From the front page at Rosetta if you haven’t seen it:
[i]February 17, 2006
Outage Notice: The project will be down starting today at 3pm PST for maintenance. The server should be back online later in the evening.
We will be updating the rosetta application today. There are a number of new features:
Work units will have a default cpu run time of 8 hours, and users will have the option to change the cpu run time as a project specific preference. The length of work units will no longer depend on the number of predicted structures. This option was added to allow participants to reduce bandwidth usage per work unit and maintain consistent run times.
Users will also have the option to change the frame rate and cpu use for graphics.
A new graphics version will be available for Mac OSX users. [/i]
Should be interesting to see the new default 8 hour run time WU’s.
“Work units will have a default cpu run time of 8 hours”
Does that mean that the WU will be sized to run in 8 hours, or that a WU will only be given 8 hours of run time per day? I’m not a Rosetta cruncher, but I’ve been thinking about it, so this is of interest.
Given the wide range of crunching capabilities from one machine to the next, they presumably can’t adjust the size of the WU so that it takes 8hrs to crunch. So I guess it will be a maximum crunch time (and then maybe timeout?).
I can’t find anything on their forums about it, but that doesn’t surprise me.
What you said makes sense, but overall the concept seems strange since you can set a project’s percentage of CPU time with the BOINC client. And if this is the case, then I wonder if it will cause problems if you are dual crunching Rosetta and something else (say SETI)? In a typical situation each project is given a percentage value of 100, which amounts to 50% CPU time for each. If the BOINC client gives time to Rosetta, but the Rosetta app isn’t crunching because the 8 hours is up, would you lose the remaining 4 hours of the 12 allotted to Rosetta? Hopefully, the Rosetta app can report to the BOINC client that it is suspended so BOINC can start the other project and not lose the time. Idle CPU time is bad CPU time.
Work units will have a default cpu run time of 8 hours
It will be the time that the work unit is actually executing which adds up to 8 hours, not the real time which might be longer if the work unit is suspended for a while.
At a guess you carry on crunching the work unit (and thus the number of predicted structures) in that time…
rather than having a set amount, you do 8 hours worth of structures…
The length of work units will no longer depend on the number of predicted structures.
Which to me means a variable length unit and the user reports back all completed work at the end of the crunch run. The 8hrs time looks like it is a user variable, so you could set it to a much smaller value if you wanted.
Don’t know how the back-end servers keep track of all the bits and whether they put them back together again.
On a different tack…major software changes on a Friday afternoon? :nono:
So, they will run a model as far as the donor will stand, regardless of expected results?
Will wait for the real explanation on the Rosetta boards.
Sounds like “player power” to me, and I don’t like it…
& echo the Friday afternoon thing, as unlike us they don’t work 7/7
:nod: exactly, I may just be being cynical, I hope I am.
Maybe there’s a better/more scientifically valid reason for it, not much news on the boards yet.
Anyway, one unit complete under Linux so it works…
The web changes are made, but I don’t think the new app is running yet. This is on the Roestta front page:
February 17, 2006
Work is flowing again. Today we upgraded our database server. Unfortunately, we will be delaying the application update for a day or two to work out a few minor issues. See Technical News for details.
and in the technical news:
February 17, 2006
Today we backed up our database and upgraded our production database server which now uses mysql-max and has improved I/O performance by using a SCSI controller serving a RAID10 of 14 drives. We will be delaying the release of the updated rosetta application due to some remaining issues with the cpu run time preference that are apparent on our RALPH test project. A fix will be made and tested immediately. You should still be able to set the new project specific preferences but they will not take effect until the application gets updated.
Friday afternoon strikes again
[edit]
This thread started in the Rosetta Science forum by The Major
Not so sure about these PRODUCTION_ABINITIO_INCREASECYCLES… Looks like they will take 4.5 hours to run with credit in 120 cobble ploppers range (of course X 4)
I just had to abort 8 of the production wu’s that were at their deadline. (on a AMD 3000+ that runs 24/7 with SETI @ 66% and Rosetta @ 33% and the swith between time at 120 minutes)
I’m still not understanding the new “cpu run time” thing
If I set it to 2 hours, and the wu is susposed to be an 8 hour unit, does it finish at the end of 2 hours of crunching, give me 1/4 credit and the rest of the wu wasted?
