One of my crunching machines has an Atom processor. GFAM and SN2S are now taking around 80 hours on it. No problem for me, since the machine runs 24/7, but if that was somebody's netbook and they ran BOINC only when it's turned on for other reasons, those certainly would be "no reply."

I know -- it's crazy to crunch on an Atom. But that's all some people (such as students whom we really want to hook on crunching) have. I crunch on mine because I own it and it seems crazier to let it sit idle on the table.

Congrats and NI!

andzgrid, I'm thinking the same thing. These bigger WUs are set up better for the big crunchers with new technology that is significantly faster than most smaller crunchers provide from their homes or possibly small office environments.

Now my laptop that is coming up on its 3 year anniversary is still keeping up pretty well. And a second home laptop that is about one year old is hanging in there, but I bet there are a number of smaller crunchers out there with desktops/laptops that say are pushing 5+ years old that are still running fine, but with these new bigger workloads may be pushing their limits on how much they can get done in a 10 day period.

So either like you said there may be a need to put out a second smaller WU group for crunching, or maybe the 10 day limit needs to be extended. If the workloads are 40% to 90%+ bigger (depending on project), maybe we need to go to a 14 or maybe even a 20 day limit for crunchers to come in with their results. This may raise the PV numbers quite a bit, but could be one option if the WCG technical team finds that does a better job of balancing the workloads across servers.

We may be getting away from the original intent of grid style computing, which is, as I remember it, to use your spare computer cycles for research. As grid computing has become more widespread there has been a growth in the 24/7 computing devoted entirely to crunching. I am one one of these. There are a variety of reasons why people do this - social conscience, competition, willingness to advance science/computing etc. All this being said, there should still be a place for the casual cruncher with an older rig who maybe only gets 3 - 5 cpu hours a week. Even though their contribution is small they are still contributing. Thankfuly there are still projects with workunits which can be successfully crunched in this time frame. At such a time when this is no longer possible, either the base requirements can be raised which would exclude the smallest, slowest systems, or perhaps the server code could be tweaked to appropriately sort and size units for these systems. The cost is time and effort to do this needs to be weighed against the total benefit which may accrue. If the total contributed by the smallest contributors is so small as to be below some arbitrary number (1%,.1% or .01%) the cost of providing very small short units exclusively to the slow systems may not be worth the cost. I believe as long as the present system has at least one project which works for these systems, no changes are necessary at this point. Just my two cents worth.
Cheers

Sigh. It *cannot* be done, not longer, not shorter. The Techs went about last month to extend what they could. Would it have been possible, would it not have been done before, considering the performance issues we suffered in previous months?