Workshops & Reports

Most of the workshops I participate in are just “out-of-the-box” events organised under the umbrella of conferences. Every now and then, I however maintain a dedicated workshop page hosting more info than the standard speaker-abstract tuple. You find this info here.

Upcoming Workshops

We will host an ExaHyPE 2 user workshop in October 2020.

SIAM CSE 21 is the next large conference I plan to join.

As soon as things are (more or less) back to a (virtual) normal, I’ll also continue to organise the Durham HPC meeting. For the time being, you might want to browse through the item in the menu for a selection of past workshops (workshop archive) for which we had set up dedicated webpages.

Reports

Besides the workshops, I write (again only occasionally) reports on workshop outcomes. I try to inject them into “mainstream”ish media, but at least I make them available here. These reports are not peer-reviewed in a classic sense or went into a journal. They are pieces of my own opinion and express my view of the world.

ExaHyPE’s OpenMP GPGPU Port—Lessons Learned

In Summer 2020, the ExaClaw project team joined the Sheffield GPU Hackathon. We ported the ExaHyPE engine onto GPUs, but this was by no means a smooth ride. Therefore, I decided to summarise some of our key observations in a brief report.

A high-level characterisation and generalisation of communication-avoiding programming techniques

In 2019, I was invited by the Societa Italiana di Fisica to the International School of Physics “Enrico Fermi” Big Data Workshop. I presented some of my work orbiting around communication-avoiding techniques (though not in a classic LA sense) to a bunch of PhD students. Later on, I summarised these thoughts in the above report. It was supposed to become part of a longer report collection, but I think that one never materialised unfortunately. Anyway, the report remains a good summary of quite a lot of work (and its context) that I’ve done over the past decade.

The power of Lo(o)sing Control – When does a re-implementation of mature simulation fragments with HPC DSLs pay off?

In 2018, the ExaHyPE project was slowly coming to an end, and I found that I have to write a new (fourth) generation of my Peano software soon if I want it to remain useful for the exascale era. I started to wonder whether others have to reimplement their core workhorses from time to time, too, and what their impressions and lessons learned are. So I assembled a small, nice cohort of speakers at ISC 2018, organised this workshop, and summarised my insights. It later got featured by HPC Wire.

Form Follows Function – Do algorithms and applications challenge or drag behind the hardware evolution?

In 2016, I have organised a workshop at ISC that did orbit around aspects of what people call co-design. There was an interesting cohort of speakers and interesting opinions, too, so I decided to summarise my key observations and take-aways in a report. Later on, this summary got featured in HPC Wire and on some other pages. The workshop pages are archived here, but I think the summary still is relevant and valid.