We just noticed that our research log forum thread has been viewed more than 50,000 times. Pretty neat!

Yesterday, Prof. Seth Marder from Georgia Tech gave a talk on “Organic Semiconductor Chemistry” as part of the Excitonics Seminar Series down the road at MIT.
http://www.rle.mit.edu/excitonics/12.13.12.marder.html

(Maybe this should be in the Suggestions/Feedback thread, but it's specific to CEP2, and it's related to Rob's last post above, so ...)
Sekerob: "Soon you'll have to update your tweak-sheet to add the up and coming 'app_config.xml' feature currently being tested with client 7.0.4x. It will allow seamless CEP2 crunching on a limited number of cores for maximum throughput/efficiency."
Sounds like one of the things I'd really like for crunching CEP2. I try to run CEP2 on just 1 core of my non-HT quad-core machines, and 2 threads of my HT 4- and 6-core machines, using micro-management. After 7 years' CPU time and 11,900 CEP2 results it's becoming a PITA.
Please keep us posted on this (in Member News?).

I also shut down my 45nm machines (Q9650 + QX9650) during peak electricity cost periods (2-8pm M-F, A$0.53 = US$0.56 / kWh), because they are less efficient than the newer 32nm devices. To ensure that no crunching time is lost due to the very sparse checkpointing of CEP2, I can't start up any new CEP2 WUs beyond the extra one I can sometimes queue up at bedtime.
I could fit in some more CEP2 crunching if there was a BOINC Manager function/button "Suspend at Next Checkpoint".
The vacated machine core would automatically move on to any WU that was Waiting to Run or Ready to Start, as per the standard Suspend function.
I'm sure this would also help many others who run WCG on machines that don't run 24/7 and who want to run CEP2 for you.

A "Checkpoint Now and Suspend" function/button would be better still, but as has already been explained, this has been found to be too difficult to implement for CEP2.

Just be forward and get 7.0.42 and add the app_config.xml to the wcg project folder with the lines [in case you want to run 2 on your Quad-HT

<name>cep2</<name>
<max_concurrent>2</max_concurrent>

See client config, app configuration section: http://boinc.berkeley.edu/wiki/Cc_config.xml

In the device profile set a value of tasks allowed to be in progress, just a little over the the cache size. To KISS the sample:

2 concurrent, cache 0.5 days [Minimum Work Buffer], average run time 8 hours would yield a setting of 3 on the profile, but to be sure if there are the interspersed shorties of 6 hours, set to 4 or 5.

On an octo, the rest will be filled with whatever other sciences you wish to run. The CEP2 tasks will age in the buffer, since never more than 2 are allowed to run, and always get first turn to start if a CEP2 finishes [FIFO with the breaks on]. Presently with a 3 day cache on octo laptop, 1 thread, have 10 in the profile, which is the sweat-point at this time to return the oldest in just over 3 days... stalest was 81 hours per WCGDAWS.

There's still some contemplation if/how to move the "in progress" limit to the app_config file at which point it could be used for any science. More on this later... sometime the new year... we're hopeful.

edit: correct tag typo, before the wrong one gets copied.

The Harvard CEP team wishes everyone out there a Merry Christmas and happy holidays! We hope you all get to enjoy a great time with friends and family.