The best place to post that suggestion would be here I think.

I was not trying to be funny. GPUGrid is a wrapper host just like WCG, Rechenkraft and others. They host multiple sciences for third parties. I am not sure what such a parterning would accomplish. If one wants to crunch GPUGrid, just do it. Not sure GPUGrid would like to convert to "dog-year points". (Now I am trying to be funny. )

Conversion to/creation of GPU applications are completed by the Project Scientific teams, not the administrative WCG layer. GPUGRID is no different, they are run by the University Biomedical Informatics Research Unit of the University of Barcelona.

Each Project Team receives its funding and support from its own Institutional Sponsor, typicaly a University or Institutional Research facilty. The sponsors of such faculties or Institutions have little inclination to contribute funding to other Project application development. In any case the GPU apps, if done at all as some science Projects are not feasible in a GPU app, are nearly always so specialised in conversion of their specialised modelling there is little if any similarity at code level where collaboration could occur.

Where such detailed code rewriting is not needed, the BOINC wrapper exists where Projects can place their application without too much code revision - at least in theory rofl ..... many a project team modelling specialist would squirm at the latter statement! Climateprediction Project is a typical example - far far from unique - in using a BOINC wrapper around the models provided by the UK weather bureau which are in fact written in Cobol and impossible (in econmic terms) to convert to GPU apps and still have a tangible gain at the end of that potentialy protracted 2-3 years conversion process, even if someone agreed to fund such a conversion, its just far too uneconomic.

Maybe when coding standards such as OpenCl yaddie yadda become more mature than the current crude implementations - and thats many many years away yet, if ever - then coding teams will be more encouraged to use them. However even the latter kind of development will never get around the often unique and highly specialised kind of coding needed for scientific Projects that need far better and more sophisticated methods than the likes of OpenCL etc etc could ever dream of providing. For example the ClimatePrediction Models have decades of Cobol development invested in them, the UK Weather Bureau is not going to throw that away any time soon just for BOINC

Regards
Zy

Hello Hypernova, hello everyone else,
I think, that a kind of partnering is possible, now.

It is just as simple as it can be:

If a member stops crunching after a short period of time, there is high probability that the reason is that he or she has found out, that he uses the computer to little time a day or the computer is too old.

It is not difficult to send an e-mail to those members who have stopped contributing.

The e-mail could be:
Dear member, for some reason you stopped crunching.

This is your free decision, which we respect.

If you stopped crunching because of the fact, that your

PC did not manage to calculate the Workunits in time, please note,

that there are other Grid Computing projects in the net, helping to heal

deseases, too.

Some of them are less demanding as far as system resourches are concerned.

One example is docking at home.

World Community Grid does bear any responsiblility for other projects.

We just hand on information.
If you decide returning to Worldcommunity Grid, we will be very pleased."

Greetings to erveryone, crunch on !
Martin Schnellinger

No problem Dataman. We are friends contributing as best as we can for a common cause.


Thanks Zydor for your detailed explanations. It seems that the issue is really not so simple.

I have decided to use two of my existing Asus Geforce boards, the GTX 285 and GTX 280 to run for GPUGrid. I test on two machines with a common use between WCG and GPUGrid. In fact it works very well. The CPU runs 12 threads on WCG (CEP2) and in parallel the GPU runs "one thread that is one WU" on GPUGrid. I am astonished how well and without a hitch, this works. The GPUgrid WU runs about 10 hours on the GTX285.

In this way inside my machine I have reunited the best of both worlds.