I can relate. I'm just shy of a million points, and it looks like I might not make it today at all because my system is processing an x16 work unit with a preferred deadline of 6 days and zero bonus points for early completion. Making matters worse, my current a4 work unit has the CPUs running warm all night and day and it won't even finish until sometime tomorrow morning. While typing this post, I just now saw my CPU cores go to zero usage, then ramp up. It appears my a4 client died and got restarted. Could it be that my current a4 WU is bad and having trouble processing? Don't know.
The way I understand it, the x17 work units have a higher early completion reward because they will only run on more modern, faster, and more expensive graphic cards. I think of it as a reward for subjecting my $600 card to higher heat and greater risk for early failure. When we temporarily run out of x17 work units, then you simply get what's next in line; usually an x16. These work units are less efficient on your GPU, and they use more CPU too. I use Process Lasso and I can see that the CPU cores are running more work whenever I have an x16 WU. And yes, x16's seem to take forever to complete and to make matters worse, the early completion bonus is exactly or almost zero.
There's nothing we can do really, because there is no way to know what's next or when we'll see x17's again. You get whatever is next in line at the time your current GPU work unit reaches 99%. And you don't want to get into second guessing, trying to time your request for a new work unit, or worst of all cancelling work units or anything, because that slows down the progress for work units that are already less efficient anyhow and eventually, you'll get penalized for lost work units anyway.
I think the way it works is when we deliver work units to the Pande servers, they have to do some calculations in order to "create" and release back to us more work units, based on the results we sent them. The servers have to split the work up so that work units don't take six months for a modern PC to complete. I'd guess that this creates a huge batch of x17's to become available all at once, and the people folding simply get one of them, working our way through the queue of available work units. But eventually we run the queue dry and have to return all of the work before the servers can use our results to create a new batch. Just a guess of course, but this would explain the "dry patches" you and I have noticed where we'll go a day or so without any x17's.
Also, I've been wondering if maybe some point in the future, x17 bonuses will be decreased, as more and more people upgrade their folding rigs to have the necessary hardware. I don't know what the Pande lab's policy is on that, but my strategy is to try to gain a points advantage by quickly standing up another system with a couple of less-expensive HD 7970 graphic cards in them to get as much bonus as possible while it's still lucrative. Yes, I am competitive, so I'm thinking of bringing my old Q6600 back into service temporarily to see how well that quad core system will fold x17 units with a couple of $199 cheapie 7970s and nothing else on that system. If the CPU work units are too strenuous/not valuable, then I'll just let that system do GPU work instead. If this plan works well, then I will only have to spend money on the graphic cards and no money on a cpu, memory, hard drive, motherboard, or OS, and will be able to take more time to plan a newer, proper rig for exclusive folding.
Haha, so if I do that, you and I will be fighting each other even more than now for those golden x17 work units.
Sorry, I babbled too much. Just hang in there, I'll bet we see more x17's before the weekend.