Robustness of deep learning algorithms in astronomy - galaxy morphology studies
Explore the source record for details and available documents.
Ciprijanoivc, A.↗
Engineering topics
Pedro, Kevin
Publications and source records attributed to Pedro, Kevin.
Compressing deep neural networks on FPGAs to binary and ternary precision with hls4ml
We present the implementation of binary and ternary neural networks in the hls4ml library, designed to automatically convert deep neural network models to digital circuits with FPGA firmware. Starting from benchmark models trained with floating point precision, we investigate different strategies to reduce the network's resource consumption by reducing the numerical precision of the network parameters to binary or ternary. We discuss the trade-off between model accuracy and resource consumption. In addition, we show how to balance between latency and accuracy by retaining full precision on a selected subset of network components. As an example, we consider two multiclass classification tasks: handwritten digit recognition with the MNIST data set and jet identification with simulated proton-proton collisions at the CERN Large Hadron Collider. The binary and ternary implementation has similar performance to the higher precision implementation while using drastically fewer FPGA resources.
72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
FPGAs-as-a-Service Toolkit (FaaST)
Computing needs for high energy physics are already intensive and are expected to increase drastically in the coming years. In this context, heterogeneous computing, specifically as-a-service computing, has the potential for significant gains over traditional computing models. Although previous studies and packages in the field of heterogeneous computing have focused on GPUs as accelerators, FPGAs are an extremely promising option as well. A series of workflows are developed to establish the performance capabilities of FPGAs as a service. Multiple different devices and a range of algorithms for use in high energy physics are studied. For a small, dense network, the throughput can be improved by an order of magnitude with respect to GPUs as a service. For large convolutional networks, the throughput is found to be comparable to GPUs as a service. This work represents the first open-source FPGAs-as-a-service toolkit.
46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
HL-LHC Computing Review: Common Tools and Community Software
Common and community software packages, such as ROOT, Geant4 and event generators have been a key part of the LHC's success so far and continued development and optimisation will be critical in the future. The challenges are driven by an ambitious physics programme, notably the LHC accelerator upgrade to high-luminosity, HL-LHC, and the corresponding detector upgrades of ATLAS and CMS. In this document we address the issues for software that is used in multiple experiments (usually even more widely than ATLAS and CMS) and maintained by teams of developers who are either not linked to a particular experiment or who contribute to common software within the context of their experiment activity. We also give space to general considerations for future software and projects that tackle upcoming challenges, no matter who writes it, which is an area where community convergence on best practice is extremely useful.
72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
Coffea -- Columnar Object Framework For Effective Analysis
The coffea framework provides a new approach to High-Energy Physics analysis, via columnar operations, that improves time-to-insight, scalability, portability, and reproducibility of analysis. It is implemented with the Python programming language, the scientific python package ecosystem, and commodity big data technologies. To achieve this suite of improvements across many use cases, coffea takes a factorized approach, separating the analysis implementation and data delivery scheme. All analysis operations are implemented using the NumPy or awkward-array packages which are wrapped to yield user code whose purpose is quickly intuited. Various data delivery schemes are wrapped into a common front-end which accepts user inputs and code, and returns user defined outputs. We will discuss our experience in implementing analysis of CMS data using the coffea framework along with a discussion of the user experience and future directions.
72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
Integration and Performance of New Technologies in the CMS Simulation [Slides]
CMS upgrade simulation for HL-LHC may be 2–3×slower than Run 2. CMS integration tests of GeantV met all goals: 1. Co-development ensured compatible threading models & interfaces, 2. Similar speedup measured in full experiment software framework, and 3. Efficient path to integration was established. GeantV prototype is a useful demonstrator. The next step is a GeantX project for HPCs w/ GPUs.
71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
COFFEA - Columnar Object Framework For Effective Analysis [Slides]
The COFFEA Framework provides a new approach to HEP analysis, via columnar operations, that improves time-to-insight, scalability, portability, and reproducibility of analysis. It is implemented with the Python programming language and commodity big data technologies such as Apache Spark and NoSQL databases. To achieve this suite of improvements across many use cases, COFFEA takes a factorized approach, separating the analysis implementation and data delivery scheme. All analysis operations are implemented using the NumPy or awkward-array packages which are wrapped to yield user code whose purpose is quickly intuited. Various data delivery schemes are wrapped into a common front-end which accepts user inputs and code, and returns user defined outputs. We will present published results from analysis of CMS data using the COFFEA framework along with a discussion of metrics and the user experience of arriving at those results with columnar analysis.
97 MATHEMATICS AND COMPUTING↗
FPGA-Accelerated Machine Learning Inference as a Service for Particle Physics Computing
Large-scale particle physics experiments face challenging demands for high-throughput computing resources both now and in the future. New heterogeneous computing paradigms on dedicated hardware with increased parallelization, such as Field Programmable Gate Arrays (FPGAs), offer exciting solutions with large potential gains. The growing applications of machine learning algorithms in particle physics for simulation, reconstruction, and analysis are naturally deployed on such platforms. We demonstrate that the acceleration of machine learning inference as a web service represents a heterogeneous computing solution for particle physics experiments that potentially requires minimal modification to the current computing model. As examples, we retrain the ResNet-50 convolutional neural network to demonstrate state-of-the-art performance for top quark jet tagging at the LHC and apply a ResNet-50 model with transfer learning for neutrino event classification. Using Project Brainwave by Microsoft to accelerate the ResNet-50 image classification model, we achieve average inference times of 60 (10) ms with our experimental physics software framework using Brainwave as a cloud (edge or on-premises) service, representing an improvement by a factor of approximately 30 (175) in model inference latency over traditional CPU inference in current experimental hardware. A single FPGA service accessed by many CPUs achieves a throughput of 600–700 inferences per second using an image batch of one, comparable to large batch-size GPU throughput and significantly better than small batch-size GPU throughput. Deployed as an edge or cloud service for the particle physics computing model, coprocessor accelerators can have a higher duty cycle and are potentially much more cost-effective.
46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
Recent progress with the top to bottom approach to vectorization in GeantV
SIMD acceleration can potentially boost by factors the application throughput. Achieving efficient SIMD vectorization for scalar code with complex data flow and branching logic, goes however way beyond breaking some loop dependencies and relying on the compiler. Since the refactoring effort scales with the number of lines of code, it is important to understand what kind of performance gains can be expected in such complex cases. We started to investigate a couple of years ago a top to bottom vectorization approach to particle transport simulation. Percolating vector data to algorithms was mandatory since not all the components can internally vectorize. Vectorizing low-level algorithms is certainly necessary, but not sufficient to achieve relevant SIMD gains. In addition, the overheads for maintaining the concurrent vector data flow and copy data have to be minimized. In the context of a vectorization R&D for simulation we developed a framework to allow different categories of scalar and vectorized components to co-exist, dealing with data flow management and real-time heuristic optimizations. The paper describes our approach on coordinating SIMD vectorization at framework level, making a detailed quantitative analysis of the SIMD gain versus overheads, with a breakdown by components in terms of geometry, physics and magnetic field propagation. We also present the more general context of this R&D work and goals for 2018.
97 MATHEMATICS AND COMPUTING↗
Electromagnetic physics vectorization in the GeantV transport framework
The development of the GeantV Electromagnetic (EM) physics package has evolved following two necessary paths towards code modernization. A first phase required the revision of the main electromagnetic physics models and their implementation. The main objectives were to improve their accuracy, extend them to the new high-energy frontier posed by the Future Circular Collider (FCC) programme and allow a better adaptation to a multi-particle flow. Most of the EM physics models in GeantV have been reviewed from theoretical perspective and rewritten with vector-friendly implementations, being now available in scalar mode in the alpha release. The second phase consists of a thorough investigation on the possibility to vectorise the most CPU-intensive physics code parts, such as final state sampling. We have shown the feasibility of implementing electromagnetic physics models that take advantage of SIMD/SIMT architectures, thus obtaining gains in performance. After this phase, the time has come for the GeantV project to take a step forward towards the final proof of concept. This takes shape through the testing of the full simulation chain (transport + physics + geometry) running in vectorized mode. In this paper we will present the first benchmark results obtained after vectorizing a full set of electromagnetic physics models.
71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
Current and Future Performance of the CMS Simulation
The CMS full simulation using Geant4 has delivered billions of simulated events for analysis during Runs 1 and 2 of the LHC. However, the HL-LHC dataset will be an order of magnitude larger, with a similar increase in occupancy per event. In addition, the upgraded CMS detector will be considerably more complex, with an extended silicon tracker and a high granularity calorimeter in the endcap region. Increases in conventional computing resources are subject to both technological and budgetary limitations, so novel approaches are needed to improve software efficiency and to take advantage of new architectures and heterogeneous resources. Several projects are in development to address these needs, including the vectorized geometry library Vec-Geom and the GeantV transport engine, which uses track-level parallelization. The current computing performance of the CMS simulation will be presented as a baseline, along with an overview of the various optimizations already available for Geant4. Finally, the progress and outlook for integrating VecGeom and GeantV in the CMS software framework will be discussed.
72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗