This spring the meeting of the ZKI working group on supercomputing took place at the University of Hamburg, on March 25th and 26th ( https://www.conferences.uni-hamburg.de/event/585/ ).
Reminder: the ZKI („Zentren für Kommunikation und Informationsverarbeitung in Lehre und Forschung e.V.“, https://www.zki.de ) is an an association of computer centres at universities. They have thirteen working groups, one of it is on supercomputing ( https://www.zki.de/ueber-den-zki/arbeitskreise/arbeitskreis-supercomputing/ ). It meets semi-anually at varying locations.
In general, I find these meetings very valuable: imho, they are a perfect mix of admin and user reports, vendor presentations and usually one talk about an application at the corresponding location.
After the registration the event started as usual with a short introduction by the organizers Olaf Schneider and Karsten Kramer (speakers of the working group) at 1pm.
After that the local organizer, Hinnerk Stüben, gave (again as usual) a talk on the local situation regarding supercomputing, For this Hinnerk (btw: that is the Northern German version of my first name 😉 started at the very beginning, namely the history of the city of Hamburg as a member of the Hanse ( https://en.wikipedia.org/wiki/Hanseatic_League ). This was (from the 12th to the 17th century) an „association“ of Northern-European cities which had common interests, mainly in the area of trade. An important remark by Hinnerk in this context: cooperation is in mutual interest of participating parties, current politics does not seem to realize this in many countries, nowadays. Hinnerk/Heinrich/Henry also mentioned several times that Hamburg is considered a/the gate to the world, given its big harbour. He then gave an overview of the HPC infrastructure at the University of Hamburg: it is not huge and quite distributed, that is at least the impression I got.
The next presentation was an unusual one as the meetings are usually very technical: Peter Pawliki of Electronics Watch talked about „Social Sustainability in Procurements“. EW is an NGO which takes care of human rights in the electronics industry, hence also in IT and HPC. (In this respect, it is comparable to Food Watch.) PP‘s presentation was pretty shocking, especially in the beginning: he mentioned various violations of human rights (not only in Asia as one might expect), e.g. examples of forced labor. (In this context, the term „forced labor“ is used in a pretty wide sense, e.g. if a worker refuses to work overtime and he is told that he will never will work overtime this is considered a case of forced labor because he then cannot feed his family.) EW tries to improve the situation of workers in the electronics industry by performimg audits etc. It gets its money mainly from member fees: institutions that do procurements, e.g. for HPC equipment. Here I want to say a big „Thank You“ to Olaf Schneider who invited PP after searching the internet. I learned this in the first break when I talked to PP, he also told me that about eight million people are working in the electronics industry worldwide.
The next presentation titled „openiris.io – a non-profit platform for scientific resource management“ was given in English (an exception) by Robyn Brackin-Helmers of the OpenIRIS Association (she is working at the Charité, the famous big hospital in Berlin). OI is a very comprehensive management software (features included are scheduling, service requests, project management, resource restrictions, iissue tracking, pricing rules, advanced statistics, custom portals, and calendar integration) that can be used in HPC environments, too. As I cannot go into detail here: please have a look at the link given above.
During the first long break I talked to PP of EW (see above) and a guy from DESY. These breaks (called „communications pauses“) are one of the highlights of these meetings as they give the opportunity to communicate with speakers, colleagues etc.
The following session began with a presentation titled „PERSEUS – Eine Open-Source-Software für Projektmanagement und Prozessoptimierung in wissenschaftlichen HPC-Zentren“ given by Lucas Ostermann from GWDG. As the title implies there is some overlap with the OI presentation just before the break. This brings me to the question I often ask: is it really necessary to develop yet another software for this purpose as there are already solutions (open-source or commercial) that solve the same problem. The answer usually is: well, we have such specific needs…
The next talk was titled „HPC-Portal: Eine maßgeschneiderte Web-Anwendung zum User- und Projektmanagement am NHR@FAU“ and given by Christoph Kluge. Again, there was some overlap with the precious presentations…
The first day ended with the talk on „Verwaltung von Speicherebenen via semantischen Datenzugriff mit Hilfe eines Datenkatalogs“ by Hendrick Nolte which was different to the previous ones although I did understand very little. 😉
Well, this was not the real end of the first day: the participants came together for dinner at the „Blockbräu“, a large micro-brewery at the „Landungsbrücken“, the harbour. This locations was somewhat strange: on one hand they serve local specialities like fish (I had „Pannfisch“ with spinach), on the other hand they advertised their version of the Octoberfest. What I did not like was the live music: a lady sang old Northern German classics like „La Paloma“, „Wo die Nordseewellen trecken an den Strand“ etc. I sat together with Prof. Ziegler of RUB (formerly Düsseldorf, Jena and speaker of the working group), a lady from DKRZ and a gentleman from the University of Oldenburg where I gave a presentation on the EuroCC project, one year and a half ago.
The second day started as usual with the closed session where only the „real“ members of the working group are allowed to attend. Explanation: the WG and its meetings is also open for „guests“ like vendor representatives (or retirees like me) but not the closed session where the representatives of the computing centres talk about e.g. their experience with vendors. Therefore, vendor representatives are not allowed in the closed session. Btw: when I was employed at SICOS (which is strongly associated with HLRS) I was allowed to attend the closed sessions.
The talks of the second day started with „EE-HPC Projekt: Automatic energy optimization for HPC clusters within ClusterCockpit“ by Jan Eitzinger from Erlangen. I know Jan as the „inventor“ of Likwid ( https://github.com/RRZE-HPC/likwid ), a low-level tool for performance monitoring and benchmarking. Btw: when he developed Likwid he had a different lastname (which I do not remember), so he must have marrried in the meantime. As the second presenter in this session did not show up Jan used a full hour for his presentation.
During the following long break I talked (besides others) to my old friend Werner Hartz (formerly education sales rep at Dell, now with RedNet, not RedHat ;-). He confirmed that Dell is currently „farting out“ in HPC.
The last session started by a short „NHR Update“ by Dörte Sternel. Besides other points she mentioned/advertised the next NHR meeting and conference in Göttingen, in September.
The last presentation was given by Stephan Rosswog from the local Astrophysics Institute: „Multi-Messenger signals from colliding neutron stars“. This is another relatively new tradition at the meetings: a paper from a local institute that uses the local HPC resources.
Also as usual the event ended with tours through the local computing centres of the university and of DKRZ. And as also usual I skipped these tours. 😉
To summarize: again this was very interesting event, perfectly organized by the speakers of the working group (Olaf Schneider from Jena, Karsten Kramer from PIK, and Holger Marten from Kiel [not present because of family issues]) and Hinnerk Stüben.
PS: Slides of the presentations can be found here: https://www.conferences.uni-hamburg.de/event/585/contributions/ (as soon as they become available).
-$tr250331
introducing: HPCsquAIre(*) 