Rosetta@home ran out of work units lol

[RejZoR]

I've been doing Rosetta for a while, with some pauses in between, but I've never ran into situation that this project ran out of work units. I've fired up BOINC and I only have 2 WU's in process and none prepared ahead. Pressed Update and nothing. That's interesting and cool. We've done it. We've computed all of it. We finish around 700.000 units per 24 hours so those 1 million still in progress will soon be done.

 

Bakerlab, give us more to work on

[porina]

I set world community grid as one backup to rosetta, and had noticed I've only been getting WCG tasks today. Even if WCG isn't doing covid-related stuff, they can help in other areas. I still need my heating! (folding on GPUs only)

[Beowulff83]

lol, I have been running rosetta all the way up until the folding event now i switched all gears to F@H.  I was going to switch back my cpu to it when the event was over hopefully they are stocked back up by then.

[minibois]

/s

 

I'm totally kidding, just saw an oppurtunity and had to seize it  

It's so amazing seeing the tech world come together in such a huge way, supporting projects like F@H and Rosetta@H. Let's hope all this information can be put to use!

Oh and let's also hope we get more WU's 

[Godlygamer23]

Yeah, I just noticed I have no more Work Units queued up for Rosetta@home. lol.

[Cmdrd]

15 hours ago, porina said:

I set world community grid as one backup to rosetta, and had noticed I've only been getting WCG tasks today. Even if WCG isn't doing covid-related stuff, they can help in other areas. I still need my heating! (folding on GPUs only)

Same here, there is lots of valuable science to be computed in other projects as well, might as well make sure that this momentum can be carried forward. If you are wanting to prioritize one project over the other, change the resource share for each project in their respective account pages:

 

Rosetta@Home: "Rosetta@home preferences" -> "Resource Share" (will have to hit Edit at the bottom)

World Community Grid: "Settings" -> "Device Manager" -> "Device Profiles" -> Whichever device profiles you are using, probably just Default -> "Project Weight" (right at the bottom)

 

I have mine set to 1000 for R@H and 10 for WCG, the vast majority of compute time is spent on R@H but when my systems have started running short on R@H WUs it will start crunching WCG WUs.

 

Would be pretty awesome to be at a point with a lot of these projects to have multiple projects working at maximum capacity to have to output WUs, could be a huge boon to scientific research overall.

[PineyCreek]

I gave up on Rosetta@Home when I started getting at least 1/3 of my projects cancelled at random times when running ANY other project in BOINC (or F@H) that used the GPU.  R@H does not like sharing resources.  Perhaps it's my configuration.  I have zero issues running WCG projects and any other GPU project or F@H GPU folding simultaneously.

 

When I disabled all GPU folding the compute errors mostly stopped, but not completely.  I'm more productive keeping the GPU on so :: shrug ::  F@H has been (mostly) assigning me WU's so I'm sticking with that for now...only popping into BOINC when F@H sits for long stretches.

[RejZoR]

Rosetta has work to do again, mostly COVID-19 related apparently.

[yoerik]

1 hour ago, RejZoR said:

Rosetta has work to do again, mostly COVID-19 related apparently.

yep. It took a bit, but it's back, for now lol. They got so many new hosts, the queue drained.

 

We'll have to wait and see how long this fix lasts

[Cmdrd]

5 hours ago, PineyCreek said:

I gave up on Rosetta@Home when I started getting at least 1/3 of my projects cancelled at random times when running ANY other project in BOINC (or F@H) that used the GPU.  R@H does not like sharing resources.  Perhaps it's my configuration.  I have zero issues running WCG projects and any other GPU project or F@H GPU folding simultaneously.

 

When I disabled all GPU folding the compute errors mostly stopped, but not completely.  I'm more productive keeping the GPU on so :: shrug ::  F@H has been (mostly) assigning me WU's so I'm sticking with that for now...only popping into BOINC when F@H sits for long stretches.

How long is your queue set for work units? R@H has a lot of short deadlines so if you are keeping more than a day of work in your queue and not getting through it, R@H will be almost always high priority due to the shorter deadlines and you will ultimately fail to meet task deadlines. I dropped my total queue to half a day (0.25 and 0.25) and no problems with having cancelled tasks. Also R@H doesn't use the GPU, so not sure why you're having problems with GPU work units.

 

[PineyCreek]

11 hours ago, Cmdrd said:

How long is your queue set for work units? R@H has a lot of short deadlines so if you are keeping more than a day of work in your queue and not getting through it, R@H will be almost always high priority due to the shorter deadlines and you will ultimately fail to meet task deadlines. I dropped my total queue to half a day (0.25 and 0.25) and no problems with having cancelled tasks. Also R@H doesn't use the GPU, so not sure why you're having problems with GPU work units.

 

The issue has never been running out of time to complete work units, but rather units terminating abruptly with a Computing Error message.  This seems to have been alleviated by the new Rosetta 4.12 release, but I'm continuing to watch.  I pop Rosetta@Home on occasionally to see if the issue is better for me.

To further clarify, I don't have a problem with GPU units on BOINC or F@H.  The issue was that the Computing Error issue on Rosetta BOINC jobs occurred far more frequently if any GPU-based job was being performed (BOINC or F@H).  The GPU work units completed without issue (F@H, or GPUGRID on BOINC).