arXiv subjects

David Schultz

Publications and source records attributed to David Schultz.

17 recordsLinked to original sources

arXiv2025

SuperSONIC: Cloud-Native Infrastructure for ML Inferencing

The increasing computational demand from growing data rates and complex machine learning (ML) algorithms in large-scale scientific experiments has driven the adoption of the Services for Optimized Network Inference on Coprocessors (SONIC) approach. SONIC accelerates ML inference by offloading it to local or remote coprocessors to optimize resource utilization. Leveraging its portability to different types of coprocessors, SONIC enhances data processing and model deployment efficiency for cutting-edge research in high energy physics (HEP) and multi-messenger astrophysics (MMA). We developed the SuperSONIC project, a scalable server infrastructure for SONIC, enabling the deployment of computationally intensive tasks to Kubernetes clusters equipped with graphics processing units (GPUs). Using NVIDIA Triton Inference Server, SuperSONIC decouples client workflows from server infrastructure, standardizing communication, optimizing throughput, load balancing, and monitoring. SuperSONIC has been successfully deployed for the CMS and ATLAS experiments at the CERN Large Hadron Collider (LHC), the IceCube Neutrino Observatory (IceCube), and the Laser Interferometer Gravitational-Wave Observatory (LIGO) and tested on Kubernetes clusters at Purdue University, the National Research Platform (NRP), and the University of Chicago. SuperSONIC addresses the challenges of the Cloud-native era by providing a reusable, configurable framework that enhances the efficiency of accelerator-based inference deployment across diverse scientific domains and industries.

cs.DC↗

arXiv2023

Demand-driven provisioning of Kubernetes-like resources in OSG

The OSG-operated Open Science Pool is an HTCondor-based virtual cluster that aggregates resources from compute clusters provided by several organizations. Most of the resources are not owned by OSG, so demand-based dynamic provisioning is important for maximizing usage without incurring excessive waste. OSG has long relied on GlideinWMS for most of its resource provisioning needs but is limited to resources that provide a Grid-compliant Compute Entrypoint. To work around this limitation, the OSG Software Team has developed a glidein container that resource providers could use to directly contribute to the OSPool. The problem of that approach is that it is not demand-driven, relegating it to backfill scenarios only. To address this limitation, a demand-driven direct provisioner of Kubernetes resources has been developed and successfully used on the NRP. The setup still relies on the OSG-maintained backfill container image but automates the provisioning matchmaking and successive requests. That provisioner has also been extended to support Lancium, a green computing cloud provider with a Kubernetes-like proprietary interface. The provisioner logic has been intentionally kept very simple, making this extension a low-cost project. Both NRP and Lancium resources have been provisioned exclusively using this mechanism for many months.

cs.DC↗

arXiv2023

IceCube experience using XRootD-based Origins with GPU workflows in PNRP

The IceCube Neutrino Observatory is a cubic kilometer neutrino telescope located at the geographic South Pole. Understanding detector systematic effects is a continuous process. This requires the Monte Carlo simulation to be updated periodically to quantify potential changes and improvements in science results with more detailed modeling of the systematic effects. IceCube's largest systematic effect comes from the optical properties of the ice the detector is embedded in. Over the last few years there have been considerable improvements in the understanding of the ice, which require a significant processing campaign to update the simulation. IceCube normally stores the results in a central storage system at the University of Wisconsin-Madison, but it ran out of disk space in 2022. The Prototype National Research Platform (PNRP) project thus offered to provide both GPU compute and storage capacity to IceCube in support of this activity. The storage access was provided via XRootD-based OSDF Origins, a first for IceCube computing. We report on the overall experience using PNRP resources, with both successes and pain points.

physics.comp-ph↗

arXiv2023

Evaluation of ARM CPUs for IceCube available through Google Kubernetes Engine

The IceCube experiment has substantial simulation needs and is in continuous search for the most cost-effective ways to satisfy them. The most CPU-intensive part relies on CORSIKA, a cosmic ray air shower simulation. Historically, IceCube relied exclusively on x86-based CPUs, like Intel Xeon and AMD EPYC, but recently server-class ARM-based CPUs are also becoming available, both on-prem and in the cloud. In this paper we present our experience in running a sample CORSIKA simulation on both ARM and x86 CPUs available through Google Kubernetes Engine (GKE). We used the production binaries for the x86 instances, but had to build the binaries for ARM instances from source code, which turned out to be mostly painless. Our benchmarks show that ARM-based CPUs in GKE were not only the most cost-effective but were also the fastest in absolute terms in all the tested configurations. While the advantage is not drastic, about 20% in cost-effectiveness and less than 10% in absolute terms, it is still large enough to warrant an investment in ARM support for IceCube.

cs.PF↗

arXiv2023

An improved infrastructure for the IceCube realtime system

The IceCube realtime alert system has been operating since 2016. It provides prompt alerts on high-energy neutrino events to the astroparticle physics community. The localization regions for the incoming direction of neutrinos are published through NASA's Gamma-ray Coordinate Network (GCN). The IceCube realtime system consists of infrastructure dedicated to the selection of alert events, the reconstruction of their topology and arrival direction, the calculation of directional uncertainty contours and the distribution of the event information through public alert networks. Using a message-based workflow management system, a dedicated software (SkyDriver) provides a representational state transfer (REST) interface to parallelized reconstruction algorithms. In this contribution, we outline the improvements of the internal infrastructure of the IceCube realtime system that aims to streamline the internal handling of neutrino events, their distribution to the SkyDriver interface, the collection of the reconstruction results as well as their conversion into human- and machine-readable alerts to be publicly distributed through different alert networks. An approach for the long-term storage and cataloging of alert events according to findability, accessibility, interoperability and reusability (FAIR) principles is outlined.

astro-ph.HE↗

arXiv2022

The anachronism of whole-GPU accounting

NVIDIA has been making steady progress in increasing the compute performance of its GPUs, resulting in order of magnitude compute throughput improvements over the years. With several models of GPUs coexisting in many deployments, the traditional accounting method of treating all GPUs as being equal is not reflecting compute output anymore. Moreover, for applications that require significant CPU-based compute to complement the GPU-based compute, it is becoming harder and harder to make full use of the newer GPUs, requiring sharing of those GPUs between multiple applications in order to maximize the achievable science output. This further reduces the value of whole-GPU accounting, especially when the sharing is done at the infrastructure level. We thus argue that GPU accounting for throughput-oriented infrastructures should be expressed in GPU core hours, much like it is normally done for the CPUs. While GPU core compute throughput does change between GPU generations, the variability is similar to what we expect to see among CPU cores. To validate our position, we present an extensive set of run time measurements of two IceCube photon propagation workflows on 14 GPU models, using both on-prem and Cloud resources. The measurements also outline the influence of GPU sharing at both HTCondor and Kubernetes infrastructure level.

cs.DC↗

arXiv2021

Expanding IceCube GPU computing into the Clouds

The IceCube collaboration relies on GPU compute for many of its needs, including ray tracing simulation and machine learning activities. GPUs are however still a relatively scarce commodity in the scientific resource provider community, so we expanded the available resource pool with GPUs provisioned from the commercial Cloud providers. The provisioned resources were fully integrated into the normal IceCube workload management system through the Open Science Grid (OSG) infrastructure and used CloudBank for budget management. The result was an approximate doubling of GPU wall hours used by IceCube over a period of 2 weeks, adding over 3.1 fp32 EFLOP hours for a price tag of about $58k. This paper describes the setup used and the operational experience.

cs.DC↗

arXiv2021

Managing Cloud networking costs for data-intensive applications by provisioning dedicated network links

Many scientific high-throughput applications can benefit from the elastic nature of Cloud resources, especially when there is a need to reduce time to completion. Cost considerations are usually a major issue in such endeavors, with networking often a major component; for data-intensive applications, egress networking costs can exceed the compute costs. Dedicated network links provide a way to lower the networking costs, but they do add complexity. In this paper we provide a description of a 100 fp32 PFLOPS Cloud burst in support of IceCube production compute, that used Internet2 Cloud Connect service to provision several logically-dedicated network links from the three major Cloud providers, namely Amazon Web Services, Microsoft Azure and Google Cloud Platform, that in aggregate enabled approximately 100 Gbps egress capability to on-prem storage. It provides technical details about the provisioning process, the benefits and limitations of such a setup and an analysis of the costs incurred.

cs.NI↗

arXiv2020

Running a Pre-Exascale, Geographically Distributed, Multi-Cloud Scientific Simulation

As we approach the Exascale era, it is important to verify that the existing frameworks and tools will still work at that scale. Moreover, public Cloud computing has been emerging as a viable solution for both prototyping and urgent computing. Using the elasticity of the Cloud, we have thus put in place a pre-exascale HTCondor setup for running a scientific simulation in the Cloud, with the chosen application being IceCube's photon propagation simulation. I.e. this was not a purely demonstration run, but it was also used to produce valuable and much needed scientific results for the IceCube collaboration. In order to reach the desired scale, we aggregated GPU resources across 8 GPU models from many geographic regions across Amazon Web Services, Microsoft Azure, and the Google Cloud Platform. Using this setup, we reached a peak of over 51k GPUs corresponding to almost 380 PFLOP32s, for a total integrated compute of about 100k GPU hours. In this paper we provide the description of the setup, the problems that were discovered and overcome, as well as a short description of the actual science output of the exercise.

cs.DC↗

arXiv2019

An Average-Compress Algorithm for the Sample Mean Problem under Dynamic Time Warping

Computing a sample mean of time series under dynamic time warping (DTW) is NP-hard. Consequently, there is an ongoing research effort to devise efficient heuristics. The majority of heuristics have been developed for the constrained sample mean problem that assumes a solution of predefined length. In contrast, research on the unconstrained sample mean problem is underdeveloped. In this article, we propose a generic average-compress (AC) algorithm for solving the unconstrained problem. The algorithm alternates between averaging (A-step) and compression (C-step). The A-step takes an initial guess as input and returns an approximation of a sample mean. Then the C-step reduces the length of the approximate solution. The compressed approximation serves as initial guess of the A-step in the next iteration. The purpose of the C-step is to direct the algorithm to more promising solutions of shorter length. The proposed algorithm is generic in the sense that any averaging and any compression method can be used. Experimental results show that the AC algorithm substantially outperforms current state-of-the-art algorithms for time series averaging.

cs.DS↗

arXiv2017

Exact Mean Computation in Dynamic Time Warping Spaces

Dynamic time warping constitutes a major tool for analyzing time series. In particular, computing a mean series of a given sample of series in dynamic time warping spaces (by minimizing the Fr\'echet function) is a challenging computational problem, so far solved by several heuristic and inexact strategies. We spot some inaccuracies in the literature on exact mean computation in dynamic time warping spaces. Our contributions comprise an exact dynamic program computing a mean (useful for benchmarking and evaluating known heuristics). Based on this dynamic program, we empirically study properties like uniqueness and length of a mean. Moreover, experimental evaluations reveal substantial deficits of state-of-the-art heuristics in terms of their output quality. We also give an exact polynomial-time algorithm for the special case of binary time series.

cs.DS↗

arXiv2017

Optimal Warping Paths are unique for almost every Pair of Time Series

Update rules for learning in dynamic time warping spaces are based on optimal warping paths between parameter and input time series. In general, optimal warping paths are not unique resulting in adverse effects in theory and practice. Under the assumption of squared error local costs, we show that no two warping paths have identical costs almost everywhere in a measure-theoretic sense. Two direct consequences of this result are: (i) optimal warping paths are unique almost everywhere, and (ii) the set of all pairs of time series with multiple equal-cost warping paths coincides with the union of exponentially many zero sets of quadratic forms. One implication of the proposed results is that typical distance-based cost functions such as the k-means objective are differentiable almost everywhere and can be minimized by subgradient methods.

cs.LG↗

arXiv2017

Asymmetric Learning Vector Quantization for Efficient Nearest Neighbor Classification in Dynamic Time Warping Spaces

The nearest neighbor method together with the dynamic time warping (DTW) distance is one of the most popular approaches in time series classification. This method suffers from high storage and computation requirements for large training sets. As a solution to both drawbacks, this article extends learning vector quantization (LVQ) from Euclidean spaces to DTW spaces. The proposed LVQ scheme uses asymmetric weighted averaging as update rule. Empirical results exhibited superior performance of asymmetric generalized LVQ (GLVQ) over other state-of-the-art prototype generation methods for nearest neighbor classification.

cs.LG↗

arXiv2017

Nonsmooth Analysis and Subgradient Methods for Averaging in Dynamic Time Warping Spaces

Time series averaging in dynamic time warping (DTW) spaces has been successfully applied to improve pattern recognition systems. This article proposes and analyzes subgradient methods for the problem of finding a sample mean in DTW spaces. The class of subgradient methods generalizes existing sample mean algorithms such as DTW Barycenter Averaging (DBA). We show that DBA is a majorize-minimize algorithm that converges to necessary conditions of optimality after finitely many iterations. Empirical results show that for increasing sample sizes the proposed stochastic subgradient (SSG) algorithm is more stable and finds better solutions in shorter time than the DBA algorithm on average. Therefore, SSG is useful in online settings and for non-small sample sizes. The theoretical and empirical results open new paths for devising sample mean algorithms: nonsmooth optimization methods and modified variants of pairwise averaging methods.

cs.CV↗

arXiv2016

On the Existence of a Sample Mean in Dynamic Time Warping Spaces

The concept of sample mean in dynamic time warping (DTW) spaces has been successfully applied to improve pattern recognition systems and generalize centroid-based clustering algorithms. Its existence has neither been proved nor challenged. This article presents sufficient conditions for existence of a sample mean in DTW spaces. The proposed result justifies prior work on approximate mean algorithms, sets the stage for constructing exact mean algorithms, and is a first step towards a statistical theory of DTW spaces.

cs.CV↗

arXiv2015

Detecting particles with cell phones: the Distributed Electronic Cosmic-ray Observatory

In 2014 the number of active cell phones worldwide for the first time surpassed the number of humans. Cell phone camera quality and onboard processing power (both CPU and GPU) continue to improve rapidly. In addition to their primary purpose of detecting photons, camera image sensors on cell phones and other ubiquitous devices such as tablets, laptops and digital cameras can detect ionizing radiation produced by cosmic rays and radioactive decays. While cosmic rays have long been understood and characterized as a nuisance in astronomical cameras, they can also be identified as a signal in idle camera image sensors. We present the Distributed Electronic Cosmic-ray Observatory (DECO), a platform for outreach and education as well as for citizen science. Consisting of an app and associated database and web site, DECO harnesses the power of distributed camera image sensors for cosmic-ray detection.

astro-ph.IM↗

arXiv2011

Laboratory Astrophysics White Paper (based on the 2010 NASA Laboratory Astrophysics Workshop in Gatlinberg, Tennessee, 25-28 October 2010)

The purpose of the 2010 NASA Laboratory Astrophysics Workshop (LAW) was, as given in the Charter from NASA, "to provide a forum within which the scientific community can review the current state of knowledge in the field of Laboratory Astrophysics, assess the critical data needs of NASA's current and future Space Astrophysics missions, and identify the challenges and opportunities facing the field as we begin a new decade". LAW 2010 was the fourth in a roughly quadrennial series of such workshops sponsored by the Astrophysics Division of the NASA Science Mission Directorate. In this White Paper, we report the findings of the workshop.

astro-ph.IM↗