Decision was already made and shared: 1 project, 1 badge, no matter the resource on which computed. Dedicate your GPU 24 hours [elapsed] and it will clock 1 day. If the card is capable of running multiple tasks concurrent, it will clock the multiple per day.

But, someone launching ideas/suggestions is held to also come with a proposal how this than "fairly" would take shape, to serve both CPU and GPU. Got any?

Consider these stats -

Compare the Collatz and milky way with WCG - the credits were obtained with 2 GPU in about 2 months running compared to the WCG Credits which took years with 6 CPU. I am aware that the credits for some projects are inflated well above WCG, but, my point still remains about the awarding of 'rewards'.

I got the opposite out of his comments, he thinks it is WRONG to do it the same as other projects. If you crunch a gpu unit and it takes 6 seconds, okay you got 6 seconds towards your badge, the idea is you can crunch thousands of those 6 second units in a day as compared to three 8 hour units from another project using your cpu. Having more and or faster gpu's is just like having more and or faster cpu's, more work done in a day and more total credits but not more credits per unit. I currently have a little bit over 20 cpu cores here at WCG, but I don't get any more credits per unit than the next person, I get more total credits per day as a result of more resources allocated here. Sure I finish more units per day on my AMD 6 core cpu than I do on my Intel dual core pc, but that is just a function of the AMD 6 core being significantly faster and more efficient.

I will continue with WCG, POEM, malaria control and Folding regardless of badges. The badges are nice to see, but I am happy as long as my work is useful to the projects. I like to read periodic project updates to see the research progress: I don't always understand all the information, but that's OK. Credits/points just show me that my results have been added to the big results pot.

This argument is no different than the dynamic that has been occurring since the launch of WCG. An i7, or more specifically one hyperthreaded half of one core, performs far more calculations per hour than my 486-66 that first crunched the grid and yet it is rewarded with the same amount of run time per work unit.

There are multiple metrics for a reason. Run time counts run time. Points count points. Never the 'twain shall meet.