Waiting for Memory

Message boards : Number crunching : Waiting for Memory

To post messages, you must log in.

AuthorMessage
AlanG7Mc

Send message
Joined: 26 Jun 06
Posts: 4
Credit: 867,054
RAC: 136
Message 97232 - Posted: 5 Jun 2020, 2:10:25 UTC

Waiting for memory

Some Rosetta Work Units get suspended on my machine with "Waiting for memory". 'Tricks' such as rebooting to clean things up sometimes work, but many times the problem returns later for the same WU. It's as if the WU can start out using 'small' memory, progress to the point where it needs 'medium' RAM, and further along needs big memory - which I simply don't have..

Yes, I have a dinky little Win 8.1 laptop with only 2 Gig RAM, but I'd still like to help, especially with COVID research.

Since a lot of CPU time (often 9+ hours) is used before problem occurs, all of it is wasted when I have to abort the work unit.

(Fortunately, World Community Grid also sends COVID WUs.)

With very great reluctance I am declining further Rosetta WUs - even if some them - the ones not jamming my RAM, would complete. Sorry.

I do get BOINC Notices from you, so if you have a fix or "magic" settings, I'll happily resume WU downloads. Thank you.
ID: 97232 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
sgaboinc

Send message
Joined: 2 Apr 14
Posts: 282
Credit: 208,966
RAC: 0
Message 97233 - Posted: 5 Jun 2020, 2:53:09 UTC

to avoid waiting for memory, there are few things that you can do
- use at most % of cpus, try to change that to a lower % e.g. 75% if you have 4 cores, 50% if you have 2 cores
this should reduce the number of threads running at the same time
- change the memory settings to use by boinc so that it is > 50% like 70% or higher
ID: 97233 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Erich56

Send message
Joined: 11 Jan 16
Posts: 35
Credit: 1,437,503
RAC: 0
Message 97234 - Posted: 5 Jun 2020, 4:25:46 UTC

how many cores does the laptop have - I guess 2 ?
So, how many tasks are running simultaneously? 1 or 2? I guess 1 only; with 2 GB RAM you might still run into memory problems even if you run 1 task only. Once a task takes up to 1GB of RAM, the remaining 1GB is certainly not enough to run all other system components.
So, in general, I doubt that a 2GB system is able to operate Rosetta without problems now and then.
ID: 97234 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
William Albert

Send message
Joined: 22 Mar 20
Posts: 23
Credit: 1,061,020
RAC: 0
Message 97235 - Posted: 5 Jun 2020, 4:35:04 UTC

If your computer only has 2GB of RAM, I would not attach it to the Rosetta@home project, as many WUs will use 1+ GB of RAM. Even if you could get the WUs to run, they would frequently be hitting swap, and it's unlikely that you would generate a quality result.

It would be cool if Rosetta@home would partition its application into low- and high-memory variants, so that the many machines without the memory to crunch the bigger WUs can still participate (and conversely, the machines with large amounts of memory can focus on the larger WUs). I'm not sure if that would be feasible, but just throwing the idea out there.
ID: 97235 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Erich56

Send message
Joined: 11 Jan 16
Posts: 35
Credit: 1,437,503
RAC: 0
Message 97236 - Posted: 5 Jun 2020, 4:38:39 UTC - in response to Message 97235.  

It would be cool if Rosetta@home would partition its application into low- and high-memory variants, so that the many machines without the memory to crunch the bigger WUs can still participate (and conversely, the machines with large amounts of memory can focus on the larger WUs). I'm not sure if that would be feasible, but just throwing the idea out there.
+ 1
ID: 97236 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 97247 - Posted: 5 Jun 2020, 17:19:20 UTC - in response to Message 97235.  

The project already makes an effort to ensure the host machine has sufficient memory to run the WU. Unfortunately, BOINC Manager does not seem to handle it well, and you have users willing to drop the entire project when they happen to notice BOINC Manager accommodating larger WUs.

If you created distinct applications for various memory footprints, you would also create an expectation for a continuous workflow for each memory footprint. But that isn't how the research works. If you are running a study of HIV, or COVID, these are large proteins that typically have a large runtime memory requirement. You, in that specific study, have zero need for any low memory WUs. The BakerLab and RosettaCommons researchers have many studies advancing concurrently. But, much of the area where new innovations and study are required are with larger proteins.
Rosetta Moderator: Mod.Sense
ID: 97247 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Number crunching : Waiting for Memory



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