You're not wrong Sek, but I would argue that we ought to be able to take part in a project if at all possible. My slow machine takes about 48 hours to process a FAH2 WU, so why not let CEP2 WUs run that long? I know that there are arguments about lack of checkpointing, but it would then be up to me to decide to opt-out due to too much time lost because of uncheckpointed failures. I can, however, imagine a scenario where someone would keep processing CEP2 WUs in the hope of getting enough that finish step 0 in under 18 hours so as to at least gain a bronze badge. That could involve a lot of wasted processing time.

Also, any machine which runs for more than about 8 hours without a restart is likely to be going to run 24/7, isn't it? Why is 18 hours so special?

But finally, whatever the credit granting process for production, I thought beta was supposed to work differently. Not, apparently, in this case.

Thanks for listening, Keith! I look forward to reading about the response and the subsequent decision(s), even if they go against my ideas. I recognise that only you guys have the full version of the bigger picture, but it really does help make us feel needed if we're kept aware of what's going on.

Yeah, I have several old machines that I would love to run CEP2 on, but they will never finish in 18 hours. I get why they have the time limit, so that if a task goes off the deep end it will eventually end, but many computers just cant crunch fast enough to meet the deadline. Anything running a Pentium 4 definitely falls under that category, and I have a 3rd gen i3 and a 1st gen i5 that sometimes have trouble finishing. heck, even my brand new computer with a 6th gen i7 sometimes wouldnt finish in under 18 hours (this was before the WU availability issue). I would be awesome if there was an option to override WU timeout or something. Either in the device manager on the WCG end or in the BOINC manager, some way to manually extend the time limit. By default it should be set to not override, but if you wanted to you could change it.

By the way, what happens to unfinished WU's that dont throw an error? Do the CEP2 servers just finish whats left, or do they have to send the task to a new computer and start over?

Just to be sure you're on the same page as others v.v. "but many computers just cant crunch fast enough to meet the deadline." Are you talking about the 18 hour cut-off or the actual deadline a result has to be reported? In latter case, the problem is member-made by setting higher buffer values **. Remember, if you have a 2 day deep work buffer, after a while of crunching, a task is already 2 days old before started, but since CEP2 has a 10 day deadline, it would not be an issue. Those who do have a 2+ day buffer eventually would not get any repair work since results are not returned regularly in under 48 hours i.e. they do not get those tasks with the 3.5 day deadline.

** Devices that don't run 24/24 , part time, best don't have any buffer at all... just compute, complete, report and at that time let BOINC fetch a new task to occupy the core that is threatened to go idle. Any buffered task 'ages'... the deadline clock starts ticking the moment they are received.