Slow, wrong time estimation, and possibly wrong claimed credits. Granted credits ok.
My HP mini 110-1000, Intel atom N270 1.6 GHz, 1 core 2 threads, Debian Buster 32 bit, with 2 GB RAM and 500 GB SATA SSD. At first it shows ARP1 estimated time of 4 days. Finished with CPU time 169.11 hours (7.046 days), it went slightly past 7 day deadline, too slow. I got the granted credits so that was something good. I changed this computer to just SCC1 and MIP1, those runs at around 10 hours each on slow CPU.

ARP1_0025006_029
OS type   Status      Sent               Return             Hours      Credits
Linuxmint Valid       10/10/20 14:13:06  10/11/20 11:19:05  20.96   146.0 / 603.9
Debian    Valid       10/10/20 14:10:33  10/17/20 19:18:49 169.11 1,061.8 / 603.9
Linux  Server Aborted 10/17/20 14:11:30  10/17/20 20:23:25   0.00   489.6 /   0.0

Old Intel atom N270 (edit, not Z270), no ARP1, yes for other tasks. I use this computer for Linux Hostapd.

In the system requirements, the expectations and limitations are clearly defined.
Even if scientists try to take care of the contributors, the science should represent the main priority.
In case of ARP1, it is clear that based on the data volume requiring to be computed, it is not possible to build smaller WUs without making task tayloring and result validation more complex and finally less efficient.
On a 10 years old Phenom II x6 (Linux), an ARP1 WU takes currently between 22 and 28 hours. We do not speak about weeks or month.
Cheers,
Yves

146+1061.8 divided by 2 is 603.90 or thereabouts. The averaging rule work in someone's favor and bold claimant what was due, albeit 169.11 hours computing is longer than the 7 day deadline by 1.11 hours. A device not fit for purpose.

An interesting unit has just been received, A new one but only 3.5 days allowed for both me and my wingman. It is an 030 so WCG are presumably trying to get it to catch up with the rest of the 030s.

Mike

Billy

The problem with arp1 is that it is an iterative process and they can't release the next unit for that patch of Africa until that one has been returned. Nearly all are returning their units within 3 days which would keep their machine listed as reliable because that enables them to get resends which have a deadline of 3.5 days instead of the usual 7 days.

You didn't say whether your unit had a 3.5 or 7 day deadline. You also didn't say how big a cache you were holding. If you hold a large cache, the units take too long to reach the front of the queue and therefore miss the deadline. I would suggest that you reduce your cache to no more that 2 days plus 1 day in Device Profiles. Go into Device Profiles and set it to 'Custom' and then you can change the cache settings.

The only reason for having a higher cache setting is if the units are not readily available, but all projects at the moment are readily available.

That should fix your problem only 4 hours after your posting.

Incidentally, if you overrun by a few hours the next person would not have enough time to return it and so you would get the credit. If he hasn't even started the unit WCG would abort his unit so there would be no duplication. If he beats you, you would not get the credit.

Mike