Daniel Klocke, Claudia Frauen, Jan Frederik Engels, Dmitry Alexeev, Rene Redler, Reiner Schnur, Helmuth Haak, Luis Kornblueh, Nils Brueggemann, Fatemeh Chegini, Manoel Roemmer, Lars Hoffmann, Sabine Griessbach, Mathis Bode, Jonathan Coles, Miguel Gila, William Sawyer, Alexandru Calotoiu, Yakup Budanaz, Pratyai Mazumder, Marcin Copik, Benjamin Weber, Andreas Herten, Hendryk Bockelmann, Torsten Hoefler, Cathy Hohenegger, Bjorn Stevens:
Computing the Full Earth System at 1km Resolution
(In Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis (SC'25), presented in St. Louis, MO, USA, Nov. 2025) Gordon Bell Prize for Climate Modelling
Abstract
The increasing complexity of machine learning models and the proliferation of diverse hardware architectures (CPUs, GPUs, accelerators) make achieving optimal performance a significant challenge. Heterogeneity in instruction sets, specialized kernel requirements for different data types and model features (e.g., sparsity, quantization), and architecture-specific optimizations complicate performance tuning. Manual optimization is resource-intensive, while existing automatic approaches often rely on complex hardware-specific heuristics and uninterpretable intermediate representations, hindering performance portability. We introduce PerfLLM, a novel automatic optimization methodology leveraging Large Language Models (LLMs) and Reinforcement Learning (RL). Central to this is PerfDojo, an environment framing optimization as an RL game using a human-readable, mathematically-inspired code representation that guarantees semantic validity through transformations. This allows effective optimization without prior hardware knowledge, facilitating both human analysis and RL agent training. We demonstrate PerfLLM's ability to achieve significant performance gains across diverse CPU (x86, Arm, RISC-V) and GPU architectures.
Documents
download article:
Recorded talk (best effort)
BibTeX
@inproceedings{2025gbclimate, author={Daniel Klocke and Claudia Frauen and Jan Frederik Engels and Dmitry Alexeev and Rene Redler and Reiner Schnur and Helmuth Haak and Luis Kornblueh and Nils Brueggemann and Fatemeh Chegini and Manoel Roemmer and Lars Hoffmann and Sabine Griessbach and Mathis Bode and Jonathan Coles and Miguel Gila and William Sawyer and Alexandru Calotoiu and Yakup Budanaz and Pratyai Mazumder and Marcin Copik and Benjamin Weber and Andreas Herten and Hendryk Bockelmann and Torsten Hoefler and Cathy Hohenegger and Bjorn Stevens}, title={{Computing the Full Earth System at 1km Resolution}}, year={2025}, month={Nov.}, booktitle={Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis (SC'25)}, location={St. Louis, MO, USA}, source={http://www.unixer.de/~htor/publications/}, }
