By: Martin Meuer, ISC Group

My father, Hans Meuer, was a huge proponent of all kinds of ranking. With great pleasure he would recite the list of names of the best chess players along with their own ”Linpack rating“, the so-called Elo numbers, and enjoy his opponent’s astonishment afterwards. He definitely could impress those who understood that the Elo numbers, unlike the TOP500 ranking, were updated monthly.

Another memorable ritual of his was to keep an exact record of all well-wishes he received on his birthday. Calling him at 9:00 am on his special day would commonly provoke comments like: “You’re the seventh to congratulate me. Number one was (as always) your mother, number two was…” and so on – always accompanied with an invisible wink. And as was often the case, I received an updated list per email later that day.

It is not surprising to those who knew my father that he was anxious to top all sort of rankings himself. Sometimes he used tricks to achieve that. When it came to his beloved lists of congratulants, he was always a candidate for the top spot by using services that send digital birthday cards immediately after midnight. Of course, he didn’t accept the criticisms that his practice was, strictly speaking, digital cheating. 

It filled him with great pride that he was the top non-US attendee of the US-based Supercomputing Conference (SC). Up until falling seriously ill last year, he never missed a single SC conference and was therefore the only person to have attended all SC and ISC conferences, a fact that he noted with some satisfaction.

Against this backdrop it’s not surprising that in 1986 he, together with Erich Strohmaier at Mannheim University, started publishing the annual “Mannheim Supercomputing Statistics,” a list of vector computer systems in the US, Japan and Europe. When Jack Dongarra joined the project in 1993 with his Linpack benchmark, the TOP500 list of the most powerful supercomputers was born. The list has been published twice a year since then. In 2000, the list of authors was completed by adding Horst Simon.

After my father passed away, following a short but intense battle with cancer, Erich, Jack and Horst gave me the opportunity to take over my father’s role as a TOP500 author.

As the managing director of Prometeus, the company running and promoting the TOP500 website, the continuation of the project is close to my heart. On top of that I would like to do all I can to make my father’s heritage live on. So I gladly accepted the offer.

Returning to the subject of rankings: there is also one with regard to the TOP500 authors, which is relevant to the order of signatures on the certificates. As this ranking is based on time spent within the project, my father naturally had the pole position, whereas I, the rookie, will of course have to go to the end of the line.

I’m aware that TOP500 has elicited controversial discussions within the HPC community. On the one hand there is the invaluable data collected in 44 lists, which is a great instrument to predict market trends. On the other hand it’s being criticized – and rightly so – that Linpack only measures part of a computer’s performance, which often doesn’t reflect the real requirements of applications.

Of course Linpack should never be misused as the only criteria when choosing a HPC system. The fact that this happens anyway, especially in political decisions, is often held against the TOP500 authors unjustly, however, as we only provide the data and are not responsible for their (mis)interpretation.

To carry the value of the historical data forward, I think it makes sense to keep Linpack as the prevailing benchmark for the time being. That said, we should still be open towards other more “realistic” benchmarks, such as HPCG or HPGMG. How well these alternatives will be accepted in the community, however, remains to be seen. I don’t have to remind you that the NAS Parallel Benchmark, once thought of as a Linpack contender, failed in the end due to the lack of computers that it was run on.

The next TOP500 publication date is already decided: the 45th list will be published on July 13th, 2015 during the opening session of ISC High Performance, in Frankfurt, Germany. The event, by the way, is the world’s oldest HPC conference and turns 30 next year.