CONFIDENTIAL: LONDON PARALLEL APPLICATIONS CENTRE


EUROPA WORKING GROUP ON PARALLEL C++


PLENARY MEETING


Minutes

Date: Wednesday, 8 November, 1995

Venue: European Commission, Brussels


Present

The members of the Working Group present were as follow:

Apologies

Apologies were received from Dennis Gannon of the University of Indiana.

1.0 Introduction - Michael Arentoft, DGIII

The meeting was opened at 10.15 am by MA who introduced the meeting with ML, also of CEC.

MA summarised progress to date and flagged some issues to be discussed during the course of the meeting. MA encouraged links with manufacturers and with the HPC++ group in the States, he also urged expansion of the membership of the Working Group.

2.0 Presentation of the Draft Standard - Dr Peter Dzwig/Prof Heather Liddell, LPAC

Refer to meeting documentation.

In his introduction to the Standard and the Working Group, PED addressed the points detailed in MA's introduction, saying that the work that would be presented in the course of the day would address most of the issues raised.

3.0 MOPs and proxies basic components of EC++ - Denis Caromel, INRIA

Refer to meeting documentation.

4.0 Libraries and Hardware Models - Richard Kaufmann, DEC Ltd

Refer to meeting documentation.

5.0 UC++ - An Active Model - Russel Winder, UCL

Refer to meeting documentation.

6.0 ||C++s and How They Relate to EC++ - Klaus Wolf, SOFTPAR

Refer to meeting documentation.

LUNCH

7.0 ||C++s and How They Relate to EC++ - Philippe Mussi, INRIA

Refer to meeting documentation.

8.0 Parallel C++ for a SIMD machine - Peter Flanders, CPP

Refer to meeting documentation.

9.0 Next Stages in the Evolution of EC++ - Dr Peter Dzwig/Prof Heather Liddell, LPAC

PED stated that there was a need to rework the Roadmap and also the project plan and MA agreed with this. PED went on to say that, following the meeting, the EUROPA Working Group would need to refine the EC++ Draft Standard as there were a number of "holes" in it and these needed to be worked on in order that people would have something at which to target their ||C++ implementation.

DC requested that the revision be made, if possible, before he went to the HPC++ meeting at Supercomputing '95 in the States. AM was not conviced that this was possible, but felt that minor amendments were feasible in the timescale mentioned. AM added that he felt it was inappropriate to make large scale changes before they had been discussed at an Architecture SIG meeting.

PED went on to say that there was a need to get the Implementation and Applications SIGs up and running and also that it was important to get a project set up (possibly through Framework IV) to get this carried forward. PED added that, in order to move on with the Implementations and Applications SIGs, it would be necessary to expand the EUROPA WG membership.

The EUROPA WG also intended to build links with HPC++, and it was expected that this would be a fruitful relationship as link-up looked feasible on the basis of what was already known about HPC++.

PED outlined the need to set up a timescale for future meetings which had not, as yet, been formally planned and added that there was a meeting at the University of Vienna in January, 1996 which the group had been invited to co-host. PED felt that this should be given the status of a full EUROPA meeting. PED reported that there had been suggestions that we should hold a workshop at HPCN and with this in mind, stated that papers were still being accepted. PED felt it important that a further Architecture SIG meeting should be held prior to the Vienna meeting.

Both the Implementations and Applications SIGs needed to be defined and a start-up meeting should be held in the near future. RK stated that there was a need to find other hardware vendors interested in the work being done as interest would wane if DEC were the only representative of this side of the industry.

DC stated that he had been working on implementations in Nice and had found that it was "greedy" for manpower. He added his weight to the assertion that a project was needed to fund this area and thus press forward.

MA responded to this by stating that a call would be opening on 15 December, 1995 (submission deadline - 23 January, 1996) and added that it was necessary that industrial applications should be targeted. In addition, MA stated that there would also be a call for accompanying actions going out in March. PED suggested that this might be more appropriate given the level of funding which the partners might be seeking and requested that all partners should send emails with proposals to PED as soon as possible. PED observed that the sort of scale associated with big projects was not necessarily required here.

MA summed up by identifying what he felt to be the five major issues: Portability, applicability, implementability, (US) compatibility, promotion and expansion.

10.0 Open Forum

DC mentioned that there was an international workshop on ||C++ in Japan on 11/12 March, 1996.

Following discussion on the subject, PED stated that there was a need to address the SIMD point further.


LPAC Homepage
EUROPA Index
Matthew Holford <M.Holford@lpac.ac.uk>

15 May, 1996