC machines, the insights gained can also be used to provide wider benefits to users, enrich existing batch queue systems, and inform supercomputing center policy also. [ABSTRACT FROM AUTHOR]

The development of new exascale supercomputers has dramatically increased the need for fast, high-performance networking technology. Efficient network topologies, such as Dragonfly+, have been introduced to meet the demands of data-intensive applications and to match the massive computing power of GPUs and accelerators. However, these supercomputers still face performance variability mainly caused by the network that affects system and application performance. This study comprehensively analyzes performance variability on a large-scale HPC system with Dragonfly+ network topology, focusing on factors such as communication patterns, message size, job placement locality, MPI collective algorithms, and overall system workload. The study also proposes an easy-to-measure metric for estimating network background traffic generated by other users, which can be used to estimate the performance of our job accurately. The insights gained from this study contribute to improving performance predictability, enhancing job placement policies and MPI algorithm selection, and optimizing resource management strategies in supercomputers. [ABSTRACT FROM AUTHOR]