I have looked over the GPGPU threads and it seems that at least one person is running both a CPU Boinc program, as well as the GPGPU client. But when I downloaded it, the Boinc version of my CPDN software changed to the version used by the GPU client. Is anyone fimiliar with running the GPGPU Boinc and CPDN? Presently, I am running four computers on various Boinc projects (CPDN and the Clean Energy Project at WCG), but I would love to use the extra GPU power to do some useful work. If I am going to run these computers 24/7, I might as well give it all the work it can handle. Does anyone know if they will conflict? How will the other version of Boinc effect the running of my other Boinc projects? Although I can choose to have no screen saver, does anyone know if the two different screen savers will conflict (as it did when I tried to add Folding@home to my GPUs)? Great thx to anyone who can help with these questions!!!
Cheers
Bill

Some of this is a bit muddled.

the Boinc version of my CPDN software changed to the version used by the GPU client

so I'll try and read between the lines.

The cpdn software shouldn't change.
I think that what you may mean, is that you've downloaded a WCG version of BOINC, (GPGPU?), which is slightly different to the Berkeley version. How different, and in what respects, I don't know.

However, it shouldn't matter, as cpdn doesn't have, and likely never will have, GPU support.
They only difference I know of that you may see, is that now your computer will keep asking for GPU work from cpdn, which will respond by saying that there's no work of that selected type. And these requests should get less frequent.

I have looked over the GPGPU threads and it seems that at least one person is running both a CPU Boinc program, as well as the GPGPU client. But when I downloaded it, the Boinc version of my CPDN software changed to the version used by the GPU client. Is anyone fimiliar with running the GPGPU Boinc and CPDN?

There are not 2 different clients for GPU and CPU processing. One client does both types of processing.

Presently, I am running four computers on various Boinc projects (CPDN and the Clean Energy Project at WCG), but I would love to use the extra GPU power to do some useful work.

The only projects that currently utilize CUDA enabled work units are GPU Grid and Seti@Home

If I am going to run these computers 24/7, I might as well give it all the work it can handle. Does anyone know if they will conflict? How will the other version of Boinc effect the running of my other Boinc projects? Although I can choose to have no screen saver, does anyone know if the two different screen savers will conflict (as it did when I tried to add Folding@home to my GPUs)?

The GPU work units do nothing to change the operation of the CPU work units. The screen saver still functions on the CPU work unit. On my single core CPU and CUDA enabled graphics card 2 work units run, one CUDA and the other CPU based.

---

Some care needs to be taken over nomenclature:

1. Most users would agree that the part of BOINC running at CPDN or SETI is the 'server' part of BOINC, not ultimately because it provides services (lots of things do that) but because it is somewhere else.

2. The two parts of BOINC that run on the user's computer are the generic BOINC 'client' software and the specific 'science application'. The BOINC 'client' uses the services provided by the distant part of BOINC, so it's a client. However, the science application uses the services provided by BOINC client, so it's a client too - and that's where the confusion arises.

There are lots of pairs of words like 'client/server' that technical people will insist are natural and obvious but which are in fact conventional and meaningless. Verification/validation, offer/bid: sit in a room with an accountant for a few hours and seemingly familiar concepts will become utterly baffling (creditor/debtor). It's professional jargon and should be kept in the profession.

I think of the parts as Project-BOINC and User-BOINC. So long as you always make it clear that you mean the science application when that's what you mean, then people will understand.

Presently, I am running four computers on various Boinc projects (CPDN and the Clean Energy Project at WCG), but I would love to use the extra GPU power to do some useful work.

You can -- but there can be issues with various iterations of the BOINC 6.x client when doing this. I have found that the 6.4.5 client version is the least problematic in this regard. The 6.6.36 and up versions seem to really have problems dealing with the mix of CPU and GPU projects -- in part because they seem to focus on GPU projects (perhaps that reflects the influence SETI has on client development).

The only projects that currently utilize CUDA enabled work units are GPU Grid and Seti@Home

There is another project that does CUDA work --
AQUA http://aqua.dwavesys.com/

Further, if you have the right ATI GPU (38xx, 47xx or 48xx) you can utilize it with the Milkyway project using a 3rd party developed optimized client.

Milkyway http://milkyway.cs.rpi.edu/milkyway/

The Milkyway folks are just about to provide support for the CUDA GPU as well.

If I am going to run these computers 24/7, I might as well give it all the work it can handle. Does anyone know if they will conflict? How will the other version of Boinc effect the running of my other Boinc projects? Although I can choose to have no screen saver, does anyone know if the two different screen savers will conflict (as it did when I tried to add Folding@home to my GPUs)?

The GPU work units do nothing to change the operation of the CPU work units. The screen saver still functions on the CPU work unit. On my single core CPU and CUDA enabled graphics card 2 work units run, one CUDA and the other CPU based.[/quote]

---

Some care needs to be taken over nomenclature:

1. Most users would agree that the part of BOINC running at CPDN or SETI is the 'server' part of BOINC, not ultimately because it provides services (lots of things do that) but because it is somewhere else.

2. The two parts of BOINC that run on the user's computer are the generic BOINC 'client' software and the specific 'science application'. The BOINC 'client' uses the services provided by the distant part of BOINC, so it's a client. However, the science application uses the services provided by BOINC client, so it's a client too - and that's where the confusion arises.

There are lots of pairs of words like 'client/server' that technical people will insist are natural and obvious but which are in fact conventional and meaningless. Verification/validation, offer/bid: sit in a room with an accountant for a few hours and seemingly familiar concepts will become utterly baffling (creditor/debtor). It's professional jargon and should be kept in the profession.

I think of the parts as Project-BOINC and User-BOINC. So long as you always make it clear that you mean the science application when that's what you mean, then people will understand.

Dear Ian:

Just to get a better handle on the nomenclature, is the generic BOINC “client” software the part of the program that is housed in the “Program Files” Folder and the “science application” the files that are contained in the “Programdata” Folder? This would be in Vista and Windows 7.

---

EVERYTHING on a project's servers is the BOINC server side program(s).
EVERYTHING on a person's personal computer that is BOINC is the client.

The split on the PC is between the programs, (which DON'T need to write to the hard disk, and can therefore be placed in the area of Windows that doesn't upset it), and those parts of BOINC that DO need to be written to, and must therefore be placed outside of the Windows "protected zone".
These include the master file for each project, the client_state.xml files, (which get written to every few seconds!), the std... files, etc.
And then there are the many application files, some of which get written to, and some of which don't. The latter are the original zips that get downloaded for each model. But, regardless of whether or not the app files are written to, ALL of them are placed in the "outside of the protected zone" location.

But, just to confuse things, BOINC, on the PC is divided into two parts:
1) A hidden part called a "daemon" (Unix term), which is also called the "worker", or "client". (There's that word again, with a different meaning. :) )

2)The manager, who's only job is the look at what the "worker" part is doing, and display this information for the human. It's not needed for the hidden part to get on with things, and when BOINC is used in work or educational environments, it's best NOT to have the manager running, so as not to alarm the people who use the computer. Or to let them fiddle with things. :)

So, "client" has two levels of meaning.
Possibly better for most people to think of all of the BOINC parts on their PC as "the PC stuff", and the BOINC parts on the project servers as "the project stuff".

[JIM wrote:]Just to get a better handle on the nomenclature, is the generic BOINC “client” software the part of the program that is housed in the “Program Files” Folder and the “science application” the files that are contained in the “Programdata” Folder? This would be in Vista and Windows 7.

I've only processed work units from CPDN and CPDN Beta, so my cross-project experience is very limited: however, the 'Program Files' folders on my Vista machines contain only generic 'BOINC client' files, whereas the separate 'data' folder contains a mix of BOINC client data files and science application data and executable files - split in the way and for the reasons that Les has described.

Folding @ Home also runs CUDA programs using NVIDIA Cards, and a similar program using ATI cards.

http://folding.stanford.edu/English/DownloadWinOther

Note you will need to have the drivers installed first!

For NVIDIA Cards, make sure you have an up to date driver and Cuda program installed.

I have not used an ATI card, but it works just as well.

Again, the other programs are GPUGRID, Aqua, and the pointless SETI.
The Milkyway also do an ATI project and may one day finally even manage to get an NVidia effort up and running. Presently just begining the Alpha.

---