In this report from the recent PRACEdays15 meeting in Dublin, Robert Roe from Scientific Computing World looks at how Europe and Japan encourage the use of HPC.
Although Europe is a loose federation of independent countries and Japan is a single unitary state, both have come to the same conclusion about how best to improve access to high-performance computing (HPC) – HPC resources must be shared and not monopolized by the individual owners of the computer systems themselves. And both have set up bodies to coordinate that process of sharing which bear remarkable resemblances to each other.
However, in one significant area the two regions differ strongly in their policy: in Japan, compute cycles on the country’s foremost HPC systems are offered to industry to conduct commercially sensitive work, without the industrial partners having to openly publish the results of the project, as is the case in the pan-European access system.
At the PRACEdays15 meeting in Dublin at the end of May, delegates heard both Sanzio Bassini, chair of the Partnership for Advanced Computing in Europe (PRACE) council and Masahiro Seki, president of the Research Organization for Information Science and Technology (RIST) in Japan. They each discussed their respective organizations strategies for developing HPC infrastructure, and encouraging the use of HPC systems by industry and academic users.
Achievements of PRACE phase one
Bassini stated that to “maintain Europe as a world class contributor in science” governments and organizations like PRACE “must ensure that they can offer access to leading HPC systems.” He assured delegates that: “PRACE aims to offer at least one system in each architectural class.”
As 2015 marks the end of the first phase of PRACE (PRACE1.0), Bassini was keen to stress the progress made by the organization over this time. Prace now represents 25 countries across Europe and has enabled 394 projects since its inception, ranging from small projects with SMEs to huge simulations modeling the cosmos or helping to advance elementary particle physics. Some 10.2 thousand million core hours had been awarded to researchers across Europe and more than 3,000 people trained through six advanced training centers across Europe.
Japan supports HPC for industry
Much like PRACE, the goals of RIST are to improve and refine computational science and technology and to encourage the use of HPC by both academic and industrial organizations.
In certain circumstances, industrial users do not need to publish the entirety of their results, instead they can choose to publish research outcomes, keeping their own IP safe from public view. Some saw this as a key differentiator between Prace and RIST: both offer computational cycles to industry; but RIST proposes a much better deal to those concerned about protecting sensitive information – a key selling point for many industrial users.
A second point of difference is the support for industrial and academic users that has been put in place by RIST. Seki said: ‘RIST provides 17 scientific consultants, with 7 consultants in Kobe, 6 in Tokyo. The Tokyo office is setup for the support of industrial users, with two rooms that are setup to provide confidentiality to potential users.’
Again RIST is catering for its industrial users on the understanding that they have different requirements in some respects from their academic counterparts and so they expect a different kind of service compared to academic users.
By providing consultants RIST is providing some expertise to groups of users that may understand their chosen field very well but may not have the skill s and knowledge required to operate HPC systems. Because RIST provides knowledge in areas such as code optimization, industrial users can concentrate on the specific challenges that face them without having to worry about the HPC systems themselves.
Seki said: ‘The computational resource is 14 petaflops in total, 10 petaflops from K and 4 petaflops from the other machines. The second layer consists of 11 supercomputers which are located in nine major universities and two national laboratories, which are playing unique rolls in computational finance and technology.” Seki concluded: “The computers from the second layer are not fully devoted to HPC. That means that generally 30 per cent of their computational capacity are provided to the infrastructure and the rest of their capabilities is used at the discretion of their respective owners.”
Who pays for industrial HPC?
During questions from the floor, Augusto Burgueño Arjona, Head of the e-Infrastructure Unit at the European Commission, asked whether the speakers thought that HPC services for industry should be charged in order to generate income which could be fed back into the infrastructure. Burgueño Arjona said: “I asked this question again this morning about what it means to offer HPC services for free, when providers of HPC could be charging. These are issues that will have to be addressed in the future.”
Bassini was asked about the progress of PRACE moving from the end of PRACE 1.0 into the start of PRACE 2.0. He remarked that “one of the most pressing issues is the participation on all members of the association to the operational costs of the infrastructure, and the sustainability of the model moving forwards.”
Bassini raised several options, such as basing contributions to the PRACE infrastructure from each country’s GDP but another option would be to subsidize these HPC resources by selling a certain portion of the computational cycles to industry. Alternatively these strategies or some combination of ideas could be employed in tandem reducing the burden on the participating countries that provide the computational muscle to PRACE.
This is the fifth in a series of reports from PraceDays15, held in Dublin at the end of May. Tom Wilkie’s article Will a European company build Europe’s first Exascale computer? analyses European attempts to develop hardware and technology for the next generation of supercomputers. Robert’s Roe’s report on the role of HPC in the host country, Ireland, can be found here. On a similar theme, Tom Wilkie writes about how supercomputing for small companies can be made simple. The reasons why parallel programs need new maths are explained by Tom Wilkie in the first report from the conference.