PreprintCC BY-NC-ND 4.0arXiv
Towards Hadronisation on GPUs
Abstract
Complete event generation on hardware accelerators such as GPUs requires vectorising the algorithms of components such as parton showers and hadronisation. To this end, I implemented a simplified cluster hadronisation and hadronic decay model in the GPU parton shower program GAPS, with one thread per event and each stage of the algorithm as a separate kernel. I compared simulation results for e+e- → hadrons at the Z pole with the Herwig event generator and with ALEPH and L3 data, and measured execution times on an NVIDIA V100 GPU. The results agree well with Herwig, and GAPS generates one million hadron-level events (NLO hard process, shower, hadronisation and decays) in about four seconds on this GPU.
Related
- Anomalous soft photon production can be understood as photon emission during hadronisation
- Hadronisation of in-medium cc̅ pairs to the exotic X(3872)
- The role of strangeness and baryon enhancement in heavy-quark hadronisation from pp to Pb−Pb collisions with ALICE
- Heavy-flavour production and correlations in pp collisions: precision tests of pQCD and hadronisation with ALICE
- MadVfold: accelerating NLO event generation and reducing negative weights with SIMD vectorization and GPUs
BibTeX
@misc{sule2026towards,
title = {{Towards Hadronisation on GPUs}},
author = {Siddharth Sule},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.02318},
eprint = {2610.02318},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.02318},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - UNPB
TI - Towards Hadronisation on GPUs
AU - Siddharth Sule
PY - 2026
DA - 2026/10/01
PB - arXiv
DO - 10.48550/arxiv.2610.02318
UR - https://arxiv.org/abs/2610.02318
AB - Complete event generation on hardware accelerators such as GPUs requires vectorising the algorithms of components such as parton showers and hadronisation. To this end, I implemented a simplified cluster hadronisation and hadronic decay model in the GPU parton shower program GAPS, with one thread per event and each stage of the algorithm as a separate kernel. I compared simulation results for e^+e^- → hadrons at the Z pole with the Herwig event generator and with ALEPH and L3 data, and measured execution times on an NVIDIA V100 GPU. The results agree well with Herwig, and GAPS generates one million hadron-level events (NLO hard process, shower, hadronisation and decays) in about four seconds on this GPU.
ER -