GPU coprocessors as a service for deep learning inference in high energy physics

Krupa, Jeffrey; Lin, Kelvin; Flechas, Maria Acosta; Dinsmore, Jack; Duarte, Javier; Harris, Philip; Hauck, Scott; Holzman, Burt; Hsu, Shih-Chieh; Klijnsma, Thomas; Liu, Mia; Pedro, Kevin; Rankin, Dylan; Suaysom, Natchanon; Trahms, Matt; Tran, Nhan

doi:10.48550/arxiv.2007.10359

Cited by 5 publications

(7 citation statements)

References 49 publications

(57 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Continued development in this direction may allow such algorithms to be used effectively in future computing workflows [32] and the Level-1 trigger at the LHC. In future work, we plan to study detailed comparisons of the two implementations based on the same model, as well as comparing to GPU coprocessors [33]. Other optimizations of the GNN model may also be possible, such as more efficient architectures [30] and use of quantization-aware training [31,34] to reduce the necessary precision.…”

Section: Discussionmentioning

confidence: 99%

Accelerated Charged Particle Tracking with Graph Neural Networks on FPGAs

Heintz,

Razavimaleki,

Duarte

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

We develop and study FPGA implementations of algorithms for charged particle tracking based on graph neural networks. The two complementary FPGA designs are based on OpenCL, a framework for writing programs that execute across heterogeneous platforms, and hls4ml, a high-level-synthesis-based compiler for neural network to firmware conversion. We evaluate and compare the resource usage, latency, and tracking performance of our implementations based on a benchmark dataset. We find a considerable speedup over CPU-based execution is possible, potentially enabling such algorithms to be used effectively in future computing workflows and the FPGA-based Level-1 trigger at the CERN Large Hadron Collider.

show abstract

Section: Discussionmentioning

confidence: 99%

Accelerated Charged Particle Tracking with Graph Neural Networks on FPGAs

Heintz,

Razavimaleki,

Duarte

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…The feasibility of the as-a-service computing model for HEP workflows has been previously demonstrated using SONIC to interact with a GPU-based server for inference [17]. The server/client design employed within this paper is similar to previous work, allowing for a direct comparison of the performance.…”

Section: Related Workmentioning

confidence: 99%

“…Summary of the performance of FaaST servers in terms of events and inferences per second, and bandwidth. Results for performance on GPUs are taken from Ref [17]…”

mentioning

confidence: 99%

FPGAs-as-a-Service Toolkit (FaaST)

Rankin,

Krupa,

Harris

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

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.

show abstract

“…This includes GPUs and potentially even fieldprogrammable gate arrays (FPGAs) or ML-specific processors such as the GraphCore intelligence processing units (IPUs) [27] through specialized ML compilers [28][29][30]. These coprocessing accelerators can be integrated into existing CPU-based experimental software frameworks as a scalable service that grows to meet the transient demand [31][32][33]. Fake rate…”

Section: Charged Hadronsmentioning

confidence: 99%

MLPF: Efficient machine-learned particle-flow reconstruction using graph neural networks

Pata,

Duarte,

Vlimant

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

In general-purpose particle detectors, the particle flow algorithm may be used to reconstruct a coherent particle-level view of the event by combining information from the calorimeters and the trackers, significantly improving the detector resolution for jets and the missing transverse momentum. In view of the planned high-luminosity upgrade of the CERN Large Hadron Collider, it is necessary to revisit existing reconstruction algorithms and ensure that both the physics and computational performance are sufficient in a high-pileup environment. Recent developments in machine learning may offer a prospect for efficient event reconstruction based on parametric models. We introduce MLPF, an end-to-end trainable machine-learned particle flow algorithm for reconstructing particle flow candidates based on parallelizable, computationally efficient, scalable graph neural networks and a multi-task objective. We report the physics and computational performance of the MLPF algorithm on on a synthetic dataset of tt events in HL-LHC running conditions, including the simulation of multiple interaction effects, and discuss potential next steps and considerations towards ML-based reconstruction in a general-purpose particle detector.

show abstract

GPU coprocessors as a service for deep learning inference in high energy physics

Cited by 5 publications

References 49 publications

Accelerated Charged Particle Tracking with Graph Neural Networks on FPGAs

Accelerated Charged Particle Tracking with Graph Neural Networks on FPGAs

FPGAs-as-a-Service Toolkit (FaaST)

MLPF: Efficient machine-learned particle-flow reconstruction using graph neural networks

Contact Info

Product

Resources

About