Where are all the Core17 units?

[HaloJones]

I'm talking about Core17 not BA. Old units with crap efficiency for both power and folding. Core17 is supposed to be the Great White Hope so where is the work?

[bruce]

I suppose Santa's core17 elves are busy at the workshop ... or home for the holidays.

There's no doubt that core17 will be the Great White Hope but like in that story, the transition from previous ways of doing things will smooth out in time but it's a slow process and some glitches can be expected along the way.

[VijayPande]

We have several new core17 projects on the way as well as more core17 WUs for existing projects. I'll get an update from team members to see where that stands.

[bcavnaugh]

Maybe they need to work harder this week, and get them out

[bollix47]

Last Friday (20th) the core 17 server was down for a few hours. While this was taking place Windows clients would have been assigned to a core 15 work server for nvidia cards and to a core 16 work server for AMD/ATI. Native Linux GPU folding would have had to wait for work. Looking at some of your stats does suggest that's when your problems started (actually your drop in points is more obvious the following day which would make sense).

If you haven't already tried the suggestion made in this post please give it a try. Doing a client restart after finishing and uploading a core 15 or 16 work unit has usually (always?) returned my client(s) to the core 17 work server. I have received nothing but core 17 work units before and after Friday's brief outage (one of my Windows clients did receive a core 15 WU and another one received a core 16 WU during the outage but both returned to core 17 after setting them to finish, waiting for them to upload and, after pausing any other slots for that client, quitting and restarting the client ... without the client restart the procedure has not been successful).

Personally, I believe this appears to be an assignment server idiosyncrasy (i.e. the AS keeps returning the client to the same WS unless you restart the client as described above or make certain config changes ... perhaps done in an attempt to speed up the assignment process) rather than a shortage of work.

My speculations may be way off base but using the procedure has worked successfully for me on a number of occasions.

[HaloJones]

Tried that a number of times including every client-type setting with empty WORK directories and it hasn't helped although I now have two 8900 units on two of my cards. Personally I find it interesting that this happened immediately after 7627 got moved to Advanced.

[7im]

...

Personally, I believe this to be an assignment server idiosyncrasy (i.e. the AS keeps returning the client to the same WS unless you restart the client as described above or make certain config changes ... perhaps done in an attempt to speed up the assignment process) rather than a shortage of work.

This behavior was noted in v6 client at times, but never explained. However, V7 requests work before the previous WU is done. There is no performance to be gained, and no previous communication to link the two like in v6. In V7, it's simply the result of assignment logic. And with more virtualized servers, and less total servers, the chances of bouncing around are even less.

[VijayPande]

From what we see here, there are (and have been since I posted above that we'd check) many core17 WUs. At least that's what we see on our side. What donors seeing on their side?

[uncle fuzzy]

When I can connect to the servers, I'm getting all core 17 for 8 GPUs, and have been for several days.

[IsraeliRD]

My 660Ti is getting core 15 (p8018) only unless I add client-type advance, then it switches to p8900. When removing that config from the slot setting it switches back to core 15.

[HaloJones]

For me, the loss of Core17 started on the 20th with none at all until midday on the 24th. No setting for client-type made the slightest difference to the core I got with all of them being Core15. On beta I got 7627, on Advanced I got 8018 and on null I got 7660.

I'm now set on beta and getting Core17 again.

[parkut]

I was assigned an 8900 on the two GTX 550 ti I have, after nearly a month since the last time any were assigned. No restart was required.

[Mstenholm]

he he. There are no information about avail WUs for that server so please check facts before ask others to do it.

The lack of data for that one server does NOT change the standard process of how to find the WUs available. I gave good info. Not my problem if Standford screwed up. That's their problem. Criticize them for not following standard procedures.

Indeed you gave the standard answer. The difference between your mentality and mine is that I check as good as I can before I give an answer. It was a new guy that asked...

Edit: It was a "new guy" question. Sorry, my very bad.

[7im]

Yes, that is true. The difference between your mentality and mine is that you would simply look up the number of available work units, and then post that number in the forum like most people. I, on the other hand, instruct people how to find their own answers... in essence, I teach them how to fish instead of just giving them the fish. I'm okay with that, either way.

[Blasko9]

From my POV work assignment does not seem to be priority based.

Running 2 650t1 and 1 Fermi card (switched from 550ti to a 580) in the same box.

Since Core17 went to full release (no advanced flag required) I have been assigned p8018 exclusively.

Prior to full release I would alternate 7-10 days with p8018 and then 7-10 days with a mix of mostly Core17 and some Core15.

If priorities change from time to time I would expect to see a more normal distribution of projects especially when compared to others on my team. While I see p8018 another GPU folder on my team sees only p7627. Shouldn't I see 7627 mixed with 8018?