Could be there are so few in the HR group [of CPU's], that there's yet a new [reliable rated] wingman to be found. Saw there are but 30-40 of those Bulldozers at WCG and them being in their own for DSFL/GFAM... the possible reason, and why techs hate to narrow down CPU matching.

Whilst, [almost] 2 days sitting there may need a nudge or a wrench :D

--//--

edit: insert [reliable rated]

Now coming up on 6th day waiting to send, and 24 hrs from overdue, if HR matching is that hard, maybe something is wrong with the rules, you can't tell me I have the only reliable AMD XP class CPU in all of WCG, if that is the case, send it back to me I have corrected the HP issue I had last week and will get it done

We are looking into it. As Sekerob indicated above, it is stuck because an eligible host hasn't requested work from the server. This is because the size of the HR class for these is very small. I queried the database and there are 30 other workunits in the same state. There were some additional computers contributing that should be part of this class that weren't being identified and I've added those in so we will see if that is enough.

Knreed,
Moving the discussion from the beta thread to here, it seems to me that there are still quite a few of these machines producing for WCG but the issue appears to be coordination.

With yet another VINA application soon to hit production, I would like to propose another shot at this redundancy class. This time, however, inform the crunchers prior to making the class live. Perhaps an e-mail to all known socket A machines and see if it isn't viable then.

I and several other that I am aware of had already deselected all of the VINA applications before you implemented the new HR class. If we had known you made this we would have contributed and probably made the rule viable.

Anyway, something to think about and I hope to hear from you.

knreed,

Any grid is going to have to handle a declining population of a CPU class one way or the other anyway, so how about the idea of combining a declining population of an HR-class with some other class? For the matter on hand: if Socket-A can't justify having its own HR-class, how about combining Socket-A into other AMD (or Intel? If that works regardless of the OS, why not) HR classes? I'm not sure though how the implementation details particular to the Socket-A situation may impact the goal of using as much of what hardware is out there as much as possible to the level of 'until death do us part' kind of thing.
;