Folding Forum

I don't see in the FAQs any discussion of why a project chooses to use a CPU or a GPU. Since the development of Core 17, I thought that the GPU could do everything a CPU could. Is it just the size of the project that determines which route is chosen, so as to accommodate users with different hardware, or is the science better done on one platform versus another?

The same answer applies to "Why did Stanford bother to create a client for Chrome?"

If somebody's computer doesn't have a gpu, should they be prohibited from contributing to F@H?

One of my computers folds with an internal gpu. It doesn't earn a lot of points, and folding with all but one of my cpus adds an appreciable amount of points to what that computer can produce.

I think you are trying to say that a project can run on either, which I already knew was supposed to be the case, but you don't provide any insight into the criteria for selecting one or the other. Thanks anyway.

A given project runs on a unique core which runs on specific hardware, not on "either" although I suppose the researcher could create two closely related projects if that makes sense. Are you asking what scientific criteria the Pande Group uses to decides which hardware best suits a particular project?

As a Donor, if your hardware gives you a choice, you can choose one or the other or both, based on PPD or power/heat criteria. V7 establishes a default configuration by enabling the GPU(s) and after reducing the CPU count sufficiently to service the GPUs, allocating the remainder to a CPU slot.

bruce wrote:A given project runs on a unique core which runs on specific hardware, not on "either" although I suppose the researcher could create two closely related projects if that makes sense. Are you asking what scientific criteria the Pande Group uses to decides which hardware best suits a particular project?

Thanks. I think that is sufficient, though if someone wants to chime in on what criteria they use, that would be interesting too. It just was not clear to me whether they chose just one type of hardware, or whether they issued the same type of work for both. Some projects do the latter, so that a user can use only a CPU if they want, while those that have a suitable GPU can do the same work, only faster.