I also got one of these awkward tasks -- ARP1_0034097_092 -- as part of a triplet. It took 22 hours CPU time on a Ryzen 3700X (running Linux) that usually finishes ARP1 tasks in 7.5 to 9 hours; that's indicative of a significant increase in the number of iterations per task! :-)

It'll be interesting to see if these tasks need to stay on the shorter time slices to keep them running properly, or whether the time slices can be relaxed again at some future point... If the issue was "geographic" I suspect the short time slices will be here to stay...

To track this over a longer period, it might be useful to have the "cell numbers" of the problem tasks (as eventually the low batch numbers won't be a clue any longer!) -- as mentioned, I've just seen 0034097, adriverhoef reported on 0033475 (also mentioned by spRocket), and Sgt.Joe reported on 0033946 -- that's three I'll set a watch for, and I'll be happy to know what the others are!

As for the 32-bit versus 64-bit thing, I've just modified the daemon I have that tracks completion of ARP1 tasks so that it will report the platform as well as the software version; if I do get any [more] 32-bit tasks now, that ought to warn me about them!

Cheers - Al.

P.S. that Ryzen currently runs 3 ARP1, 5 MCM1, 1 HST1 (if I can get it!), 1 OPN1 and a GPU job (usually not OPNG, as I don't deliberately try to get work if it isn't offered!) -- I have one or two other things running (such as monitoring software!) and if I don't leave at least 25% of my cores free, I find the BOINC client can get seriously blocked up when ARP1 or OPN1/OPNG tasks finish, even with my best efforts to stagger task completions!

[Edit - fixed a link and a typo...]

That is 3 out of only 57. Please keep them coming. I have logged them on a spreadsheet. It may be geographic because those are clustered about 33/34 000. The generations involved are currently 086 - 103 but anything from 079 - 110 will be unstuck units.

Mike

Hi Mike,

I've got 0034176_092, 0034177_094, and 0034706_090, which are all triplets.

Cheers,
Mark

entity

I can identify the 6 ultras from their generation numbers and Kevin isn't doing the remaining stuck ones until next week. We have a few days to identify them.

I am wondering if these 3 groups have distinct unit ranges?

Mike

Mike, if you don't have these results yet in your spreadsheet, this is copied from text on my computer screen (scrolling back ):

* ARP1_0019736_068_0  Linux Fedora  Valid  2021-12-27T07:04:16  2021-12-27T21:07:18   12.89/13.31    637.8/652.8
  ARP1_0019736_068_1  Linux Debian  Valid  2021-12-27T07:01:11  2021-12-28T06:51:41   12.50/12.50    667.8/652.8 
  ARP1_0019736_068_2  Arch Linux    Valid  2021-12-27T07:01:10  2021-12-28T11:31:49   14.46/15.10    710.8/652.8 

  ARP1_0033949_098_0  Linux Ubuntu  InPrg  2021-12-28T21:05:21  2022-01-02T09:05:21    0.00/0.00       0.0/0.0   
  ARP1_0033949_098_1  Linux Ubuntu  InPrg  2021-12-28T21:33:23  2022-01-02T09:33:23    0.00/0.00       0.0/0.0   
* ARP1_0033949_098_2  Linux Fedora  P.Val  2021-12-28T21:18:26  2021-12-29T20:42:53   15.70/16.18    786.5/0.0