Apis Tintinnambulator Close, it’s 14days ;) It would be more accurate to think of a GPU as a several hundred low power CPU threads with more limited abilities in the one unit. While cards can run one task at a time very well, one task does not necessarily use all the resources. AMD’s GPU’s are especially capable of running many apps simultaneously.

It’s already been stated that there will Not be a separate badge for HCC. Why? Apparently, because GPU’s will continue the existing work. Does this mean there will no longer be HCC for CPU’s? Don’t know, but I guess so, otherwise what would be the point of saying there is 52days to go?

So on the one hand it's the same project.
On the other it's moving from CPU’s to GPU’s (different processing devices) and using different apps. Does sounds very much like a second phase of a project to me! I bet there will be a separate paper on the move to GPU and how greatly it improves turnover/performance.

The only problem I could see with having a different project is if CPU processing will actually continue for this ‘same project’. Then having a separate GPU exclusive project, would inherently be elite. Something I’m not keen on here. That said, existing crunchers would already have made a contribution.

There is also the bigger (long term) picture to consider (future GPU projects); will this decision mean other potential GPU project have to be add-on projects or replacement projects? Well, now is certainly the time to consider such things, and make informed decisions, for now and the future.

Re badges, I have already expressed my concerns about 1day runtime credit for high end GPU’s, verses for example, 4days runtime credit for an atom CPU!

I also made what I thought was a more reasonable and respectable suggestion, based on core config or NVidia’s SM count. Whether this was even considered, I can’t say. If a 980X can get 12days runtime why can’t a GTX580 even get a respectable 16days? We know their relative performances on projects that can utilize both.
Anyway, it's not my decision and I will try to respect whatever decision is made. My decision is to participate, or not, and will be based on my opinion of the HCC project's merit, rather than WCG's longevity based badge system (and I'm obviously more points orientated anyway).

Regarding the one day runtime credit, I would point out that the OpenCL/GL apps might use considerable amounts of CPU. So while you might get a full day for your high end GPU, you would lose some CPU credit and runtime. How much will depend on the app (no info on system usages), the CPU performance, and GPU. If the app does need considerable CPU support then the faster the GPU, the less runtime credit you will get from the supporting CPU. This is one of the reasons I suggested a GPU badge system based on points.
When some crunchers start running multiple apps, and they will if they can, this will undermine the credibility of the badge system - a second reason to move to a points based GPU badge system.
A third reason would be if you plan to support future GPU projects. Over-rewarding CPU’s relative to GPU’s in a badge system seems counterproductive. If you really intend to facilitate future GPU projects, stick to credit, and why don’t you have a separate WCG-GPU project, with separate stats, badge systems and so on?

WRT GPU memory capacity; GPU's can allocate system memory, and do so on operating systems such as Vista and W7, so low memory shouldn’t be a barrier to multi-apping. Also, the apps will most likely adjust for available memory.
Most high end GPU's have over 1GB GDDR5 on-card, and some, such as AMD's Antilles 6990 have 4GB – plenty of room there to run multiple apps simultaneously. Even some medium or entry level GPU’s can have relatively high memory amounts (GT520 2GB). 3 GT520's running 4tasks each would give 12days, but (app utilization dependent) do much less work than a GTX570 running one app at a time. Do you want this?
In March AMD will release it’s latest flagship, the NZ 7990. It will ship with 6GB. The lesser 7970 will ship earlier with 3GB and the 7950 will have 2GB on-card. While the NZ 7990’s 4096shaders (1GHz) obviously can’t be used for 4096days, do you really want this dual card to count as two days CPU runtime, less the CPU runtime loss to support the card (less than half of an atom)?

Lets see...the objective is to find a cure? or at least help find one. Like someone said above, our decision is to participate or not. I vote for participation with all the CPUs and GPUs I can muster. If I get more or less points or badges than someone else, I will still feel the same inside and I will still get the same number of Christmas Presents.

One of your friends in Texas

Fascinating. So it seems it comes down to who has how much control of those resources. If WCG's WUs (BOINC?) can request the best use of those resources, and decide for themselves whether to use them all or only to use some and to let (an)other WU(s) use the rest, then the resources will be well used whether the "card" is doing one day per day or many more. Brilliant. That seems fair to me.

If the user can tweak parameters so that each WU only gets a minimum of resources and the card runs a zillion WUs slowly in parallel then they'll get a blue badge in no time. Fine. That just shows that they're good at tweaking (and they're probably not doing the maximum feasible amount of work). It's vaguely annoying but I won't loose any sleep over it.

A separate GPU badge system would be good for those who like to boast about how fantastic their 'puters are, but it doesn't help humanity. And setting it up will probably take time away from the work the techs do to get new projects running on the grid.

I regard the badges simply as showing our commitment to help by providing runtime resources. I don't care if you have a palmtop or a roomful of crunchers -- just crunch what you can and be proud that you're helping.

I agree about the system inequality, but remember that we have THREE credit systems:

• Total Run Time;
  
• Points Generated; and
  
• Results Returned.
  

For whatever reason it was decided to assign badges to the first one only. Not the other two, and not a combination.

I think that that was the correct choice. The other two depend on money. The first depends on commitment to help. That's what we need.

Perhaps something worth reconsidering then, before the GPU project starts?
I would be concerned about promoting and encouraging the use of systems and components that contribute less; older systems are usually a lot less energy efficient. Using more energy and doing less work hardly helps humanity.