What kind of machine is it taking 30+ hours to run on? What kind of processor, memory, etc...
I have an older dual processor AMD Athlon machine that I am on right now, and the FAAH jobs have been stating the "To Completion" times of 20+ hours, but the jobs usually finish in around 12-15 hours. If it is an older machine you are using, you might want to consider adding in your preferences the Rice project that just opened in mid May. Those jobs run 8 hours no matter what. Other than the Rice jobs your best bet would be to run the DDDT jobs. Those seem to finish quickest on my oldest machines. The FAAH and HPF2 jobs are monsters on my older computers. I just recently lost a pair of Pentium 4s and haven't had the time to try to bring them back to life, and those were my oldest.
If you are running an older P4 or even P3 processors, you should probably stop and weigh how efficient the machine is (i.e. how much power it is consuming) versus how much work is getting done. Every little bit helps, but it's probably not the "green" thing to do for some of these older machines.

Hello DGA Director,
'processor percentage' is a heat control measure for laptops. It decides how many seconds (out of 10 seconds) the CPU will run. 50% will run 1 second on, 1 second off. 75% will run 7 seconds on, 3 seconds off. I use a desktop so I run mine at 100%.

Lawrence

If the below graph is a reference, take your mean run time and multiply by the average change between May 23 and a few days ago. According the data job this went from 8.035 hours and topping out at 12.041 hours i.e. 50%. So a machine doing normally 20 would have them doing about 30 hours.

The problem with long computation times is strictly related to the complexity of the ligand to be docked into the HIV protein.

Generally, ligands crunched with FA@H tend to be "hits", potentially active molecules that can be further optimized and made more complex to become a lead compound (the grandparent of the final drug). These hits does not require too much time due to the relatively simple molecular structure.

Some of current experiments involves ligands bigger than usual hits, and most likely the long times you noticed are related to them. The molecules showed an interesting activity, then the reason to further investigate on their interactions. Unfortunately small increase of structural complexity can easily lead to big increase in computational time.

Concerning the errors, it would be interesting to have some information on jobs id's so we can try to figure out which issue is responsable of calculation failure.


Hope this helps,


Stefano

Was it just one result? If you have many "Error message: "too many normally harmless exit(s)".", time to look if something is interfering, security software the most obvious item to look at.