Message boards : Number crunching : Rosetta 4.83
Author | Message |
---|---|
Rossmor35 Send message Joined: 24 Sep 05 Posts: 4 Credit: 84,870 RAC: 0 |
Does the new version of Rosetta allow you to do away with leaving the Rosetta app in memory when switching to work at a another project after 1hr? |
Fuzzy Hollynoodles Send message Joined: 7 Oct 05 Posts: 234 Credit: 15,020 RAC: 0 |
Does the new version of Rosetta allow you to do away with leaving the Rosetta app in memory when switching to work at a another project after 1hr? NO! Don't do it. Let it stay in memory. Over at Ralph, they are working with Windows Application 4.93, and this one seems stable to the "not be left in memory while preempted" problem, but it hasn't been implemented here yet. You can follow the development in this thread: http://ralph.bakerlab.org/forum_thread.php?id=4 I have reported crashed Rosetta WU's due to this problem myself here. So let Rosetta WU's stay in memory while preempted. I'm sure they'll let us know when it can be changed, as Rom is working hard on the problems. [b]"I'm trying to maintain a shred of dignity in this world." - Me[/b] |
Divide Overflow Send message Joined: 17 Sep 05 Posts: 82 Credit: 921,382 RAC: 0 |
I beg to differ. The recent Rosetta application change includes a fix for removing a WU from memory that was already successful on RALPH. Check the Technical News page here for details. https://boinc.bakerlab.org/rosetta/rah_technical_news.php If you are interested in removing a WU from memory, try giving it a shot and see how it goes. I'd pay close attention for a little while to confirm that the fix is good! (The continued application changes for RALPH are zeroing in on the 1% bug, among other fixes.) |
Fuzzy Hollynoodles Send message Joined: 7 Oct 05 Posts: 234 Credit: 15,020 RAC: 0 |
I beg to differ. The recent Rosetta application change includes a fix for removing a WU from memory that was already successful on RALPH. Check the Technical News page here for details. Thanks. I wasn't aware of this, as I've followed the threads about this problem both on Ralph and here. I've emptied my cache due to a hardware upgrade tomorrow or Monday, and after this I'll certainly give it a try. It is after all this, I signed in over at Ralph for. :-) [b]"I'm trying to maintain a shred of dignity in this world." - Me[/b] |
doc :) Send message Joined: 4 Oct 05 Posts: 47 Credit: 1,106,102 RAC: 0 |
remember that you still will loose all work done after the last checkpoint if you remove it from memory, even if it does not crash anymore :) |
Fuzzy Hollynoodles Send message Joined: 7 Oct 05 Posts: 234 Credit: 15,020 RAC: 0 |
remember that you still will loose all work done after the last checkpoint if you remove it from memory, even if it does not crash anymore :) Yes, and because of this, it's a good idea to set the switch time between the projects to at least 2 hours. But I thought they had overcome some of those problems with the checkpoints in the beginning, where the checkpoints were very scarce? Now it seems like there're checkpoints quite often. [b]"I'm trying to maintain a shred of dignity in this world." - Me[/b] |
doc :) Send message Joined: 4 Oct 05 Posts: 47 Credit: 1,106,102 RAC: 0 |
there are more jumps early in each run now, like 1.02 and 1.03, but it still can take a long time to reach the next model. i did not really watch that close yet though to be sure if there are much more checkpoints or not :) |
R/B Send message Joined: 8 Dec 05 Posts: 195 Credit: 28,095 RAC: 0 |
Should I go and change my 'leave in memory' setting to 'no' now? Would it matter? Thanks in advance. Founder of BOINC GROUP - Objectivists - Philosophically minded rational data crunchers. |
doc :) Send message Joined: 4 Oct 05 Posts: 47 Credit: 1,106,102 RAC: 0 |
"should"? my opinion, no, cause you will still loose everything that happened since the last checkpoint. but you can remove it from memory without crashing the WU now if you want. :) |
Message boards :
Number crunching :
Rosetta 4.83
©2024 University of Washington
https://www.bakerlab.org