Search NASA⌕ Search

Engineering topics

Piperov, Stefan

Publications and source records attributed to Piperov, Stefan.

Deployment of inference as a service at the US CMS Tier-2 data centers

Coprocessors, especially GPUs, will be a vital ingredient of data production workflows at the HL-LHC. At CMS, the GPU-as-a-service approach for production workflows is implemented by the SONIC project (Services for Optimized Network Inference on Coprocessors). SONIC provides a mechanism for outsourcing computationally demanding algorithms, such as neural network inference, to remote servers, where requests from multiple clients are intelligently distributed across multiple GPUs by a load-balancing service. This talk highlights the recent progress in deploying SONIC at selected U.S. CMS Tier-2 data centers. Using realistic CMS Run3 data processing workflows, such as those containing transformer-based algorithms, we demonstrate how SONIC is integrated into the production-like environment to enable accelerated inference offloading. We will present developments from both the client and server sides, including production job and data center configurations for NVIDIA and AMD GPUs. We will also present performance scaling benchmarks and discuss the challenges of operating SONIC in CMS production, such as server discovery, GPU saturation, fallback server logic, etc.

Holzman, Burt↗

Calorimeter Simulation with Hadrons in CMS

CMS is using Geant4 to simulate the detector setup for the forthcoming data from the LHC. Validation of physics processes inside Geant4 is a major concern in view of getting a proper description of jets and missing energy for signal and background events. This is done by carrying out an extensive studies with test beam using the prototypes or real detector modules of the CMS calorimeter. These data are matched with Geant4 predictions using the same framework that is used for the entire CMS detector. Tuning of the Geant4 models is carried out and steps to be used in reproducing detector signals are defined in view of measurements of energy response, energy resolution, transverse and longitudinal shower profiles for a variety of hadron beams over a broad energy spectrum between 2 to 300 GeV/c. The tuned Monte Carlo predictions match many of these measurements within systematic uncertainties.

43 PARTICLE ACCELERATORS↗

Geant4 validation with CMS calorimeters test-beam data

CMS experiment is using Geant4 for Monte-Carlo simulation of the detector setup. Validation of physics processes describing hadronic showers is a major concern in view of getting a proper description of jets and missing energy for signal and background events. This is done by carrying out an extensive studies with test beam using the prototypes or real detector modules of the CMS calorimeter. These data are matched with Geant4 predictions. Tuning of the Geant4 models is carried out and steps to be used in reproducing detector signals are defined in view of measurements of energy response, energy resolution, transverse and longitudinal shower profiles for a variety of hadron beams over a broad energy spectrum between 2 to 300 GeV/c.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗