ALICE : online-offline processing for Run 3

TL;DR

This paper discusses ALICE's new online-offline data processing strategy for Run 3, which involves handling significantly increased data rates with GPU acceleration for real-time and post-processing reconstruction tasks.

Contribution

It introduces a dual-phase reconstruction approach for high-rate data, leveraging GPUs for both online and offline processing to meet the challenges of increased data volume and complexity.

Findings

Successful implementation of GPU-accelerated online reconstruction

Effective data compression and calibration techniques for high-rate data

Enhanced processing pipeline for continuous readout data

Abstract

ALICE will increase the data-taking rate for Run 3 significantly to 50 kHz continuous readout of minimum bias Pb--Pb collisions. The foreseen reconstruction strategy consists of 2 phases: a first synchronous online reconstruction stage during data-taking enabling detector calibration, and a posterior calibrated asynchronous reconstruction stage. The main challenges include processing and compression of 50 times more events per second than in Run 2, sophisticated compression and removal of TPC data not use for physics, tracking of TPC data in continuous readout, the TPC space-charge distortion calibrations, and in general running more reconstruction steps online compared to Run 2. ALICE will leverage GPUs to facilitate the synchronous processing with the available resources. In order to achieve the best utilization of the computing farm, we plan to offload also several steps of the asynchronous reconstruction to the GPU. This paper gives an overview of the important processing steps during synchronous and asynchronous reconstruction and of the required computing capabilities.