The ability to set the Maximum number of work units a gpu can process at one time, mainly If in a mixed gpu system, Like the following example:

GTX460(2 per gpu, 3 or more if ram allows)
GTX295(1 per gpu)

Set for 400 and Newer(Fermi) with a drop down menu selection, 200 series cards and older set to 1 maximum per gpu.

I have one GTX460 on the way for the latest graphics and I have 5 GTX295 cards, All in one water cooled Evga P55 Classified 200 PC and run with 2 psus(a Silverstone ST1500 1500w psu and a Thermaltake 650w Drive bay mounted video card psu).

---

Play MULE, 42 is irrelevant

I used the GTX460 only as an example, As this would also apply to any 460, 470, 480, 560, 570, 580, 590, etc..., If Boinc were hardwired to not allow older gpus like the GTX/GTS-200, 9000 and 8000 series gpus to do no more than one wu per gpu(Work Unit), then the existing advanced controls would work just fine and then only for the Anonymous Platform.

---

Play MULE, 42 is irrelevant

I asked the developers a while ago about this, their answer was "Why?".
I can see what they're going at, as BOINC is made to run one task per piece of hardware, so one task per core of the CPU, one task per GPU per videocard.

You can run more than one task per CPU, be it with the debug option built into BOINC, or with an anonymous platform file; that's then your own choice.

So for the GPU as well, the function's built in that we'll run work on it. You want to run more than one task on it, then use the options to your disposal for that. But it's going to be an advanced users thing, not a default, "let's make the new user mad as hell" kind of thing.

Because as ever, people are forgetting what BOINC was made for. It's time you start remembering that. It's not here to run science 24/7/365 on as much hardware as it can run on, to pump out credits by the gazillion. It's here to use the spare cycles on the hardware.

So, as a result, BOINC's default behavior is that science applications are not run while the computer is in use. This to stop it from slowing down the whole computer when people start using it. And we all know that work done on a GPU will slow down the computer, especially the drawing of the desktop. Some more so than others. And then the users either complain at the BOINC forums or at the project forums and they uninstall the software as it interferes with their usage patterns. Not what we want.

---

I asked the developers a while ago about this, their answer was "Why?".
I can see what they're going at, as BOINC is made to run one task per piece of hardware, so one task per core of the CPU, one task per GPU per videocard.

You can run more than one task per CPU, be it with the debug option built into BOINC, or with an anonymous platform file; that's then your own choice.

So for the GPU as well, the function's built in that we'll run work on it. You want to run more than one task on it, then use the options to your disposal for that. But it's going to be an advanced users thing, not a default, "let's make the new user mad as hell" kind of thing.

Because as ever, people are forgetting what BOINC was made for. It's time you start remembering that. It's not here to run science 24/7/365 on as much hardware as it can run on, to pump out credits by the gazillion. It's here to use the spare cycles on the hardware.

So, as a result, BOINC's default behavior is that science applications are not run while the computer is in use. This to stop it from slowing down the whole computer when people start using it. And we all know that work done on a GPU will slow down the computer, especially the drawing of the desktop. Some more so than others. And then the users either complain at the BOINC forums or at the project forums and they uninstall the software as it interferes with their usage patterns. Not what we want.

I had changed the idea to the advanced position before You replied Ageless, Although not for the reasons You state, Fermi on Seti(GPU) is what I was talking about, I never mentioned anything about the cpu of course, Unless You consider a Fermi GPU a CPU.

---

Play MULE, 42 is irrelevant

Oh, is there rules now? You changed what you said so everyone ignore the rest of the message, that what you mean?

If Boinc were hardwired to not allow older gpus like the GTX/GTS-200, 9000 and 8000 series gpus to do no more than one wu per gpu(Work Unit), then the existing advanced controls would work just fine and then only for the Anonymous Platform.

You're missing out on what the anonymous platform is for. It's there to allow you to run work on 'not standard hardware', 'not recognized operating systems' or to do work differently than how it's done by default. In all cases, it's there for the other than new or average user.

You're contradicting your self in your 'request'. First you ask for the option to be hardwired into BOINC that it only recognizes hardware X and not hardware Y to be able to do this, then you continue to saying you want the controls through the advanced view (preferences) and then you want it through anonymous platform... which overrides all what we had before.

Now, the last bit of a problem here is:
a) No other science application has its specific controls through any of the preferences (global or local), so why should it be done for Seti?
b) BOINC is more than just Seti alone. Please consider that even if it were possible to do this, it would have to work for all projects out there, not just Seti. And then work for them without bringing them into problems.
c) No work is ever done by BOINC, but always by the science application. So for any of this to work, you'd better change the behaviour of the science application, or add one to it with its own app_plan.

I never mentioned anything about the cpu of course, Unless You consider a Fermi GPU a CPU.

Please go re-read what I wrote, as I never said anything like that, nor even hinted to it.

---