Rosetta@Home: Got my hopes up

[Insidious]

David Baker's journal

Lots of neat stuff coming up this week!

Rhiju, Bin and David K. think they have the watchdog and checkpointing all working properly. This means
(1) No stuck jobs ever!!
(2) Little time wasted when Rosetta is taken out of memory--process resumes in the middle of whatever structure was being calculated.

I'm excited to see the results of jobs running now testing the improved sidechain sampling. The outcome should be clear by next monday/tuesday. Rhiju will then be testing an exciting combination of the "jumping" protocol invented by Phil Bradley here, the star of CASP6, with fullatom refinement.

Will Sheffler, a graduate student here, has developed a "smoother" version of the energy function which we hope will make possible the finding of deep minima from further away. We should be able to start testing this next week as well.

Other good news is that Microsoft has generously agreed to cover Rom Walton's consulting fees. This means that Rom will be back soon to put in more robust backtracing on ralph and rosetta to allow him and us to track down the remaining access violation errors. He will also fix the 0 credit win98 problem as promised some time ago. He is deeply involved in getting the latest boinc release ready for prime time, so it may be a little while before you see him back in action here.

I have never stopped being impressed at Dr. Baker's candor and appreciation for those of us who participate in Rosetta@Home.

It sounds like there is realistic hope that our babysitting days may be comming to an end.
With all the positives of this project... (Great Science, Congenial Project Developers, HONEST feedback from the project when there are issues or news to share, etc.) Add on a client that behaves itself... I hope so!

-Sid

 

[TAandy]

I'll second that Sid :beer:

 

[Fardringle]

That's good news. I hope it will work. I've had at least one stuck machine every day since I started doing this project in March.

 

[Insidious]

I'm going to keep watching pretty closely for a while. (I still have some 5.01 and 5.06 version Work Units in my queues)

I think it's the 5.07's that Dr. Baker is talking about. If you find one of those going nutz, please be sure to post it so we will know not to let our gaurds down.

:beer: (<-- Cheers!)

-Sid

 

[BadThad]

Fingers are crossed!

 

[RobertE]

Really have to hand it to the Rosetta crew. They by far are the best project as far as communication and interaction with the crunchers. So much so, I think if you were to take all the other projects combined, they would still not touch Rosettas interactions. Kudos to them! :beer::thumbsup:

Other projects could and should (Predictor) take a note for the Rosetta book on how to keep people involved.

 

[biodoc]

So far so good with 5.07

Thanks for the info Sid!:thumbsup:

 

[petrusbroder]

5.07 is also running good on my Mac - no problems so far.

 

[Dalephi]

Today is the first time in a long time I didn't have at least one stuck work unit. All of my computers should be on 5.07 by now.

 

[JWMiddleton]

Originally posted by: biodoc
So far so good with 5.07

Thanks for the info Sid!:thumbsup:

Is download of the new client automatic? Or, do I need to do something?

Yesterday I found a stuck WU sitting at 2.12% for 120 hours, last week I had one that wasted 80 hours. I sure hope this round fixes it.

 

[biodoc]

Originally posted by: JWMiddleton

Originally posted by: biodoc
So far so good with 5.07

Thanks for the info Sid!:thumbsup:

Is download of the new client automatic? Or, do I need to do something?

Yesterday I found a stuck WU sitting at 2.12% for 120 hours, last week I had one that wasted 80 hours. I sure hope this round fixes it.

Yes, download of the new client is automatic.

 

[The Reaper]

I've been hit pretty hard with stuck WU. 5.06, and now the 5.07 clients have been running very smooth so far.