ISC-HPC impression

3 minute read


ISC-HPC, “The Event for High Performance Computing, Machine Learning, Data Analytics & Quantum Computing” is a yearly conference focused on HPC but also more and more on the adjacent topics above.

Main conference

The amount of quantum focus in this years ISC was astonishing. A non-exhaustive list of events were the full-day QML tutorial on Sunday, full day workshop on Thursday, several posters, sessions on the EuroHPC efforts, the HPCQS and similar projects, the ETP4HPC quantum project, and a session on industry usecases. The level was all-throughout very high.

Although mostly very positive impression I do have one critique. I’m not sure the European projects are all structured correctly. The efforts appear quite siloed off and I’m not sure the HPC community is providing a useful platform. Rather, there are projects there which end up in the HPC world which would perhaps be better off as pure quantum focused projects rather than having the forced HPC mix. This would also perhaps let them be tackled more by quantum computing experts. In quantum we have much to learn from HPC, and the HPC world could be a very valuable link between the pure quantum world and our user-base. But I’m not sure the projects in their current form will maximize the utility.

One thing that I feel is sorely missing from the discussion there is all the work that goes into trying to get quantum algorithms run on near-term (say 5-10 years) hardware. That is sort of a cross functional area that overlaps with every part of the process and the European HPC projects appear to me to not dedicate any resources to this. This I fear is a mistake.

I’m also concerned that a few loud and interested voices in the HPC community are allowed to dictate much of the discourse and projects. They don’t always have the broadest experience or interest in the quantum space and depending on how quickly we achieve FTQC this may or may not be something regrettable.

The HPC players also seem very set on quantum abiding by strict standards fairly early on. I’m not sure this is the best way forward. From the computing history and from HPC I think it has shown that it’s best to let things mature more naturally before establishing the standards (see e.g. programming languages or MPI/OpenMP). I think this eagerness for standards is closely linked with the above point of the European funded projects that might be a bit siloed and not always tackling all the questions.

Otherwise is was a great conference with lots of networking opportunities and interesting discussions.

Quantum workshop

This years addition of the QQCC workshop at ISC was far more quantum heavy compared to the year before. In fact, there were mostly no evidence at all that this was part of an HPC event rather than a pure quantum event and several of the talks presented interesting algorithmic work.

In terms of intellectual curiosity this was by far the best part of my week in Hamburg.