I'd like the project to add an ASAP (As Soon As Possible) setting for the target CPU time, please. Technically this would be called ASAP and numerically would specify a time of 1 sec.

The effect would be that all running results would terminate gracefully at the earliest opportunity - at the end of the current model.

Any unstarted result would run for a single model only, and would thus also terminate as soon as possible given that they are gong to do anything at all (if the user doesn't want anything at all, the current abort facility is fine).

The use would be that when someone wanted to run down Rosetta in a hurry, they'd set that time setting, and update the project.

For someone with several boxes, like me, I could set one venue to this setting, and when I wanted to run that box down, just change its venue.

In my case I'd like to do this as I run LHC as my main project on many boxes, with Rosetta to fill in the frequent gaps in work on LHC. When LHC work comes I put Rosetta into "No New Work" and it would be nice to be able to encourage it to finsih of the work held more quickly.

It may seem counter-productive to allow people to crunch less for this project. I'd suggest this is a short term view and in the long term the opposite effect happens.

For one thing, at present I tend to abort any unstarted Rosetta work, and with a 3 day cache and 24-hour run times that means that 2/3rds of the results held get sent back unstarted whenever LHC work appears. With an ASAP option I would be likely to let the results run through one model each

Secondly, there will be, in my opinion, a growing need for projects to fill the gaps made by other projects. Gaps can be expected and frequent (like LHC), expected and occasional (like if there turns out to be a gap between s4 and s5 data on Einstein), and unexpected (like project downtime). The more friendly you make Rosetta for people seeking a "second priority" project, then the more of that time you will get.

Whether you take up this suggestion or no, thanks anyway for being already the friendliest project for those seeking fill-in work.

---

This would also be useful for the person (been there!) that has a multi-day cache (I've seen some hosts with 20 WUs on board), and changes their preference from the default to, say 24hrs. And then when you update to the project to bring down the change, it grabs more WUs, but it doesn't fully realize yet that the WUs are going to run for so long, and so it pulls down 10 times more WUs than you need for your cache.

River~~ for your case, why not just set up your "end gracefully" venue with the minimum runtime preference? Don't they have it down to an hour now? So, that would give you your one model & finish up scenerio. At least on most WUs. And even if they crunch 3 or 4 models in that hour, it's still very close to what you are asking for.

You should also be able to achieve your objective by giving Rosetta a low resource share. That way it will only pull down one WU at a time, and once a project with higher share that's been deprived (due to their own inability to produce WUs) BOINC will go back to running that. I mean you REALLY would prefer to establish a project priority right? (something BOINC doesn't support) Rather than resource share?

And if that single R@H WU gets started, and interrupted by your other projects, BOINC will eventually see the R@H WU has a deadline looming and finish it up. But it would just be the one WU. So, one WU could be your fill-in for more than a week. And if other projects have long outages, BOINC will get a second WU from Rosetta and continue to tally debt to your favored project, so it gets more crunch time later when work is available.

Try setting R@H to around a 10% resource share in the venue where you wish to give preference to another project.

---

Add this signature to your EMail:
Running Microsoft's "System Idle Process" will never help cure cancer, AIDS nor Alzheimer's. But running Rosetta@home just might!
https://boinc.bakerlab.org/rosetta/

This may be a bit off topic perhaps. But I try not to make new threads. Why have none of my machines downloaded the latest Rosetta client? One d/l a new unit a few hours ago but it's still the old version. Furthermore, one unit seems to be sticking and the messages say it is operating on the earliest deadline first premise. Mainly though I'm interested in why my machines haven't d/l the newest version yet.

---

Founder of BOINC GROUP - Objectivists - Philosophically minded rational data crunchers.

This may be a bit off topic perhaps. But I try not to make new threads. Why have none of my machines downloaded the latest Rosetta client? One d/l a new unit a few hours ago but it's still the old version. Furthermore, one unit seems to be sticking and the messages say it is operating on the earliest deadline first premise. Mainly though I'm interested in why my machines haven't d/l the newest version yet.

Are you running a large queue? My machines connect about every 2 hours and the 5.07 application came down within 4 hours of the deployment with the first new Work Unit that arrived. However, my queue was empty. In any case you will get it sooner or later, because the 5.06 application is no longer being sent out, and eventually you will run out of 5.06 work.

---

Moderator9
ROSETTA@home FAQ
Moderator Contact

This may be a bit off topic perhaps. But I try not to make new threads. Why have none of my machines downloaded the latest Rosetta client? One d/l a new unit a few hours ago but it's still the old version. Furthermore, one unit seems to be sticking and the messages say it is operating on the earliest deadline first premise. Mainly though I'm interested in why my machines haven't d/l the newest version yet.

Are you running a large queue? My machines connect about every 2 hours and the 5.07 application came down within 4 hours of the deployment with the first new Work Unit that arrived. However, my queue was empty. In any case you will get it sooner or later, because the 5.06 application is no longer being sent out, and eventually you will run out of 5.06 work.

Well, the machine I'm on now is 5.07 but the machine in question is still 5.01. It is a dialup and I just manually turned in another completed unit. Since it is on dialup I naturally have it set to 24 hour crunch time. It's not sticking now apparently. I was just curious because about 12 hours ago it d/l a casp 6 unit under 5.01. Doesn't really matter to me I suppose as long as it's doing 'its thing'. Thanks, Mod9.

There's a 3rd machine but I haven't checked it yet. I can look at that one later.

---

Founder of BOINC GROUP - Objectivists - Philosophically minded rational data crunchers.

This ASAP setting would be helpful for Ralph testing too. Once a new version comes out, it would be a graceful way to transition to the new version. Or you could set your Rosetta work to complete so you have time to crunch Ralph.

---

Add this signature to your EMail:
Running Microsoft's "System Idle Process" will never help cure cancer, AIDS nor Alzheimer's. But running Rosetta@home just might!
https://boinc.bakerlab.org/rosetta/