Computing demands for large scientific experiments, such as the CMS experiment at the CERN LHC, will increase dramatically in the next decades. To complement the future performance increases of software running on central processing units (CPUs), explorations of coprocessor usage in data processing hold great potential and interest. Coprocessors are a class of computer processors that supplement CPUs, often improving the execution of certain functions due to architectural design choices. We explore the approach of Services for Optimized Network Inference on Coprocessors (SONIC) and study the deployment of this as-a-service approach in large-scale data processing. In the studies, we take a data processing workflow of the CMS experiment and run the main workflow on CPUs, while offloading several machine learning (ML) inference tasks onto either remote or local coprocessors, specifically graphics processing units (GPUs). With experiments performed at Google Cloud, the Purdue Tier-2 computing center, and combinations of the two, we demonstrate the acceleration of these ML algorithms individually on coprocessors and the corresponding throughput improvement for the entire workflow. This approach can be easily generalized to different types of coprocessors and deployed on local CPUs without decreasing the throughput performance. We emphasize that the SONIC approach enables high coprocessor usage and enables the portability to run workflows on different types of coprocessors.
Note: When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external site maintained by the publisher.
Some full text articles may not yet be available without a charge during the embargo (administrative interval).
What is a DOI Number?
Some links on this page may take you to non-federal websites. Their policies may differ from this site.
-
Abstract Free, publicly-accessible full text available December 1, 2025 -
Abstract A measurement is presented of a ratio observable that provides a measure of the azimuthal correlations among jets with large transverse momentum
. This observable is measured in multijet events over the range of$$p_{\textrm{T}}$$ –$$p_{\textrm{T}} = 360$$ based on data collected by the CMS experiment in proton-proton collisions at a centre-of-mass energy of 13$$3170\,\text {Ge}\hspace{-.08em}\text {V} $$ , corresponding to an integrated luminosity of 134$$\,\text {Te}\hspace{-.08em}\text {V}$$ . The results are compared with predictions from Monte Carlo parton-shower event generator simulations, as well as with fixed-order perturbative quantum chromodynamics (pQCD) predictions at next-to-leading-order (NLO) accuracy obtained with different parton distribution functions (PDFs) and corrected for nonperturbative and electroweak effects. Data and theory agree within uncertainties. From the comparison of the measured observable with the pQCD prediction obtained with the NNPDF3.1 NLO PDFs, the strong coupling at the Z boson mass scale is$$\,\text {fb}^{-1}$$ , where the total uncertainty is dominated by the scale dependence of the fixed-order predictions. A test of the running of$$\alpha _\textrm{S} (m_{{\textrm{Z}}}) =0.1177 \pm 0.0013\, \text {(exp)} _{-0.0073}^{+0.0116} \,\text {(theo)} = 0.1177_{-0.0074}^{+0.0117}$$ in the$$\alpha _\textrm{S}$$ region shows no deviation from the expected NLO pQCD behaviour.$$\,\text {Te}\hspace{-.08em}\text {V}$$ Free, publicly-accessible full text available August 1, 2025 -
Abstract A search for
and$${\text {Z}{}{}} {\text {Z}{}{}} $$ production in the$${\text {Z}{}{}} {\text {H}{}{}} $$ final state is presented, where H is the standard model (SM) Higgs boson. The search uses an event sample of proton-proton collisions corresponding to an integrated luminosity of 133$${\text {b}{}{}} {\bar{{\text {b}{}{}}}{}{}} {\text {b}{}{}} {\bar{{\text {b}{}{}}}{}{}} $$ collected at a center-of-mass energy of 13$$\,\text {fb}^{-1}$$ with the CMS detector at the CERN LHC. The analysis introduces several novel techniques for deriving and validating a multi-dimensional background model based on control samples in data. A multiclass multivariate classifier customized for the$$\,\text {Te}\hspace{-.08em}\text {V}$$ final state is developed to derive the background model and extract the signal. The data are found to be consistent, within uncertainties, with the SM predictions. The observed (expected) upper limits at 95% confidence level are found to be 3.8 (3.8) and 5.0 (2.9) times the SM prediction for the$${\text {b}{}{}} {\bar{{\text {b}{}{}}}{}{}} {\text {b}{}{}} {\bar{{\text {b}{}{}}}{}{}} $$ and$${\text {Z}{}{}} {\text {Z}{}{}} $$ production cross sections, respectively.$${\text {Z}{}{}} {\text {H}{}{}} $$ Free, publicly-accessible full text available July 1, 2025 -
Abstract The measurement of Z boson production is presented as a method to determine the integrated luminosity of CMS data sets. The analysis uses proton–proton collision data, recorded by the CMS experiment at the CERN LHC in 2017 at a center-of-mass energy of 13
. Events with Z bosons decaying into a pair of muons are selected. The total number of Z bosons produced in a fiducial volume is determined, together with the identification efficiencies and correlations from the same data set, in small intervals of 20$$\,\text {Te\hspace{-.08em}V}$$ of integrated luminosity, thus facilitating the efficiency and rate measurement as a function of time and instantaneous luminosity. Using the ratio of the efficiency-corrected numbers of Z bosons, the precisely measured integrated luminosity of one data set is used to determine the luminosity of another. For the first time, a full quantitative uncertainty analysis of the use of Z bosons for the integrated luminosity measurement is performed. The uncertainty in the extrapolation between two data sets, recorded in 2017 at low and high instantaneous luminosity, is less than 0.5%. We show that the Z boson rate measurement constitutes a precise method, complementary to traditional methods, with the potential to improve the measurement of the integrated luminosity.$$\,\text {pb}^{-1}$$ Free, publicly-accessible full text available January 1, 2025 -
Abstract The strange quark content of the proton is probed through the measurement of the production cross section for a W boson and a charm (c) quark in proton–proton collisions at a center-of-mass energy of 13
. The analysis uses a data sample corresponding to a total integrated luminosity of 138$$\,\text {Te}\hspace{-.08em}\text {V}$$ collected with the CMS detector at the LHC. The W bosons are identified through their leptonic decays to an electron or a muon, and a neutrino. Charm jets are tagged using the presence of a muon or a secondary vertex inside the jet. The$$\,\text {fb}^{-1}$$ production cross section and the cross section ratio$$\hbox {W}+\hbox {c}$$ are measured inclusively and differentially as functions of the transverse momentum and the pseudorapidity of the lepton originating from the W boson decay. The precision of the measurements is improved with respect to previous studies, reaching 1% in$$R_\textrm{c}^{\pm }= \sigma ({\hbox {W}}^{+}+\bar{\text {c}})/\sigma (\hbox {W}^{-}+{\textrm{c}})$$ . The measurements are compared with theoretical predictions up to next-to-next-to-leading order in perturbative quantum chromodynamics.$$R_\textrm{c}^{\pm }= 0.950 \pm 0.005\,\text {(stat)} \pm 0.010 \,\text {(syst)} $$ Free, publicly-accessible full text available January 1, 2025 -
Abstract The mass of the top quark is measured in 36.3
of LHC proton–proton collision data collected with the CMS detector at$$\,\text {fb}^{-1}$$ . The measurement uses a sample of top quark pair candidate events containing one isolated electron or muon and at least four jets in the final state. For each event, the mass is reconstructed from a kinematic fit of the decay products to a top quark pair hypothesis. A profile likelihood method is applied using up to four observables per event to extract the top quark mass. The top quark mass is measured to be$$\sqrt{s}=13\,\text {Te}\hspace{-.08em}\text {V} $$ . This approach significantly improves the precision over previous measurements.$$171.77\pm 0.37\,\text {Ge}\hspace{-.08em}\text {V} $$ -
Abstract A generic search is presented for the associated production of a Z boson or a photon with an additional unspecified massive particle X,
, in proton-tagged events from proton–proton collisions at$${\textrm{pp}}\rightarrow {\textrm{pp}} +{{\textrm{Z}}}/\upgamma +{{\textrm{X}}} $$ , recorded in 2017 with the CMS detector and the CMS-TOTEM precision proton spectrometer. The missing mass spectrum is analysed in the 600–1600 GeV range and a fit is performed to search for possible deviations from the background expectation. No significant excess in data with respect to the background predictions has been observed. Model-independent upper limits on the visible production cross section of$$\sqrt{s}=13\, \textrm{TeV}$$ are set.$${\textrm{pp}}\rightarrow {\textrm{pp}} +{{\textrm{Z}}}/\upgamma +{{\textrm{X}}} $$ -
Abstract The production of Z bosons associated with jets is measured in $$\text {p}\text {p}$$ pp collisions at $$\sqrt{s}=13\,\text {Te}\hspace{-.08em}\text {V} $$ s = 13 Te V with data recorded with the CMS experiment at the LHC corresponding to an integrated luminosity of 36.3 $$\,\text {fb}^{-1}$$ fb - 1 . The multiplicity of jets with transverse momentum $$p_{\textrm{T}} > 30\,\text {Ge}\hspace{-.08em}\text {V} $$ p T > 30 Ge V is measured for different regions of the Z boson’s $$p_{\textrm{T}} (\text {Z })$$ p T ( Z ) , from lower than 10 $$\,\text {Ge}\hspace{-.08em}\text {V}$$ Ge V to higher than 100 $$\,\text {Ge}\hspace{-.08em}\text {V}$$ Ge V . The azimuthal correlation $$\varDelta \phi $$ Δ ϕ between the Z boson and the leading jet, as well as the correlations between the two leading jets are measured in three regions of $$p_{\textrm{T}} (\text {Z })$$ p T ( Z ) . The measurements are compared with several predictions at leading and next-to-leading orders, interfaced with parton showers. Predictions based on transverse-momentum dependent parton distributions and corresponding parton showers give a good description of the measurement in the regions where multiple parton interactions and higher jet multiplicities are not important. The effects of multiple parton interactions are shown to be important to correctly describe the measured spectra in the low $$p_{\textrm{T}} (\text {Z })$$ p T ( Z ) regions.more » « less
-
Abstract Multijet events at large transverse momentum ( $$p_{\textrm{T}}$$ p T ) are measured at $$\sqrt{s}=13\,\text {TeV} $$ s = 13 TeV using data recorded with the CMS detector at the LHC, corresponding to an integrated luminosity of $$36.3{\,\text {fb}^{-1}} $$ 36.3 fb - 1 . The multiplicity of jets with $$p_{\textrm{T}} >50\,\text {GeV} $$ p T > 50 GeV that are produced in association with a high- $$p_{\textrm{T}}$$ p T dijet system is measured in various ranges of the $$p_{\textrm{T}}$$ p T of the jet with the highest transverse momentum and as a function of the azimuthal angle difference $$\varDelta \phi _{1,2}$$ Δ ϕ 1 , 2 between the two highest $$p_{\textrm{T}}$$ p T jets in the dijet system. The differential production cross sections are measured as a function of the transverse momenta of the four highest $$p_{\textrm{T}}$$ p T jets. The measurements are compared with leading and next-to-leading order matrix element calculations supplemented with simulations of parton shower, hadronization, and multiparton interactions. In addition, the measurements are compared with next-to-leading order matrix element calculations combined with transverse-momentum dependent parton densities and transverse-momentum dependent parton shower.more » « less