Technical Information
The requirements for applying for a sprint are summarized in a separate document.
Useful insights e.g. lessons learned as well as FAQ are continuously added.
Information from the natESM hands-on session of the DKRZ User Workshop October 2022:
Input |
Download |
1- CUDA, OpenACC, and more: Many ways lead to GPUs Dr. Andreas Herten, JSC |
|
2- Performance analysis in a nutshell Michael Knobloch, JSC |
Executed Sprints
Already accepted sprints can be found here.
For information about the closed sprints beyond the PDF, please have a look in the Gitlab Wiki and in the Lessons learned.
Modell |
Service Description |
Documentation |
Analysis of ART code for GPU porting |
||
Online coupling mHM into ICON using YAC |
||
Optimise the data transfers between host (CPU) and device (GPU) |
||
Port FESOM 2.1 to JUWELS booster and Levante- GPU |
||
Port ParFlow to AMD GPUs, Inspection of RAPID Memory Manager and Hipification, Performance Analysis |