I can't speak for the older Core 2 based Xeons (I retired mine years ago), but your e3 and e5 times look like you're running every thread. On LLR, you want to either turn off HyperThreading or set Boinc to use only 50% of cores. Your 2670 should run around 1.1-1.2ks/task (assuming you are properly cooled and there is no thermal throttling) and the 1230 even less than that (700-800).

Also, I see you have a super large number of cached tasks. Around here, setting boinc to 0 day cache=more chance of being primary prime finder.

Hey, I got this in my event messages:

444470 PrimeGrid 13/04/2017 1:20:27 PM Not accepting requests from this host

What is this about? Previously I was able to get work on this machine, but as of recently I am unable to

Short answer: I banned four of your hosts.

Longer answer: There are more than the normal amount of malfunctioning hosts running TRP Sieve jobs. They had been so "productive" that we had a half-million results that errored out in about 3 seconds, at which point the host would grab another task. So, for any computer that showed 50 or more TRP Sieve results that ran but errored out in less than 10 seconds, I've blacklisted that host. I haven't been looking closely at who I banned, figuring that anyone who was a normal user would eventually notice and say something.

Running a quick query, your hosts 610371, 610487, 613080 and 613537 are among the 1690 total hosts banned. I'm allowing them again, but if they get above 50 TRP Sieve tasks where they show a nonzero runtime but the task errored out, they'll automatically get banned again. I don't have the ability to ban by subproject, it's an all or none setting.

As to why I started doing this - we were getting into trouble on the database. The result table had grown so large that queries were taking too long. It was still manageable, but we couldn't let it get any worse. If you look on the Server Status page at the strip chart right below Filesystems, you can see how we had 3.5 million results and that's now dropped to under 3 million. That makes a huge difference in how well the system works.

I apologize for the inconvenience to you, but we had single hosts with 70,000 bad results apiece. Most had errors where they couldn't download the sieving program and so were never going to work. I had to pick a threshold and chose 50 bad results. You're the first one to notice and say anything about it. Chances are I'll unban all the hosts after TRP Sieve is completed so that nobody else is in this situation.

444470 PrimeGrid 13/04/2017 1:20:27 PM Not accepting requests from this host

I see on the status page, we're upto 83P, so what is the sieving depth that we are doing for TRP again?

Our most recent estimate is 206.