Everyone is having this problem, and has been since the project started. It has to do with the interval between some of the checkpoints. There is no fix. If you want to make any appreciable progress on CEP2 you have to run 24/7.

I do not have this problem after my computers have been in hibernation or sleep mode. So running 24/7 is not necessary.

Yes, I have the same problem.
I do not want my Computer on 24/day and this error ought to fixed - surely
saving the result more frequently could reduce the times lost.

Expecting a 24/Day run is not accectable and so I have removed this Project from
my Project list.

Why not make a suggestion to the Authors of the Project to fix this,
thus increasing the number of people running the Project.

Alll in all very, very disppointing for such an interesting project,

The researchers at Harvard are already well aware of the checkpointing situation but finding a solution is difficult as lavaflow explains. The latest Harvard post in another thread says they are trying to find an acceptable solution.

How your laptop crunches on and on and is not able to complete an at-most 18 hour result, no matter how powerful the device, pass, but one guess as part of this is, that the task unloads each time the screensaver comes off and u using the computer. As i wrote in my previous post -with- 'leave application in memory when suspended' aka 'laim' on, you would not have that. Progress is retained in memory and eventually written the virtual memory on storage to free up ram for the user. Of course booting does go back to last checkpoint, but hibernating/sleeping does not. It's like closing lid and opening the next day and continuing exactly where the task was. Same as any other open user application, doc, worksheet, webbrowser etc.

On the percent reversal from 40 to < 37, say 3 percent loss, this indicates it was not far passed the last checkpoint. There's 8 jobs in 1 cep2 task, so pure mathematically each job represents 12.5 percent. It's though the jobs are ordered in hardest the easiest, so once past the first, the reboot/resume losses are smaller. If a task has not checkpointed yet, laim is automatic, so the going back from 40 to 37 indicates it completed at least one job, the most important one.

There are those not very amused, yes, and those probably did not do the conscious reading before opting in on the demands and resource requirements, or? Right in the device profile where you choose cep2 it says

• Please review the system requirements before opting to participate in this project

There the memory/disk maxes are detailed and footnote 5 spelling out:

5. Users who choose to run this project are encouraged to set the 'Leave applications in memory while suspended' option in their device profile

Sorry, but all key 'must take note' information is right there, plus, there's the technical discussions in the cep2 forum how to max out on performance and returns.