Rosetta is using > 700MB of RAM, which is all the available RAM on my laptop. I tried to change the preferences to use only 30% of my RAM, but it's ignoring them. How do I fix this? I had to disable BOINC because it's making my computer unusable.

---

Rosetta is using > 700MB of RAM, which is all the available RAM on my laptop. I tried to change the preferences to use only 30% of my RAM, but it's ignoring them. How do I fix this? I had to disable BOINC because it's making my computer unusable.

I figured out what part of the problem was, I need to update with Rosetta to get the new prefs; not just with BAM.

But now I have a new problem, all of my work units are getting a Computation Error. Is this because Rosetta can only run if it has > 500MB of RAM for it to use?

---

Hi raccoonone

I'm suprised your're having WUs error out on you - the error is:

<core_client_version>5.8.16</core_client_version>
<![CDATA[
<message>
Maximum memory exceeded
</message>
]]>

However, I'm running rosetta on a number of PCs that have 340-448MB RAM and they don't error out like this. You have had one success today, but then another failure after too. Have you allowed Rosetta to use more than 30% of RAM when the computer isn't in use? It may need more RAM than this at some point before it will attempt the jobs.

---

Hi,

I noticed the same thing, the memory requirements of gp04_FOLD_and_DOCK is way off the usual limits (600-700K per workunit). I noticed it when my dual core system slowed down when Rosetta started to use most of my 2GB RAM. I do not know if this is an expected behavior of these type of WUs or rather a bugged memory leak for the latest client version, but it needs looking at in my opinion.

from Rhiju in another thread:

"Hi Everybody:
Sorry for checking in a little late on this thread. I'm a bit puzzled that the FOLD_AND_DOCK_SUBSYSTEM workunits are taking up so much memory, but I've canceled all those jobs, and won't send any more out until we reduce the memory requirement! Apologies! Thanks for posting so quickly about the problem. It wasn't apparent on ralph.

Also: if you have one of these workunit in your queue, please feel free to cancel it rather than risk a system slowdown due to virtual memory problems."

---

I also had two of them today, trying to use ~850 Megs of memory each. Result they both were suspended with the message Waitting for Memory and QMC running slower on the other core.

If BOINC would have aborted them to free up the memory all would have been OK and run other units.

That is a change to send to the BOING development team to see if they could impliment it. But if they did something like that many people running with smaller amounts of memory could have to many aborted. Looking at my results it looks as if I waisted very little time on these, but would have stayed there with them not running if I did not chack that system, which I ofter don't for a coupple days at a time. I was luckey this time.

EDIT
Since that system is always on and not switching from one project to another (always has Rosetta on one core and QMC on the other) I set the preferences not to keep in memory when preempted.