Dodgy work units?

[adrianxw]

This morning, I noticed three ecm 705.02 work units in my queue with unusual remaining values. All three are showing 23 days and small change. Now, these work units normally complete in less than a day on here, (4GHz i7), they are not the fastest, but these three do seem to be somewhat excessive, especially with deadlines next week.

[Michael H.W. Weber]

Yes, I observed something similar, yesterday...

Michael.

[adrianxw]

Those three have had a good few hours running time each, but are still in single figures in terms of percentage. I've multiplied their priority by ten so they should be going 24/7 from today, they still won't finish by the deadline, and the way BOINC works, they'll be sent out again to someone else. I remember in the early BOINC days, projects like climateprediction's work units used to run routinely for weeks.

[yoyo]

I'm pretty sure, they do not need 23 days.
What is you Duration Correction Factor?

[adrianxw]

I found the figure for all projects except one was set to 1.000000 as expected. Yoyo is set for 3.811421. I set it to 1.000000 and restarted, the figure reverted to the previous larger value.

[adrianxw]

These six tasks with considerable run times have crashed recently...

117764197 83362159 24 Apr 2026 20:02:03 UTC 25 Apr 2026 10:06:05 UTC Over Client error Compute error 7,380.77 125.10 ---
117762886 83360854 24 Apr 2026 13:36:42 UTC 24 Apr 2026 23:48:44 UTC Over Client error Compute error 4,084.06 69.22 ---
117762793 83360725 24 Apr 2026 13:36:42 UTC 25 Apr 2026 11:07:01 UTC Over Client error Compute error 16,233.70 275.15 ---
117759931 83358690 24 Apr 2026 13:36:42 UTC 25 Apr 2026 7:03:14 UTC Over Client error Compute error 12,038.89 204.05 ---
117758934 83357675 24 Apr 2026 5:49:58 UTC 25 Apr 2026 2:53:52 UTC Over Client error Compute error 23,321.56 395.28 ---
117755987 83354985 23 Apr 2026 19:12:15 UTC 23 Apr 2026 22:01:53 UTC Over Client error Compute error 4,564.05 77.36 ---

I have seen an error box saying...

The instruction at 0x0000000000436F88 referenced memory at 0x0000000000000000. The memory could not be read.