Of course it is. The evil Checkpoint Reset Bug is everywhere. Don't forget to look under your bed before you go to sleep.

Aurum's referencing an issue that he had with using third party software. BoincTasks's "suspend after checkpoint" function apparently wasn't working quite right just after launch. By timeslicing with another project, he lost some work because the units reset to the last checkpoint. No diagnostic info provided for that, but that could happen if the application data wasn't stored in memory when the other units were crunching. Finally, he seems to have alleged an issue where the sixth checkpoint was not created on a task (at 75%, a sixth checkpoint should be created, because a checkpoint should be created every 12.5%), but I don't remember any data on that task, or comparative data, actually being posted. If he can post data, I'm sure the techs would love to see it so the issue could be identified. However, there's a chance whatever third party software he's running may be interfering. I don't remember anyone else having these problems running any sort of vanilla BOINC installation.

There's also general complaints arising from work lost because an unexpected shut down occurred and work was reset to a previous checkpoint. But that's not a bug; it's just the limitations of the software and the models being used.

Can't say for sure, but your error doesn't look like it has anything to do with the 1st checkpoint. Not sure what a segmentation violation is per se, but hopefully someone can look at it and make sure it was a one-off error and not something systematic. Hope that answered one of your questions!

I suspect that "segmentation violation" is due to hardware, and memory in particular. It doesn't mean that it is bad, only unstable. I sometimes see it when four memory modules are used. Yesterday, I just pulled out two 16 GB modules and left two in place, since I was seeing that error once in a while on CPDN, where I did not see it on any of my other machines. I think that will fix it.

The speed rating of memory modules is for only two by the way. You take your chances with four. You might need to reduce the speed to the motherboard default (probably 2133 MHz these days) in order to use four reliably.

1. Not third party software but BOINC 7.9.3 & 7.14.2.
2. Not using "Suspend at Checkpoint" from BoincTasks.
3. Using timeslicing because ARP1 WUs can only be supplied for less than 20% of CPU threads. I added Universe@home and even with "leave application in memory" checked for every computer it dropped back to the last checkpoint when BOINC switched from WCG leading to Universe. Since there are not enough ARP1 WUs to occupy every CPU thread it adds BHspin2 to fill out the list. When the lead switches to Universe it's all BHspin2 WUs. When it switches back to WCG leading all ARP1 WUs restart at their last checkpoint. This does not happen when ARP1 is filled out with HST1, MIP1, and MCM1 because it does not seem to switch ARP1 WUs off.
4. This is from the test of BoincTasks 1.80's Suspend at Checkpoint which does not work. Others confirmed that it may never have worked.