RSE Meetup

Thursday 30 Nov 2023, 15:00 → 17:15 Europe/Berlin

HZDR Conference Building

Do you deal with software for science? -- Let's talk about it!

---

We aim to foster connections and ignite discussions among our community. Do you have a burning topic that you're eager to share with us? Here's your chance! 

We're offering 3-5 slots for 5-minute pitch talks to pique everyone's interest. Afterward, we'll break into smaller groups for more in-depth discussions. Submit your topic now and let's spark engaging conversations! Topics may include, but are not limited to:

• Interesting Features in the World of Software Engineering
• Unraveling Your Toughest SE Challenges
• Sharing Your Brilliant Solutions to SE Conundrums
• Exploring Any SE-Related Topics

 

Your input is invaluable to us! Feel free to contribute your topic via an issue right here. We've even provided slide templates in the repository to help you along the way.

Our ultimate goal is to create a framework that welcomes both experts and beginners who share an interest in software-related topics. Join us as we cultivate meaningful connections and knowledge-sharing within our community.

---

TLDR

Add your topic here (direct link)

---

Link to event overview presentation

---

 

Zoom Link

 

 

 

15:00 → 15:15 Welcome

Speaker: Thomas Foerster (HZDR)

15:15 → 15:35 Pitch Talks

15:15 [Pitch] Collaborative Graph Data Editing

Research institutions can connect all their data in one large graph (OLG) if they manage to design and agree on common semantics and graph structures collaboratively. We have planned and begun implementing a software stack that will enable such a scenario.
In this stack, a graph data editor will allow users to edit their data as relatively small and contain "graph data sets" which are effectively sub-graphs of the OLG. A git+GitLab management system will enable the collaborative work of multiple people on the same graph data sets. However, in order to maintain the OLG as the single source of truth, a central synchronization management needs to be developed.

Speaker: Leon Steinmeier (Helmholtz Institute Freiberg)

RSE_meetup_2023-11-30_pitch_Leon_Steinmeier.pdf

15:20 [Pitch] There is no excuse why your software is not running on CPU and GPU!

In the ever-evolving landscape of research, the execution of scalable software on the largest computer clusters is indispensable.
However, the vast array of available hardware and programming models poses a challenge, not only in terms of finding proficient research software engineers but also in navigating the rapidly changing terrain of technology. Gone are the days when we could afford the luxury of periodically porting our software to the latest hardware.
In such a situation, there is a huge risk of becoming locked in a single vendor ecosystem, leaving us at the mercy of the vendor's financial and technical fortunes, thereby restricting our ability to tackle scientific challenges independent of the available hardware.
Enter the realm of alpaka, a revolutionary C++ library. This powerhouse provides a zero-overhead kernel abstraction, offering the flexibility to run applications on any contemporary High-Performance Computing (HPC) system. Whether you're dealing with cutting-edge RISC-V and ARM CPUs, GPUs, or legacy X86 CPUs, alpaka empowers you to harness the computational of the fastest systems worldwide seamlessly.

Speaker: Rene Widera (HZDR – Helmholtz-Zentrum Dresden-Rossendorf)

rse-meetup-alpaka_30-11-2023.pdf

15:25 [Pitch] Testing a Library that Supports Different Target Architectures with GitLab CI

Our C++ library alpaka allows to write a function once and execute it on different target architectures like x86 and ARM CPUs, Nvidia, AMD, and Intel GPUs, and FPGAs. In addition, the library supports Linux, Windows, and macOS. Therefore, we need to support a wide range of SDKs from different vendors on different systems. Automatic testing is mandatory for this and is a complex problem that cannot be solved by hand. The presentation introduces the core idea of our CI - the job generator.

Speaker: Simeon Ehrig

rse-meetup-alpaka-ci.pdf

15:30 [Pitch] Searching for a memory leak

I am currently developping a software to analyze laser pulses. It records an image from a spectrometer. This is in the frequency domain and to judge on the quality of the recorded data it needs to be transfered into the time domain. This is done by doing a fast fourier transformation of the image data using libfftw via libfftwpp.

Speaker: Joern Dreyer

RSE-Meetup_3011.pdf

15:35 → 16:00 Group Discussion

One group per pitch talk.

15:35 [Discussion] Group Dreyer

Speaker: Joern Dreyer

15:35 [Discussion] Group Ehrig

Speaker: Simeon Ehrig

15:35 [Discussion] Group Steinmeier

Speaker: Leon Steinmeier (Helmholtz Institute Freiberg)

15:35 [Discussion] Group Widera

Speaker: Rene Widera (HZDR – Helmholtz-Zentrum Dresden-Rossendorf)

16:00 → 16:15 Summary and Closing

Speaker: Thomas Foerster (HZDR)