Folding Forum

whocrazy wrote:I have another question, can the folding @home cuda program be made to work on all projects, like the normal folding@home application program can do?

I'm not sure what you mean by this. Are you referring to the GPU client vs. the Classic CPU client? Are you referring to the Nvidia-specific viewer vs. the viewer included with the Classic systray client? If your concern is about the viewer, then don't bother with it. It's recommended to not use any of the viewers because A) certain versions are buggy and can cause the client to crash and B) the viewers take up computing time and slow down the WU processing.

bruce wrote:

Baowoulf wrote:What if turning on that option gave less or no points and it was more for people who are like you may have a reason for a specific project. Like if someone had a family member who died or was afflicted with a diease that the project looked into? I'd be willing to do that type of work and for people who can't afford to have big folding farms it might let them feel like they are contributing more then say a big time folder could because it's 0 points but the project has a more centralized goal in mind.

That's an interesting suggestion and as far as I know, you're the first one to suggest such an option. I'll pass it on.

The Pande Group, of course, makes their own decisions about what features to add or not to add, but I do know that their first criteria is whether it will result it more science being done compared to what would be done without the feature. This one focuses more on Donor satisfaction than scientific production, but that's part of their decision process, too. Happy donors do recruit their friends.

So if we modify this idea to include a quota system, it might solve the issue of other Projects:

Lets say you allocated a quota of 30 WUs from a particular disease (it can include a single Project or many Projects), once you reach that limit, you will then be allocated 15 random WUs (if PG notices a particular Project lacking resources, they can issue WUs from those Projects then) after which you can again have your 30 WUs quota. This can continue in a cycle. If you are found to be dumping the 15 random WUs, your quota will be suspended or you can simply be banned from the quota status and can only fold random WUs like the present.

whocrazy wrote:I have another question, can the folding @home cuda program be made to work on all projects, like the normal folding@home application program can do?

I'm not sure what you mean either, but I'll try to answer anyway.

The Classic client works only on uniprocessor assignments. Some additions were made to that client to support SMP hardware and new Projects were created for what was called a beta SMP client. Some different additions were made to the classic client to support GPU hardware and new Projects were created for what called the beta GPU client.

Can all of those features be integrated into a single client? Probably, but we don't know yet. The upcoming Version 7 client has been called a unified client, which seems to imply that there will be only one type of client that can support any type of hardware. We'll have to wait for more information on V7 before we know if that has been accomplished, but it's certainly an identified goal.

There's also a question of Operating Systems. A client for Windows is not the same as a client for Linux or as a client for MacOS or for the PS3. I don't see any possibility of integrating those into a single version. The PS3 is unique, but hopefully the x86/x64 features (Windows/Linux/MacOS) will be equivalent.

On the other side of the same question, you might be asking about about specific Projects. Work intended for a uniprocessor configuration is different than work designed for a SMP configuration or for a GPU configuration so I don't expect that a GPU will work on all projects. Moreover, CUDA is "owned" by NVidia, so using it restricts work to nVidia GPUs.

With respect to GPUs in general, work is currently divided into three types of Projects, those for ATI, those for NVidia g80 series hardware and those designed for NVidia Fermi series hardware. A lot of work has been going into the future support of OpenCL. The goal of OpenCL is to allow one type of software to work with any GPU or with many types of CPU hardware. How fully that dream will be realized depends on efforts by many different companies/teams, including the developers of FAH. We'll have to wait and see how that develops.

I kind of like the idea of folding for a specific diseases or biomedical problems that F@H is currently studying; however, I suspect that most of the results returned from the Work Units we are processing can be applied to more than one specific disease or problem. Of course I could be totally wrong in my thoughts, but most of the project descriptions that I have read make no mention of a specific disease.

While it is nice to fold for a specific cause (grandmother had Alzheimer's Disease), I am not sure that is the most effective way to go as far as F@H goes. I prefer to let the scientist make the calls.