Oh well as I had said , I have only really ever cared about Results Returned.
And I think once GPUs are in the mix then I should be able to produce
a boatload of results in a short space of time. I guess that's all that really matters.

Good to hear that under-clocking will be accounted for. I hope a 10% OC will not get 1.1days.

Well, it's down to the code in the system used to grant 60% on a 60% throttled card. How does it know it's being throttled? From the GFlops peak and an average performance expectation of that card. So, does it work both ways or is there a 100% cap?

I can only say what is done elsewhere...each Project writes the software to use the gpu on their own, often with help from Nvidia, and decides how much and exactly how to use the gpu. Most decide on a minimum requirement and then write the software to use just short of all of that. So if the minimum is say a 256 megabyte gpu then their software will use like 250 megabytes or so. In Boinc there are a TON of ways to play with the available options, MOST of the time running more units at the same time on a gpu with more than the minimum amount of memory is not helpful, but on some projects it is. So if a workunit fits into memory and takes only 780 megabytes of memory but the gpu has 2 gigabytes of memory, then on some few projects it IS better, as far as points generated and workunits returned, to run 2 workunits at once. MOST people do NOT have 2 gigabytes, or more, memory on their gpu's, MOST of us have 1 gigabyte or less. The cost factor can be a huge deterrent when going beyond 1 gigabyte, but when you need more memory it is extremely helpful. Video processing for example can benefit alot by having a 4 gigabyte gpu, but they can be VERY expensive!

In the end we will all crunch with what WCG, and any other project, provides us with the means to. We will do it their way because it is their project and we are just here for our own reasons. If we don't like it we can always move on elsewhere, there are LOTS of other Boinc Distributed Computing Projects out there. The key for us crunchers is to KEEP CRUNCHING!!

If we are going to talk about badging instead of GPU crunching my suggestion would be to cap the max runtime per user at 24 hours per day and have absolutley nothing to do with performance. Yes, I actually mean per user. Regardless of how many or what flavor hosts, cores, shaders, and whatever comes next, purely and simply showing an individual's commitment to how long they have been actively crunching with WCG on a per project basis.

In the end we have the stats sites and sig blocks for epeen so I don't see why the WCG badges get everyone one in an uproar. Can you imagine what some of the posts would be like if KR ever said that a new science app was delayed "X" months so they could focus reworking badges and converting UD credits ... how about if there was some form of outage or configuration glitch ... sorry, working on badges right now, let me get back to you in a month or so???

I appreciate the WCG staff for all they do and how well they balance the tightrope of doing everything necessary to keep WCG the best ever multi project BOINC host