Search

Due to the increased demand of network traffic expected during the HL-LHC era, the T2 sites in the USA will be required to have 400Gbps of available bandwidth to their storage solution. With the above in mind we are pursuing a scale test of XRootD software when used to perform Third Party Copy transfers using the HTTP protocol. Our main objective is to understand the possible limitations in the software stack to achieve the target transfer rate; to that end we have set up a testbed of multiple XRootD servers in both UCSD and Caltech which are connected through a dedicated link capable of 400 Gbps end-to-end. Building upon our experience deploying containerized XRootD servers, we use Kubernetes to easily deploy and test different configurations of our testbed. In this work, we will present our experience doing these tests and the lessons learned., Comment: 8 pages, 4 figures, submitted to CHEP'23

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., Comment: 6 pages, 3 figures, Submitted to Proceedings of CHEP23

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., Comment: 7 pages, 3 figures, 1 table, To be published in Proceedings of CHEP23

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., Comment: 5 pages,3 tables, Submitted to proceedings of CHEP23

Modern large-scale physics experiments create datasets with sizes and streaming rates that can exceed those from industry leaders such as Google Cloud and Netflix. Fully processing these datasets requires both sufficient compute power and efficient workflows. Recent advances in Machine Learning (ML) and Artificial Intelligence (AI) can either improve or replace existing domain-specific algorithms to increase workflow efficiency. Not only can these algorithms improve the physics performance of current algorithms, but they can often be executed more quickly, especially when run on coprocessors such as GPUs or FPGAs. In the winter of 2023, MIT hosted the Accelerating Physics with ML at MIT workshop, which brought together researchers from gravitational-wave physics, multi-messenger astrophysics, and particle physics to discuss and share current efforts to integrate ML tools into their workflows. The following white paper highlights examples of algorithms and computing frameworks discussed during this workshop and summarizes the expected computing needs for the immediate future of the involved fields., Comment: Whitepaper resulting from Accelerating Physics with ML@MIT workshop in Jan/Feb 2023

Compute resource providers often put in place batch compute systems to maximize the utilization of such resources. However, compute nodes in such clusters, both physical and logical, contain several complementary resources, with notable examples being CPUs, GPUs, memory and ephemeral storage. User jobs will typically require more than one such resource, resulting in co-scheduling trade-offs of partial nodes, especially in multi-user environments. When accounting for either user billing or scheduling overhead, it is thus important to consider all such resources together. We thus define the concept of a threshold-based "canonical unit" that combines several resource types into a single discrete unit and use it to characterize scheduling overhead and make resource billing more fair for both resource providers and users. Note that the exact definition of a canonical unit is not prescribed and may change between resource providers. Nevertheless, we provide a template and two example definitions that we consider appropriate in the context of the Open Science Grid., Comment: 6 pages, 2 figures, To be published in proceedings of PEARC23

GitHub is a popular repository for hosting software projects, both due to ease of use and the seamless integration with its testing environment. Native GitHub Actions make it easy for software developers to validate new commits and have confidence that new code does not introduce major bugs. The freely available test environments are limited to only a few popular setups but can be extended with custom Action Runners. Our team had access to a Kubernetes cluster with GPU accelerators, so we explored the feasibility of automatically deploying GPU-providing runners there. All available Kubernetes-based setups, however, require cluster-admin level privileges. To address this problem, we developed a simple custom setup that operates in a completely unprivileged manner. In this paper we provide a summary description of the setup and our experience using it in the context of two Knight lab projects on the Prototype National Research Platform system., Comment: 5 pages, 1 figure, To be published in proceedings of PEARC23

Unique scientific instruments designed and operated by large global collaborations are expected to produce Exabyte-scale data volumes per year by 2030. These collaborations depend on globally distributed storage and compute to turn raw data into science. While all of these infrastructures have batch scheduling capabilities to share compute, Research and Education networks lack those capabilities. There is thus uncontrolled competition for bandwidth between and within collaborations. As a result, data "hogs" disk space at processing facilities for much longer than it takes to process, leading to vastly over-provisioned storage infrastructures. Integrated co-scheduling of networks as part of high-level managed workflows might reduce these storage needs by more than an order of magnitude. This paper describes such a solution, demonstrates its functionality in the context of the Large Hadron Collider (LHC) at CERN, and presents the next-steps towards its use in production., Comment: Submitted to the proceedings of the XLOOP workshop held in conjunction with Supercomputing 22

Fusion simulations have traditionally required the use of leadership scale High Performance Computing (HPC) resources in order to produce advances in physics. The impressive improvements in compute and memory capacity of many-GPU compute nodes are now allowing for some problems that once required a multi-node setup to be also solvable on a single node. When possible, the increased interconnect bandwidth can result in order of magnitude higher science throughput, especially for communication-heavy applications. In this paper we analyze the performance of the fusion simulation tool CGYRO, an Eulerian gyrokinetic turbulence solver designed and optimized for collisional, electromagnetic, multiscale simulation, which is widely used in the fusion research community. Due to the nature of the problem, the application has to work on a large multi-dimensional computational mesh as a whole, requiring frequent exchange of large amounts of data between the compute processes. In particular, we show that the average-scale nl03 benchmark CGYRO simulation can be run at an acceptable speed on a single Google Cloud instance with 16 A100 GPUs, outperforming 8 NERSC Perlmutter Phase1 nodes, 16 ORNL Summit nodes and 256 NERSC Cori nodes. Moving from a multi-node to a single-node GPU setup we get comparable simulation times using less than half the number of GPUs. Larger benchmark problems, however, still require a multi-node HPC setup due to GPU memory capacity needs, since at the time of writing no vendor offers nodes with a sufficient GPU memory setup. The upcoming external NVSWITCH does however promise to deliver an almost equivalent solution for up to 256 NVIDIA GPUs., Comment: 6 pages, 1 table, 1 figure, to be published in proceedings of PEARC22

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., Comment: 6 pages, 2 tables, 1 figure, to be published in proceedings of PEARC22

The XRootD system is used to transfer, store, and cache large datasets from high-energy physics (HEP). In this study we focus on its capability as distributed on-demand storage cache. Through exploring a large set of daily log files between 2020 and 2021, we seek to understand the data access patterns that might inform future cache design. Our study begins with a set of summary statistics regarding file read operations, file lifetimes, and file transfers. We observe that the number of read operations on each file remains nearly constant, while the average size of a read operation grows over time. Furthermore, files tend to have a consistent length of time during which they remain open and are in use. Based on this comprehensive study of the cache access statistics, we developed a cache simulator to explore the behavior of caches of different sizes. Within a certain size range, we find that increasing the XRootD cache size improves the cache hit rate, yielding faster overall file access. In particular, we find that increase the cache size from 40TB to 56TB could increase the hit rate from 0.62 to 0.89, which is a significant increase in cache effectiveness for modest cost.

Scientific collaborations are increasingly relying on large volumes of data for their work and many of them employ tiered systems to replicate the data to their worldwide user communities. Each user in the community often selects a different subset of data for their analysis tasks; however, members of a research group often are working on related research topics that require similar data objects. Thus, there is a significant amount of data sharing possible. In this work, we study the access traces of a federated storage cache known as the Southern California Petabyte Scale Cache. By studying the access patterns and potential for network traffic reduction by this caching system, we aim to explore the predictability of the cache uses and the potential for a more general in-network data caching. Our study shows that this distributed storage cache is able to reduce the network traffic volume by a factor of 2.35 during a part of the study period. We further show that machine learning models could predict cache utilization with an accuracy of 0.88. This demonstrates that such cache usage is predictable, which could be useful for managing complex networking resources such as in-network caching.

HTCondor has been very successful in managing globally distributed, pleasantly parallel scientific workloads, especially as part of the Open Science Grid. HTCondor system design makes it ideal for integrating compute resources provisioned from anywhere, but it has very limited native support for autonomously provisioning resources managed by other solutions. This work presents a solution that allows for autonomous, demand-driven provisioning of Kubernetes-managed resources. A high-level overview of the employed architectures is presented, paired with the description of the setups used in both on-prem and Cloud deployments in support of several Open Science Grid communities. The experience suggests that the described solution should be generally suitable for contributing Kubernetes-based resources to existing HTCondor pools., Comment: 6 pages, 3 figures, to be published in proceedings of PEARC22

This paper describes a vision and work in progress to elevate network resources and data transfer management to the same level as compute and storage in the context of services access, scheduling, life cycle management, and orchestration. While domain science workflows often include active compute resource allocation and management, the data transfers and associated network resource coordination is not handled in a similar manner. As a result data transfers can introduce a degree of uncertainty in workflow operations, and the associated lack of network information does not allow for either the workflow operations or the network use to be optimized. The net result is that domain science workflow processes are forced to view the network as an opaque infrastructure into which they inject data and hope that it emerges at the destination with an acceptable Quality of Experience. There is little ability for applications to interact with the network to exchange information, negotiate performance parameters, discover expected performance metrics, or receive status/troubleshooting information in real time. Developing mechanisms to allow an application workflow to obtain information regarding the network services, capabilities, and options, to a degree similar to what is possible for compute resources is the primary motivation for this work. The initial focus is on the Open Science Grid (OSG)/Compact Muon Solenoid (CMS) Large Hadron Collider (LHC) workflows with Rucio/FTS/XRootD based data transfers and the interoperation with the ESnet SENSE (Software-Defined Network for End-to-end Networked Science at the Exascale) system., Comment: contribution to Snowmass 2022

The Compact Muon Solenoid (CMS) experiment at the Large Hadron Collider (LHC) is one of the largest data producers in the scientific world, with standard data products centrally produced, and then used by often competing teams within the collaboration. This work is focused on how a local institution, University of California San Diego (UCSD), partnered with the Open Science Grid (OSG) to use Azure cloud resources to augment its available computing to accelerate time to results for multiple analyses pursued by a small group of collaborators. The OSG is a federated infrastructure allowing many independent resource providers to serve many independent user communities in a transparent manner. Historically the resources would come from various research institutions, spanning small universities to large HPC centers, based on either community needs or grant allocations, so adding commercial clouds as resource providers is a natural evolution. The OSG technology allows for easy integration of cloud resources, but the data-intensive nature of CMS compute jobs required the deployment of additional data caching infrastructure to ensure high efficiency., Comment: 11 pages, 5 figures, to be published in proceedings of ICOCBI 2021

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., Comment: 2 pages, 2 figures, to be published in proceedings of eScience 2021

HTCondor is a major workload management system used in distributed high throughput computing (dHTC) environments, e.g., the Open Science Grid. One of the distinguishing features of HTCondor is the native support for data movement, allowing it to operate without a shared filesystem. Coupling data handling and compute scheduling is both convenient for users and allows for significant infrastructure flexibility but does introduce some limitations. The default HTCondor data transfer mechanism routes both the input and output data through the submission node, making it a potential bottleneck. In this document we show that by using a node equipped with a 100 Gbps network interface (NIC) HTCondor can serve data at up to 90 Gbps, which is sufficient for most current use cases, as it would saturate the border network links of most research universities at the time of writing., Comment: 2 pages, 2 figures, to be published in proceedings of eScience 2021

The volume of data moving through a network increases with new scientific experiments and simulations. Network bandwidth requirements also increase proportionally to deliver data within a certain time frame. We observe that a significant portion of the popular dataset is transferred multiple times to different users as well as to the same user for various reasons. In-network data caching for the shared data has shown to reduce the redundant data transfers and consequently save network traffic volume. In addition, overall application performance is expected to improve with in-network caching because access to the locally cached data results in lower latency. This paper shows how much data was shared over the study period, how much network traffic volume was consequently saved, and how much the temporary in-network caching increased the scientific application performance. It also analyzes data access patterns in applications and the impacts of caching nodes on the regional data repository. From the results, we observed that the network bandwidth demand was reduced by nearly a factor of 3 over the study period.

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., Comment: 8 pages, 7 figures, 4 tables, to be published in proceedings of PEARC21

The High Luminosity Large Hadron Collider provides a data challenge. The amount of data recorded from the experiments and transported to hundreds of sites will see a thirty fold increase in annual data volume. A systematic approach to contrast the performance of different Third Party Copy(TPC) transfer protocols arises. Two contenders, XRootD-HTTPS and the GridFTP are evaluated in their performance for transferring files from one server to an-other over 100Gbps interfaces. The benchmarking is done by scheduling pods on the Pacific Research Platform Kubernetes cluster to ensure reproducible and repeatable results. This opens a future pathway for network testing of any TPC transfer protocol., Comment: 7 pages, 8 figures

We present results from parallelizing the unpacking and clustering steps of the raw data from the silicon strip modules for reconstruction of charged particle tracks. Throughput is further improved by concurrently processing multiple events using nested OpenMP parallelism on CPU or CUDA streams on GPU. The new implementation along with earlier work in developing a parallelized and vectorized implementation of the combinatoric Kalman filter algorithm has enabled efficient global reconstruction of the entire event on modern computer architectures. We demonstrate the performance of the new implementation on Intel Xeon and NVIDIA GPU architectures.

During the first observation run the LIGO collaboration needed to offload some of its most, intense CPU workflows from its dedicated computing sites to opportunistic resources. Open Science Grid enabled LIGO to run PyCbC, RIFT and Bayeswave workflows to seamlessly run in a combination of owned and opportunistic resources. One of the challenges is enabling the workflows to use several heterogeneous resources in a coordinated and effective way.

A general problem faced by computing on the grid for opportunistic users is that delivering cycles is simpler than delivering data to those cycles. In this project we show how we integrated XRootD caches placed on the internet backbone to implement a content delivery network for general science workflows. We will show that for some workflows on different science domains like high energy physics, gravitational waves, and others the combination of data reuse from the workflows together with the use of caches increases CPU efficiency while decreasing network bandwidth use.

One of the most computationally challenging problems expected for the High-Luminosity Large Hadron Collider (HL-LHC) is determining the trajectory of charged particles during event reconstruction. Algorithms used at the LHC today rely on Kalman filtering, which builds physical trajectories incrementally while incorporating material effects and error estimation. Recognizing the need for faster computational throughput, we have adapted Kalman-filter-based methods for highly parallel, many-core SIMD architectures that are now prevalent in high-performance hardware. In this paper, we discuss the design and performance of the improved tracking algorithm, referred to as mkFit. A key piece of the algorithm is the Matriplex library, containing dedicated code to optimally vectorize operations on small matrices. The physics performance of the mkFit algorithm is comparable to the nominal CMS tracking algorithm when reconstructing tracks from simulated proton-proton collisions within the CMS detector. We study the scaling of the algorithm as a function of the parallel resources utilized and find large speedups both from vectorization and multi-threading. mkFit achieves a speedup of a factor of 6 compared to the nominal algorithm when run in a single-threaded application within the CMS software framework.

The Scalable Systems Laboratory (SSL), part of the IRIS-HEP Software Institute, provides Institute participants and HEP software developers generally with a means to transition their R&D from conceptual toys to testbeds to production-scale prototypes. The SSL enables tooling, infrastructure, and services supporting the innovation of novel analysis and data architectures, development of software elements and tool-chains, reproducible functional and scalability testing of service components, and foundational systems R&D for accelerated services developed by the Institute. The SSL is constructed with a core team having expertise in scale testing and deployment of services across a wide range of cyberinfrastructure. The core team embeds and partners with other areas in the Institute, and with LHC and other HEP development and operations teams as appropriate, to define investigations and required service deployment patterns. We describe the approach and experiences with early application deployments, including analysis platforms and intelligent data delivery systems.

The University of California system has excellent networking between all of its campuses as well as a number of other Universities in CA, including Caltech, most of them being connected at 100 Gbps. UCSD and Caltech have thus joined their disk systems into a single logical xcache system, with worker nodes from both sites accessing data from disks at either site. This setup has been in place for a couple years now and has shown to work very well. Coherently managing nodes at multiple physical locations has however not been trivial, and we have been looking for ways to improve operations. With the Pacific Research Platform (PRP) now providing a Kubernetes resource pool spanning resources in the science DMZs of all the UC campuses, we have recently migrated the xcache services from being hosted bare-metal into containers. This paper presents our experience in both migrating to and operating in the new environment.

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., Comment: 18 pages, 5 figures, 4 tables, to be published in Proceedings of ISC High Performance 2020

One of the most computationally challenging problems expected for the High-Luminosity Large Hadron Collider (HL-LHC) is finding and fitting particle tracks during event reconstruction. Algorithms used at the LHC today rely on Kalman filtering, which builds physical trajectories incrementally while incorporating material effects and error estimation. Recognizing the need for faster computational throughput, we have adapted Kalman-filter-based methods for highly parallel, many-core SIMD and SIMT architectures that are now prevalent in high-performance hardware. Previously we observed significant parallel speedups, with physics performance comparable to CMS standard tracking, on Intel Xeon, Intel Xeon Phi, and (to a limited extent) NVIDIA GPUs. While early tests were based on artificial events occurring inside an idealized barrel detector, we showed subsequently that our mkFit software builds tracks successfully from complex simulated events (including detector pileup) occurring inside a geometrically accurate representation of the CMS-2017 tracker. Here, we report on advances in both the computational and physics performance of mkFit, as well as progress toward integration with CMS production software. Recently we have improved the overall efficiency of the algorithm by preserving short track candidates at a relatively early stage rather than attempting to extend them over many layers. Moreover, mkFit formerly produced an excess of duplicate tracks; these are now explicitly removed in an additional processing step. We demonstrate that with these enhancements, mkFit becomes a suitable choice for the first iteration of CMS tracking, and eventually for later iterations as well. We plan to test this capability in the CMS High Level Trigger during Run 3 of the LHC, with an ultimate goal of using it in both the CMS HLT and offline reconstruction for the HL-LHC CMS tracker.

Commercial Cloud computing is becoming mainstream, with funding agencies moving beyond prototyping and starting to fund production campaigns, too. An important aspect of any scientific computing production campaign is data movement, both incoming and outgoing. And while the performance and cost of VMs is relatively well understood, the network performance and cost is not. This paper provides a characterization of networking in various regions of Amazon Web Services, Microsoft Azure and Google Cloud Platform, both between Cloud resources and major DTNs in the Pacific Research Platform, including OSG data federation caches in the network backbone, and inside the clouds themselves. The paper contains both a qualitative analysis of the results as well as latency and throughput measurements. It also includes an analysis of the costs involved with Cloud-based networking., Comment: 7 pages, 1 figure, 5 tables, to be published in CHEP19 proceedings

Building particle tracks is the most computationally intense step of event reconstruction at the LHC. With the increased instantaneous luminosity and associated increase in pileup expected from the High-Luminosity LHC, the computational challenge of track finding and fitting requires novel solutions. The current track reconstruction algorithms used at the LHC are based on Kalman filter methods that achieve good physics performance. By adapting the Kalman filter techniques for use on many-core SIMD architectures such as the Intel Xeon and Intel Xeon Phi and (to a limited degree) NVIDIA GPUs, we are able to obtain significant speedups and comparable physics performance. New optimizations, including a dedicated post-processing step to remove duplicate tracks, have improved the algorithm's performance even further. Here we report on the current structure and performance of the code and future plans for the algorithm., Comment: Submitted to proceedings of the 2019 Connecting the Dots and Workshop on Intelligent Trackers (CTD/WIT 2019); 6 pages, 4 figures

In the High-Luminosity Large Hadron Collider (HL-LHC), one of the most challenging computational problems is expected to be finding and fitting charged-particle tracks during event reconstruction. The methods currently in use at the LHC are based on the Kalman filter. Such methods have shown to be robust and to provide good physics performance, both in the trigger and offline. In order to improve computational performance, we explored Kalman-filter-based methods for track finding and fitting, adapted for many-core SIMD and SIMT architectures. Our adapted Kalman-filter-based software has obtained significant parallel speedups using such processors, e.g., Intel Xeon Phi, Intel Xeon SP (Scalable Processors) and (to a limited degree) NVIDIA GPUs. Recently, an effort has started towards the integration of our software into the CMS software framework, in view of its exploitation for the Run III of the LHC. Prior reports have shown that our software allows in fact for some significant improvements over the existing framework in terms of computational performance with comparable physics performance, even when applied to realistic detector configurations and event complexity. Here, we demonstrate that in such conditions physics performance can be further improved with respect to our prior reports, while retaining the improvements in computational performance, by making use of the knowledge of the detector and its geometry., Comment: Submitted to proceedings of 19th International Workshop on Advanced Computing and Analysis Techniques in Physics Research (ACAT 2019); 6 pages, 5 figures

Without significant changes to data organization, management, and access (DOMA), HEP experiments will find scientific output limited by how fast data can be accessed and digested by computational resources. In this white paper we discuss challenges in DOMA that HEP experiments, such as the HL-LHC, will face as well as potential ways to address them. A research and development timeline to assess these changes is also proposed., Comment: arXiv admin note: text overlap with arXiv:1712.06592

The High-Luminosity Large Hadron Collider at CERN will be characterized by greater pileup of events and higher occupancy, making the track reconstruction even more computationally demanding. Existing algorithms at the LHC are based on Kalman filter techniques with proven excellent physics performance under a variety of conditions. Starting in 2014, we have been developing Kalman-filter-based methods for track finding and fitting adapted for many-core SIMD processors that are becoming dominant in high-performance systems. This paper summarizes the latest extensions to our software that allow it to run on the realistic CMS-2017 tracker geometry using CMSSW-generated events, including pileup. The reconstructed tracks can be validated against either the CMSSW simulation that generated the hits, or the CMSSW reconstruction of the tracks. In general, the code's computational performance has continued to improve while the above capabilities were being added. We demonstrate that the present Kalman filter implementation is able to reconstruct events with comparable physics performance to CMSSW, while providing generally better computational performance. Further plans for advancing the software are discussed.

At the heart of experimental high energy physics (HEP) is the development of facilities and instrumentation that provide sensitivity to new phenomena. Our understanding of nature at its most fundamental level is advanced through the analysis and interpretation of data from sophisticated detectors in HEP experiments. The goal of data analysis systems is to realize the maximum possible scientific potential of the data within the constraints of computing and human resources in the least time. To achieve this goal, future analysis systems should empower physicists to access the data with a high level of interactivity, reproducibility and throughput capability. As part of the HEP Software Foundation Community White Paper process, a working group on Data Analysis and Interpretation was formed to assess the challenges and opportunities in HEP data analysis and develop a roadmap for activities in this area over the next decade. In this report, the key findings and recommendations of the Data Analysis and Interpretation Working Group are presented., Comment: arXiv admin note: text overlap with arXiv:1712.06592

Realizing the physics programs of the planned and upgraded high-energy physics (HEP) experiments over the next 10 years will require the HEP community to address a number of challenges in the area of software and computing. For this reason, the HEP software community has engaged in a planning process over the past two years, with the objective of identifying and prioritizing the research and development required to enable the next generation of HEP detectors to fulfill their full physics potential. The aim is to produce a Community White Paper which will describe the community strategy and a roadmap for software and computing research and development in HEP for the 2020s. The topics of event reconstruction and software triggers were considered by a joint working group and are summarized together in this document., Comment: Editors: Vladimir Gligorov and David Lange

Realizing the physics programs of the planned and upgraded high-energy physics (HEP) experiments over the next 10 years will require the HEP community to address a number of challenges in the area of software and computing. For this reason, the HEP software community has engaged in a planning process over the past two years, with the objective of identifying and prioritizing the research and development required to enable the next generation of HEP detectors to fulfill their full physics potential. The aim is to produce a Community White Paper which will describe the community strategy and a roadmap for software and computing research and development in HEP for the 2020s. The topics of event reconstruction and software triggers were considered by a joint working group and are summarized together in this document., Comment: Editors Vladimir Vava Gligorov and David Lange

Particle physics has an ambitious and broad experimental programme for the coming decades. This programme requires large investments in detector hardware, either to build new facilities and experiments, or to upgrade existing ones. Similarly, it requires commensurate investment in the R&D of software to acquire, manage, process, and analyse the shear amounts of data to be recorded. In planning for the HL-LHC in particular, it is critical that all of the collaborating stakeholders agree on the software goals and priorities, and that the efforts complement each other. In this spirit, this white paper describes the R&D activities required to prepare for this software upgrade.

Faced with physical and energy density limitations on clock speed, contemporary microprocessor designers have increasingly turned to on-chip parallelism for performance gains. Algorithms should accordingly be designed with ample amounts of fine-grained parallelism if they are to realize the full performance of the hardware. This requirement can be challenging for algorithms that are naturally expressed as a sequence of small-matrix operations, such as the Kalman filter methods widely in use in high-energy physics experiments. In the High-Luminosity Large Hadron Collider (HL-LHC), for example, one of the dominant computational problems is expected to be finding and fitting charged-particle tracks during event reconstruction; today, the most common track-finding methods are those based on the Kalman filter. Experience at the LHC, both in the trigger and offline, has shown that these methods are robust and provide high physics performance. Previously we reported the significant parallel speedups that resulted from our efforts to adapt Kalman-filter-based tracking to many-core architectures such as Intel Xeon Phi. Here we report on how effectively those techniques can be applied to more realistic detector configurations and event complexity., Comment: Accepted to the Proceedings of the 18th International Workshop on Advanced Computing and Analysis Techniques in Physics Research; 6 pages, 5 figures. arXiv admin note: text overlap with arXiv:1702.06359

During 2015 and 2016, the Laser Interferometer Gravitational-Wave Observatory (LIGO) conducted a three-month observing campaign. These observations delivered the first direct detection of gravitational waves from binary black hole mergers. To search for these signals, the LIGO Scientific Collaboration uses the PyCBC search pipeline. To deliver science results in a timely manner, LIGO collaborated with the Open Science Grid (OSG) to distribute the required computation across a series of dedicated, opportunistic, and allocated resources. To deliver the petabytes necessary for such a large-scale computation, our team deployed a distributed data access infrastructure based on the XRootD server suite and the CernVM File System (CVMFS). This data access strategy grew from simply accessing remote storage to a POSIX-based interface underpinned by distributed, secure caches across the OSG., Comment: 6 pages, 3 figures, submitted to PEARC17

For over a decade now, physical and energy constraints have limited clock speed improvements in commodity microprocessors. Instead, chipmakers have been pushed into producing lower-power, multi-core processors such as GPGPU, ARM and Intel MIC. Broad-based efforts from manufacturers and developers have been devoted to making these processors user-friendly enough to perform general computations. However, extracting performance from a larger number of cores, as well as specialized vector or SIMD units, requires special care in algorithm design and code optimization. One of the most computationally challenging problems in high-energy particle experiments is finding and fitting the charged-particle tracks during event reconstruction. This is expected to become by far the dominant problem in the High-Luminosity Large Hadron Collider (HL-LHC), for example. Today the most common track finding methods are those based on the Kalman filter. Experience with Kalman techniques on real tracking detector systems has shown that they are robust and provide high physics performance. This is why they are currently in use at the LHC, both in the trigger and offline. Previously we reported on the significant parallel speedups that resulted from our investigations to adapt Kalman filters to track fitting and track building on Intel Xeon and Xeon Phi. Here, we discuss our progresses toward the understanding of these processors and the new developments to port Kalman filter to NVIDIA GPUs., Comment: Submitted to proceedings of Connecting The Dots 2017 (CTD2017), Orsay. arXiv admin note: substantial text overlap with arXiv:1605.05508

Limits on power dissipation have pushed CPUs to grow in parallel processing capabilities rather than clock rate, leading to the rise of "manycore" or GPU-like processors. In order to achieve the best performance, applications must be able to take full advantage of vector units across multiple cores, or some analogous arrangement on an accelerator card. Such parallel performance is becoming a critical requirement for methods to reconstruct the tracks of charged particles at the Large Hadron Collider and, in the future, at the High Luminosity LHC. This is because the steady increase in luminosity is causing an exponential growth in the overall event reconstruction time, and tracking is by far the most demanding task for both online and offline processing. Many past and present collider experiments adopted Kalman filter-based algorithms for tracking because of their robustness and their excellent physics performance, especially for solid state detectors where material interactions play a significant role. We report on the progress of our studies towards a Kalman filter track reconstruction algorithm with optimal performance on manycore architectures. The combinatorial structure of these algorithms is not immediately compatible with an efficient SIMD (or SIMT) implementation; the challenge for us is to recast the existing software so it can readily generate hundreds of shared-memory threads that exploit the underlying instruction set of modern processors. We show how the data and associated tasks can be organized in a way that is conducive to both multithreading and vectorization. We demonstrate very good performance on Intel Xeon and Xeon Phi architectures, as well as promising first results on Nvidia GPUs., Comment: Proceedings of the 22nd International Conference on Computing in High Energy and Nuclear Physics, CHEP 2016; 8 pages, 9 figures

Power density constraints are limiting the performance improvements of modern CPUs. To address this we have seen the introduction of lower-power, multi-core processors such as GPGPU, ARM and Intel MIC. To stay within the power density limits but still obtain Moore's Law performance/price gains, it will be necessary to parallelize algorithms to exploit larger numbers of lightweight cores and specialized functions like large vector units. Track finding and fitting is one of the most computationally challenging problems for event reconstruction in particle physics. At the High-Luminosity Large Hadron Collider (HL-LHC), for example, this will be by far the dominant problem. The need for greater parallelism has driven investigations of very different track finding techniques such as Cellular Automata or Hough Transforms. The most common track finding techniques in use today, however, are those based on the Kalman Filter. Significant experience has been accumulated with these techniques on real tracking detector systems, both in the trigger and offline. They are known to provide high physics performance, are robust, and are in use today at the LHC. We report on porting these algorithms to new parallel architectures. Our previous investigations showed that, using optimized data structures, track fitting with a Kalman Filter can achieve large speedups both with Intel Xeon and Xeon Phi. Additionally, we have previously shown first attempts at track building with some speedup. We report here our progress towards an end-to-end track reconstruction algorithm fully exploiting vectorization and parallelization techniques in a simplified experimental environment., Comment: Submitted to proceedings of Connecting The Dots 2016 (CTD2016), Vienna. arXiv admin note: text overlap with arXiv:1601.08245

Power density constraints are limiting the performance improvements of modern CPUs. To address this we have seen the introduction of lower-power, multi-core processors such as GPGPU, ARM and Intel MIC. To stay within the power density limits but still obtain Moore's Law performance/price gains, it will be necessary to parallelize algorithms to exploit larger numbers of lightweight cores and specialized functions like large vector units. Track finding and fitting is one of the most computationally challenging problems for event reconstruction in particle physics. At the High-Luminosity Large Hadron Collider (HL-LHC), for example, this will be by far the dominant problem. The need for greater parallelism has driven investigations of very different track finding techniques such as Cellular Automata or Hough Transforms. The most common track finding techniques in use today, however, are those based on the Kalman Filter. Significant experience has been accumulated with these techniques on real tracking detector systems, both in the trigger and offline. They are known to provide high physics performance, are robust, and are in use today at the LHC. We report on porting these algorithms to new parallel architectures. Our previous investigations showed that, using optimized data structures, track fitting with Kalman Filter can achieve large speedups both with Intel Xeon and Xeon Phi. We report here our progress towards an end-to-end track reconstruction algorithm fully exploiting vectorization and parallelization techniques in a realistic experimental environment., Comment: Proceedings of the 2015 IEEE NSS/MIC Conference, San Diego, CA

Data access is key to science driven by distributed high-throughput computing (DHTC), an essential technology for many major research projects such as High Energy Physics (HEP) experiments. However, achieving efficient data access becomes quite difficult when many independent storage sites are involved because users are burdened with learning the intricacies of accessing each system and keeping careful track of data location. We present an alternate approach: the Any Data, Any Time, Anywhere infrastructure. Combining several existing software products, AAA presents a global, unified view of storage systems - a "data federation," a global filesystem for software delivery, and a workflow management system. We present how one HEP experiment, the Compact Muon Solenoid (CMS), is utilizing the AAA infrastructure and some simple performance metrics., Comment: 9 pages, 6 figures, submitted to 2nd IEEE/ACM International Symposium on Big Data Computing (BDC) 2015

This paper describes a programme to study the computing model in CMS after the next long shutdown near the end of the decade., Comment: Submitted to proceedings of the 21st International Conference on Computing in High Energy and Nuclear Physics (CHEP2015), Okinawa, Japan

Power density constraints are limiting the performance improvements of modern CPUs. To address this we have seen the introduction of lower-power, multi-core processors, but the future will be even more exciting. In order to stay within the power density limits but still obtain Moore's Law performance/price gains, it will be necessary to parallelize algorithms to exploit larger numbers of lightweight cores and specialized functions like large vector units. Example technologies today include Intel's Xeon Phi and GPGPUs. Track finding and fitting is one of the most computationally challenging problems for event reconstruction in particle physics. At the High Luminosity LHC, for example, this will be by far the dominant problem. The need for greater parallelism has driven investigations of very different track finding techniques including Cellular Automata or returning to Hough Transform. The most common track finding techniques in use today are however those based on the Kalman Filter. Significant experience has been accumulated with these techniques on real tracking detector systems, both in the trigger and offline. They are known to provide high physics performance, are robust and are exactly those being used today for the design of the tracking system for HL-LHC. Our previous investigations showed that, using optimized data structures, track fitting with Kalman Filter can achieve large speedup both with Intel Xeon and Xeon Phi. We report here our further progress towards an end-to-end track reconstruction algorithm fully exploiting vectorization and parallelization techniques in a realistic simulation setup.