31 Jul 2023
 | 31 Jul 2023
Status: this preprint is open for discussion.

Parflow 3.9: development of lightweight embedded DSLs for geoscientific models

Zbigniew P. Piotrowski, Jaro Hokkanen, Daniel Caviedes-Voullieme, Olaf Stein, and Stefan Kollet

Abstract. Recognizing the leap in high-performance computing with accelerated co-processors, we propose a lightweight approach to adapt legacy codes to next generation hardware and achieve efficiently a high degree of performance portability. We focus on abstracting the computing kernels at the loop levels based on the lightweight, preprocessor-based embedded Domain Specific Language (eDSL) concept in conjunction with Unified Memory management. We outline a set of code pre-adaptations that facilitate the proposed abstraction. In two geophysical code applications programmed in C and Fortran, we demonstrate the efficiency of the eDSL approach in adaptation to NVIDIA GPUs with: native CUDA and Kokkos eDSL backends achieving up to 10–30 fold speedup. Our experience suggests that the proposed lightweight eDSL code adaptation is less expensive in terms of Full Time Equivalent of effort than adaptation based on complex DSL approaches, even if no earlier GPU competence exists.

Zbigniew P. Piotrowski et al.

Status: open (until 25 Oct 2023)

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse
  • RC1: 'Comment on egusphere-2023-1079', Anonymous Referee #1, 22 Sep 2023 reply

Zbigniew P. Piotrowski et al.

Zbigniew P. Piotrowski et al.


Total article views: 277 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
209 60 8 277 3 3
  • HTML: 209
  • PDF: 60
  • XML: 8
  • Total: 277
  • BibTeX: 3
  • EndNote: 3
Views and downloads (calculated since 31 Jul 2023)
Cumulative views and downloads (calculated since 31 Jul 2023)

Viewed (geographical distribution)

Total article views: 268 (including HTML, PDF, and XML) Thereof 268 with geography defined and 0 with unknown origin.
Country # Views %
  • 1
Latest update: 30 Sep 2023
Short summary
The computer programs capable of simulation of Earth system components evolve, adapting new fundamental science concepts and more observational data on more and more powerful computer hardware. Adaptation of a large scientific program to a new type of hardware is costly. In this work we propose cheap and simple but effective strategy that enable computation using graphic processing units, based on automated program code modification. This results in better resolution and/or longer predictions.