1. Why does CPN say in their FAQs that project will be OS specific yet we should install the Oracle VM VirtualBox???
2. Since there is a chronic shortage of WUs one needs to add another BOINC Project if you want to keep contributing. CPN is such a long WU project compared to any other CPU projects the general settings do not make sense. Under Computing Preferences/Computing/Other there's two entries:

Store at least 0.25 days of work
Store up to an additional 0.5 days of work

It seems I should set additional days to a large number to try to catch a CPN WU. But that means I DL hundreds of WCG WUs. If I set it to 0.5 day for WCG will I ever DL a CPN WU???
3. The Server Status page does not indicate which OS WUs are for. It also says there 27 tasks ready to send at the top and at the bottom it says there are NO unsent WUs. I'd like to dedicate the cores on some of my rigs to CPN, the vast majority of which are Linux rigs. Any plans to make the Server Status page more informative???
4. Xeons go up to 22c44t. Does CPN use multiple CPU threads to make WU calculations or can it only use one thread per WU???
5. Has anyone used a Xeon Phi on CPN???

BOINC should change the Usage Limits from "Use at most 75% of CPUs" to "Use xx threads".

---

CPDN does not use VirtualBox, which is used by LHC@home and SourceFinder@home.
Tullio

---

Maybe if they used VirtualBox they wouldn't have to do OS specific projects.

---

1. While they suggest downloading boinc+virtualbox, virtualbox is not used in cpdn. Perhaps whoever put that up was thinking that it might be possible in the future, but nothing about going to virtualbox is in any concrete plans for the future (that I know).

2. It would be nice if one could go ahead and set the download extra work settings by project, but it is not to be. Otherwise one has to change those preferences periodically and pay attention to when new releases of models are for cpdn if one wants to get extra cpdn work without problems with other projects.

3. The 27 tasks have been a semi-permanent feature since cpdn upgraded to this version of boinc server software. It is meaningless. One has to look at the list of specific applications farther down on the server status page to see if there are actual tasks to download.

4. One thread per work unit. Maximum total model throughput (completions) occurs with multiple cores running multiple models, while the completion time for an individual model would be fastest if it was running as several threads in a multithreaded app. The decision was made to go for maximum completions rather than trying to finish an individual models from a batch very fast.

One problem. If you DL a large number of WU’s from WCG and try to run them alongside CPDN, if you set the work buffer for more than 1 day the Boinc manager will panic about the short deadlines on the WCG WU’s and run nothing else regardless of what the settings are. It won;t let CPDN run at all so long as there are unfinished WCG WU's present.

---

One problem. If you DL a large number of WU’s from WCG and try to run them alongside CPDN, if you set the work buffer for more than 1 day the Boinc manager will panic about the short deadlines on the WCG WU’s and run nothing else regardless of what the settings are. It won;t let CPDN run at all so long as there are unfinished WCG WU's present.

The only way I have found around this is to micro manage BOINC and suspend some WCG or other non CPDN tasks manually. This is a pain but is what I do on the machine that I use on a daily basis. I fiddle a bit less often with the others.