There is no real difference in speed for an application program whether it is managed by the UD client or the BOINC client. Some other factor is at work.

We'll verify with the technicians, as the colour switch of a protein combo was thought to be the point of saving work. See Here . Power or Reset switch operation (how do you 'reboot'?) will with great probability prevent the progress status files being stored properly.

UD Monitor is an unsupported 3rd party utility and although it's proven it's value to many in the area of buffering work similar to BOINC, it has not been maintained for a long time (last beta rev feb.'06) and may not 100% work with the newer sciences and 'could' even impair the proper functionality of the AGENT-DAEMON-SCIENCE interaction. We don't know as WCG has never endorsed it!

As an aside, UDmon is also responsible for the large individual distortion, for all i know, of member device count inflation as each slot represents a device. 12 slots is 12 devices has been my experience. There are other pros and cons to the use. I've fired it up, see screen-print below and monitored for a while with the 'Last Write' action selected. It generated this record for the 5th save "07-02-20 14:48:49: #5: Save (+ 1:39:26)". Is a boon to know when to stop the agent, before closing down windows the proper way for the UD users, even if the buffer function is not used.

Question: How does one get the Client Log to work. It was always blank when running it, whatever options selected. Does it require installation in the WCG agent directory (where it is installed now)?

HCMD has had some bugs, and may be shut down part of the time.

If your computer needs to be shut down, it may be preferable to pick a project that saves a lot of checkpoints, so that little will be lost each time. Total backoff to 0% is a symptom of not being able to complete enough each time before the computer is shut down.

On the other hand, I just let my computers crunch all day and all night, 24/7. Checkpoints cause no problems at all.

I have used UDMonitor in the past, but found that it required a lot of babysitting. (note to new users: since the default is grid.org, you have to make sure to use wcg.bat to start it for these projects.) BOINC is much easier to deal with. You can basically set it to however many days of work you want and forget it, unless you want to watch. Since more and more projects, including these, now have graphics, you don't have to give up anything by running the BOINC version.

I have had the same problem several times, which is especially frustrating on long computations such as Human Proteome Folding 2, a project I would like to participate in.

I finally dropped out of Human Proteome Folding 2 due to MANY (>100) hours lost.

I have also limited the days/hours that WCG runs so that my SETI/BOINC client has higher priority. I do not experience hours lost on SETI.

I would suggest that the real problem is the lack of a requirement on researchers to produce user friendly software applications. We should not have to go to graphs or other application software to second guess when a backup is/has occured. Further, I would suggest that such a requirement be placed on research software; i.e. backup should be automatic after some reasonable time period, like every 1-2 hours. I can afford to lose 2 hours of computation time. As volunteer clients, we have a number of choices of projects where we can donate our unused computer resources; i.e. this is a competitive business - it would be well worth WCG's time to implement auto backup on all current/future projects.

Other than this complaint, I do like the WCG interface and am happy I can help the researchers.

cfpops

Hello cfpops,

I can understand very well your frustration. I experienced the "restart from 0%" problem several times since end of January. I noticed also that "snoozing" the task before login out or reboot does not really help.
I had also very very very long computation especially for the first task I received (over 300 hours).
Being honnest, I did experience such problems not only with HCMD but also with HPF2. In between, the situation is better and the computation activities make significant progress.

I agree with your statement regarding the necessity to have proper design of both software agent as well as task in order to not waste "volunteer" computation time (and given money: electricity cost).

Because, I am strongly convinced that working on such research programs is important for supporting medical research, I can only encourage you to support further such activities and to still participate to medical research projects.

Sincerely,

Yves