Title: GPU-based First Collision Detection in Parton Cascade in Heavy-ion Collisions

Year of Publication: Nov - 2014
Page Numbers: 12-15
Authors: Qing-Jun Liu, Fang Liu, Ning-Ming Nie, Chun-Bao Zhou, Wei-Qin Zhao
Conference Name: The International Conference on Electrical, Electronics, Computer Engineering and their Applications (EECEA2014)
- Malaysia

Abstract:


Zhang's Parton Cascade (ZPC) is an essential module in A Multi-Phase Transport (AMPT) model, which applies Monte-Carlo methods for simulating ultra-relativistic heavy-ion collisions. Our ultimate goal is to accelerate ZPC, and we start with the acceleration of the detection of two-parton collisions in the cascade. In this work, using CUDA C, we parallelized and then tested the code for the detection of the first two-parton collision in the ZPC-based simulation of parton cascade. We compared results from the parallelized code and the serial code programmed in C. The comparison has shown for the first time that a significant speedup can be achieved by utilizing GPU as a co-processor for the detection of the first two-parton collision in the ZPC-based simulation of parton cascade in heavy-ion collisions of Pb - Pb at sqrt(sNN)=2.76 TeV, which is the top colliding energy for heavy-ion collisions at the Large Hadron Collider (LHC). The parallelized code is readily available per request, and will be integrated into GPU-based simulation of parton cascade in the near future. This study clearly indicates that ZPC may be accelerated and so is the event generation of Monte-Carlo events for heavy-ion collisions at the LHC.