how many a2 SMP WUs are available?

[alpha754293]

how many a2 SMP WUs are available?

[7im]

Not many at the moment. And I think we have atleast one thread going on this already.... http://foldingforum.org/viewtopic.php?f=44&t=8995

[alpha754293]

Not many at the moment.

Is an actual count available?

[toTOW]

We don't have any way to know that information, sorry

[alpha754293]

We don't have any way to know that information, sorry

No problem. Thanks.

[7im]

And even if the Project Summary page showed us which projects were a2s, and then we looked for the total WUs available on the Work Server sending out those projects on the Server Status page, the available total would change from day to day, hour to hour, as people downloaded them, and as Pande Group added more to fold.

Plus we don't know if and when the a2 core will be replaced by a3 or whatever. a2s could be around a long time, or a short time.

[alpha754293]

And even if the Project Summary page showed us which projects were a2s, and then we looked for the total WUs available on the Work Server sending out those projects on the Server Status page, the available total would change from day to day, hour to hour, as people downloaded them, and as Pande Group added more to fold.

Yes, I understand that. But at least there might be some account or rough guestimates as to how many there are available so that I'd be able to figure on the precentage available, and then the probability of getting one relative to an a1 WU.

[7im]

With the ever changing mix of work units and projects running, the probability changes way too often and too much to make any prediction meaningful, IMO.

But if you could, what purpose would it serve?

[alpha754293]

With the ever changing mix of work units and projects running, the probability changes way too often and too much to make any prediction meaningful, IMO.

But if you could, what purpose would it serve?

Well, I wasn't looking for any sort of dynamic probability. To do that, it isn't difficult, you can probably do it with MATLAB and you just have a simple text file that states how many a1s in/out, how many a2s in/out and run with that. And total WUs.

With that, you can have at least a first-order calculation. If you want to get uber fancy, you can always decompose each of the servers into the various types, and then use 6th order chaos for your probabilty calculations. overkill, but it would work.

I would have thought that there would be a quick way for you guys to check the systems so that you would know what's going on roughly, and naturally, there'd always be presumptions with it and in the end, you're going to only get an approximation anyways.

In fact, even the simulations are approximations. Nature of the beast.

So, by that decree, you can always ask "what's the point of doing ANY computations?"

[7im]

...
In fact, even the simulations are approximations. Nature of the beast.

So, by that decree, you can always ask "what's the point of doing ANY computations?"

While I can't give you the probability of getting an a2 work unit, the probablity of contracting one of the many diseases studied here is surprisingly high, and getting higher each day as we live longer lives. Factor in a spouse and offspring, and the numbers start to get ugly even just for your immediate family. That's my point for doing this... and everyone has their own.

In the US, the probability of developing cancer by age 65 is over 15% and goes up sharply with age. At age 75, 1 in 2000 will get throat cancer, 1 in 800 gets stomach cancer, 1 in 400 gets bladder cancer, 1 in 225 gets colon cancer.*

For Alzheimer's at age 65, it's 1 in 10 and increases to 1 in 7 by age 75.**
Huntington's Disease (HD) is 1 in 20,000.***
Parkinson's is about 3% by age 75.****

*http://www.cancer.org/downloads/AA/CancerAtlas10.pdf
**http://www.provistals.com/AboutAlzheimerDisease.aspx
***http://en.wikipedia.org/wiki/Huntington ... idemiology
****http://ezinearticles.com/?Who-Gets-Park ... &id=152916

The life you save may be your own.

[bruce]

There is information on the server status page about how many WUS AVAIL for each SERVER, but I've never been able to learn enough by studying it to make reasonable conclusions. In general, each server manages a number of projects and I don't know any way to figure out if most of the WUs belong to a few projects or if they're distributed across more of the projects that belong to that server.

As 7im said, each project may be managing a certain number of Run/Clone combinations but at any given time, most of them SHOULD BE assigned and actively being processed by someone. When the results are returned, a new WU is (almost always) generated and available for assignment. I do know that the Pande Group does add new WUs to a project, but from watching the numbers on the servers, I think that''s a rare occurrence. It seems like new projects are created more often than old projects are extended. [But like stock brokers like to tell people, past performance is no assurance of future events.]

Then, too, you'll find that your probabilistic calculations will have to figure out how the assignment process uses the WEIGHT setting. It's some kind of priority setting, but with numbers like those, I'm not sure how they're used and it's certainly not clear what projects they're applied to.

If you learn anything new, be sure to let me know, too.

[susato]

Thanks 7im for reminding us about the probabilities that really matter in the long run.

[alpha754293]

Well, the end result or the objective of the project is quite well documented. I'm presuming this much only because I've never really bothered to read it, I just run it.

So, given that we kind of already know what the end results are; therefore, we can step away from the macroscopic picture and start looking at the microscopic picture of running it.

Naturally, asking a macroscopic picture question at the microscopic level is kinda pointless/senseless. So, ergo, rather than asking for the probability of my system working on a WU that would be relevant to a particular disease; I figure that the general classification for SMP clients might be easier to determine. Apparently, I was wrong. Oh well. What you do. Spit happens.