New kind of app on Ralph

Message boards : Number crunching : New kind of app on Ralph

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4

AuthorMessage
Profile eiernacken1983

Send message
Joined: 30 May 19
Posts: 5
Credit: 27,668,118
RAC: 82,558
Message 102004 - Posted: 2 Jun 2021, 15:46:18 UTC - in response to Message 102001.  

I already had an app_config for setting a general task limit for Rosetta with <project_max_concurrent>....

Adding the limitation for the python tasks obviously confused the BOINC manager. It started requesting (and getting) a lot of additional tasks. The amount of tasks being ready for execution now easily exceeds my limit of 0.4 days. Did it on two out of three machines. Those two machines were flooded with new tasks. The third machine hasn't VM installed, so there was no need for changing app_config an it didn't get too much tasks.
ID: 102004 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Jim1348

Send message
Joined: 19 Jan 06
Posts: 881
Credit: 52,257,545
RAC: 0
Message 102005 - Posted: 2 Jun 2021, 15:47:50 UTC

Actually, I would set up a machine that did only the Pythons if they allowed me to select. And it would have enough memory to do the job.
But as it is, a lot of them probably fail. There appear to be some resends going out, but not much more.
ID: 102005 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile [VENETO] boboviz

Send message
Joined: 1 Dec 05
Posts: 1860
Credit: 8,152,819
RAC: 8,197
Message 102006 - Posted: 2 Jun 2021, 15:51:48 UTC - in response to Message 101998.  

My Threadripper is running some Rosetta python projects 1.02 (vbox 64):
The work units are now running for 2 days and 6 hours. Progress is 99.8 %, and further progress is advancing extremely slow.


That's strange.
I've crunched about ten python wus, with almost the correct runtime (4 hs "standard" wus, 6 hs python wus).
ID: 102006 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile eiernacken1983

Send message
Joined: 30 May 19
Posts: 5
Credit: 27,668,118
RAC: 82,558
Message 102007 - Posted: 2 Jun 2021, 16:02:54 UTC - in response to Message 102006.  
Last modified: 2 Jun 2021, 16:09:08 UTC

I had succesfull python tasks too running about 6 - 8 hours (target runtime 8 h). Now the first 2day+ tasks have exceeded their deadline. Running only 3 of them concurrently for the last 3 hours didn't speed them up.

I aborted one of them. Here is the result:
https://boinc.bakerlab.org/rosetta/result.php?resultid=1388271825
ID: 102007 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Jim1348

Send message
Joined: 19 Jan 06
Posts: 881
Credit: 52,257,545
RAC: 0
Message 102008 - Posted: 2 Jun 2021, 16:45:14 UTC - in response to Message 102007.  

Running only 3 of them concurrently for the last 3 hours didn't speed them up.
Did you reboot after reducing the number?
If you have "leave application in memory" enabled, the old ones could still be taking up memory space.
ID: 102008 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
biodoc

Send message
Joined: 19 Feb 06
Posts: 14
Credit: 30,298,003
RAC: 8,546
Message 102011 - Posted: 3 Jun 2021, 9:23:07 UTC

I wonder if the the "rosetta python projects" app is an implementation of PyRosetta for boinc.
ID: 102011 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Grant (SSSF)

Send message
Joined: 28 Mar 20
Posts: 1481
Credit: 14,589,068
RAC: 14,761
Message 102012 - Posted: 3 Jun 2021, 9:44:36 UTC - in response to Message 102007.  

I had succesfull python tasks too running about 6 - 8 hours (target runtime 8 h). Now the first 2day+ tasks have exceeded their deadline. Running only 3 of them concurrently for the last 3 hours didn't speed them up.

I aborted one of them. Here is the result:
https://boinc.bakerlab.org/rosetta/result.php?resultid=1388271825
Looks like there was a problem with that Task
Run time 2 days 14 hours 37 min 23 sec
CPU time                  6 min 20 sec
The CPU wasn't actually doing any processing of the Task.


This is from one you completed & Validated without issue.
Run time 4 hours 11 min 46 sec
CPU time 4 hours 11 min 39 sec

Grant
Darwin NT
ID: 102012 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile [VENETO] boboviz

Send message
Joined: 1 Dec 05
Posts: 1860
Credit: 8,152,819
RAC: 8,197
Message 102015 - Posted: 3 Jun 2021, 13:34:14 UTC - in response to Message 102011.  
Last modified: 3 Jun 2021, 13:35:00 UTC

I wonder if the the "rosetta python projects" app is an implementation of PyRosetta for boinc.


Old post (April 2020) from Rosetta admins:
We have a version of TrRosetta (a model much like the published version but also including PDB templates) that has been benchmarked and continues to be benchmarked on CAMEO as a hidden server. It undoubtedly performs better than the current prediction method used by Robetta for medium to hard targets, which is (Robetta) consistently the best performing server among public servers. We plan to add the latest protocols that will be tested in this coming CASP to Robetta in the near future and open it up to the public. This will not only improve the prediction quality for most targets, but will also significantly reduce the cpu computing requirements. We also are looking into the possibility of running these ML models and minimization modeling strategies to R@h which will make use of GPUs. This may require the ability to run python, Tensor Flow, and PyRosetta on BOINC clients.

ID: 102015 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
mmonnin

Send message
Joined: 2 Jun 16
Posts: 54
Credit: 20,058,207
RAC: 4,375
Message 102018 - Posted: 5 Jun 2021, 13:09:49 UTC - in response to Message 101995.  
Last modified: 5 Jun 2021, 13:11:29 UTC

... as well as being able to run more than once at a time. If the PC has memory.
But have you seen more than one running at a time?


For the 3rd time. YES. This is possible. I have seen it with my own eyes on my own PC. 64gb of memory on that PC and usage was not even 50%.
ID: 102018 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Jim1348

Send message
Joined: 19 Jan 06
Posts: 881
Credit: 52,257,545
RAC: 0
Message 102019 - Posted: 5 Jun 2021, 19:40:10 UTC - in response to Message 102018.  

For the 3rd time. YES. This is possible. I have seen it with my own eyes on my own PC. 64gb of memory on that PC and usage was not even 50%.

You haven't been reading my posts. I had a discussion of it.
And you probably did not see the part about "leave application in memory".
Who knows what state you are in.
ID: 102019 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Falconet

Send message
Joined: 9 Mar 09
Posts: 350
Credit: 1,017,068
RAC: 435
Message 102055 - Posted: 10 Jun 2021, 19:02:04 UTC

First official info on TrRosetta at Rosetta@home: https://twitter.com/RosettaAtHome/status/1403057126541447170
ID: 102055 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Billy

Send message
Joined: 29 May 06
Posts: 13
Credit: 1,494,571
RAC: 773
Message 102056 - Posted: 11 Jun 2021, 1:22:14 UTC

On my old iMac, I had 2 of these running at once with 2 regular Rosetta tasks. These new ones take up a lot of memory. As long as I was away for the day and the computer wasn't touched, they ran to completion. I had 2 others running at the same time and when I started to look at videos and do my email they sort of choked on the memory constraints and stopped themselves with an error message on Boinc Manager. When I shut everything off and restarted the computer to get them going again, they appeared to attempt to restart but failed. I don't remember now what happened after that.
ID: 102056 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile [VENETO] boboviz

Send message
Joined: 1 Dec 05
Posts: 1860
Credit: 8,152,819
RAC: 8,197
Message 102057 - Posted: 11 Jun 2021, 7:04:56 UTC

Yesterday they released the 0.21 version of Python app
ID: 102057 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile dcdc

Send message
Joined: 3 Nov 05
Posts: 1829
Credit: 115,451,918
RAC: 54,317
Message 102058 - Posted: 11 Jun 2021, 9:21:01 UTC - in response to Message 102055.  

First official info on TrRosetta at Rosetta@home: https://twitter.com/RosettaAtHome/status/1403057126541447170


This needs posting in "News"! I should get a chance to do that later today.

Does anyone know if there's a way to see which of your computers have Virtualbox installed without checking each one manually?
ID: 102058 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile [VENETO] boboviz

Send message
Joined: 1 Dec 05
Posts: 1860
Credit: 8,152,819
RAC: 8,197
Message 102087 - Posted: 17 Jun 2021, 9:06:08 UTC

I'm testing some 0.21 on Ralph and there is a problem
Can admins read this thread and say what i have to do?
ID: 102087 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Jim1348

Send message
Joined: 19 Jan 06
Posts: 881
Credit: 52,257,545
RAC: 0
Message 102089 - Posted: 17 Jun 2021, 11:21:28 UTC - in response to Message 102087.  

Can admins read this thread and say what i have to do?

Try this one:
https://boinc.bakerlab.org/rosetta/forum_thread.php?id=1000
ID: 102089 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile [VENETO] boboviz

Send message
Joined: 1 Dec 05
Posts: 1860
Credit: 8,152,819
RAC: 8,197
Message 102091 - Posted: 17 Jun 2021, 19:37:48 UTC - in response to Message 102089.  

Try this one:https://boinc.bakerlab.org/rosetta/forum_thread.php?id=1000


Thank you. But I posted this message on other Ralph thread, before reading your suggestion:

Some tests:
- i reduced the runtime from 6 hours to 2 hrs.
- i crunched correctly 3 wus (5323336, 5323342, 5323341) in 3 hrs each.
- i notice that the valid wus use about 8-10% of my system (a 12 cores). These use correctly a core per wus
- seems that "infinite" wus don't use cpu.

Now i will pass from 2hrs to 4 to see how these wus work

ID: 102091 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Previous · 1 · 2 · 3 · 4

Message boards : Number crunching : New kind of app on Ralph



©2024 University of Washington
https://www.bakerlab.org