Skip to main content

Debugging Atmospheric Chemistry Models on HPC Clusters

We supported the RIFS research group in analyzing and resolving intermittent errors after migrating complex atmospheric chemistry models to a new high-performance computing cluster. In doing so, we bridged the gap between scientific application logic and low-level system debugging.

Client/Company/Industry

Research Institute for Sustainability (RIFS)

Duration

2 weeks

Product

Service

Expertise

Software Development

Goal

The goal of the project was to identify and resolve the cause of intermittent errors that occurred after the migration to a new high-performance computing cluster. This was intended to restore the reliable execution of the model simulations on the new infrastructure.

Tasks

  • Analyzing intermittent errors after the migration of the simulation software to a new high-performance computing cluster
  • Familiarizing ourselves with the domain context and the intended behavior of the scientific code
  • Performing an in-depth analysis of the interaction between application code, system libraries, and cluster hardware
  • Identifying the technical root cause of the problem
  • Resolving the identified issue
  • Presenting the results in a clear and understandable way for the research team and the cluster administrators

Challenges

A key challenge was that, after the transition to a new high-performance computing cluster, an incompatible MPI configuration caused runtime issues. The root cause was therefore not in the scientific model itself, but in the interaction between the scientific code, system libraries, and the cluster environment. This required an in-depth analysis of the low-level system configuration.

Programming Languages

Bash

Technologies

MPI

Project Image

Schematic representation of a high-performance computing cluster running scientific simulation software.

Similar problem?

Contact us

Takeaway

The project identified and resolved the cause of the errors, allowing the model simulations to run reliably again on the high-performance computing cluster. It demonstrated the value of bridging scientific software development and low-level HPC analysis.

Similar Projects

Project Image

RIM2D - Highly Efficient 2D Hydraulic Simulation of Fluvial, Pluvial, and Urban Flooding

Hydrodynamic Simulation Web Application Geodata GPU Computing

RIM2D is an existing, highly efficient 2D hydraulic simulation model for fluvial, pluvial, and urban flooding. As part of a strategic partnership, we supported the extension of the research code with a web application and a cloud-based GPU simulation environment, enabling its transition into a market-ready product.

Project Image

Data Lake for Geoscience Research Data Management

Datalake Research Data Management Geodata Cloud-Native Open-Source

We developed an Open-Source S3-based data lake solution for the centralized ingestion, categorization, and searchability of data. The goal was to automate and improve manual data management through an integrated architecture with workflow orchestration, data cataloging, and access control.

Back To Top