Speaker
Description
Computing centres, including those used to process High-Energy Physics data and simulations, are increasingly providing significant fractions of their computing resources using hardware architectures other than x86 CPUs, with GPUs being a commonly available alternative. GPUs can provide excellent computational performance at a good price point for tasks that can be suitably parallelized. Charged particle (track) reconstruction is a computationally expensive component of HEP data reconstruction, and thus needs to use available resources in an efficient way.
In this talk, we will present an implementation of a full Kalman filter for track fitting using CUDA and running on GPUs. This utilizes the ACTS (A Common Tracking Software) toolkit; an open source and experiment-independent toolkit for track reconstruction. The implementation details and parallelization approach will be presented, along with the specific challenges for such an implementation. Detailed performance benchmarking results showing encouraging performance gains over a CPU-based implementation for representative configurations will be highlighted. We will also discuss the challenges and future directions for these studies including more complex and realistic scenarios and anticipated developments to software frameworks and standards which may open up possibilities for greater flexibility and improved performance.
A manuscript about this work has been accepted by the journal of Computer and Software for Big Science. See the preprint in Ref.[1].
[1] Xiaocong Ai, Georgiana Mania, Heather M. Gray, Michael Kuhn, Nicholas Styles, A GPU-based Kalman Filter for Track Fitting, arXiv: 2105.01796
Significance
The talk will given an overview of a complete study for which a manuscript has been submitted to CSBS recently.
References
Xiaocong Ai, Georgiana Mania, Heather M. Gray, Michael Kuhn, Nicholas Styles, A GPU-based Kalman Filter for Track Fitting, arXiv: 2105.01796
| Speaker time zone | Compatible with Europe |
|---|
