TheWeatherMan
08-21-2002, 11:53 AM
What's New Version 1.9.4
Download Page Changes.
Changed code that reads frame times.
Added some "crude" frame prediction.
Fixed a few tracking bugs.
I am now posting two zip files when I update. One will have only the files necessary to update your current copy of EM III. Always read the link carefully to see if your current install can be updated this way. It may not work properly otherwise.
First thing is that this release does not fix the frame tracking progress bar problem. But, the good thing is that unless a frame takes over 1 hour and 49 minutes, you shouldn't see a problem. I am running a bunch of tests before I add that fix to EM. I want to make sure that even if it requires one week per frame that EM can track it. I hope we don't see things like that any time soon. Vijay understands the problems that these very long frame times cause to the user, and after PM'ing each other about this problem, I doubt we will see frame times this long again. This also may be the end of the 10 frame work units. Since even on the fastest folding system the frames were around an hour each, this is still a lot of computing time to lose if one has to shut the system down, thus shutting the client. So, with the change to 100 frames on the p180 work unit this problem should go away. The new problem this introduced was picking the correct frame time as before this, all proteins with the same name had the same number of frames. EM can now read this info properly from your csv file.
I'm just beginning to write the code that will "predict" frame times based on what the "averages" say they should be. Currently it is "very" crude. Don't expect it to be correct or even close. 1.9.5 release will hopefully be better. It can never be perfect as each protein is different even if they have the same name, and some more than others.
There were a few more minor tracking problems that rarely showed up, but I'm slowly getting this tracking code back in order. Since the introduction of the Gromacs core, it has been difficult to "reuse" the tracking code that has been good for a long time because there are some differences, and some are radical. In this version I haven't seen any problems over the last day or so. Hopefully it is back to being fairly stable.
Get it here: http://www.em-dc.com
http://www.em-dc.com/images/latestem.jpg
Download Page Changes.
Changed code that reads frame times.
Added some "crude" frame prediction.
Fixed a few tracking bugs.
I am now posting two zip files when I update. One will have only the files necessary to update your current copy of EM III. Always read the link carefully to see if your current install can be updated this way. It may not work properly otherwise.
First thing is that this release does not fix the frame tracking progress bar problem. But, the good thing is that unless a frame takes over 1 hour and 49 minutes, you shouldn't see a problem. I am running a bunch of tests before I add that fix to EM. I want to make sure that even if it requires one week per frame that EM can track it. I hope we don't see things like that any time soon. Vijay understands the problems that these very long frame times cause to the user, and after PM'ing each other about this problem, I doubt we will see frame times this long again. This also may be the end of the 10 frame work units. Since even on the fastest folding system the frames were around an hour each, this is still a lot of computing time to lose if one has to shut the system down, thus shutting the client. So, with the change to 100 frames on the p180 work unit this problem should go away. The new problem this introduced was picking the correct frame time as before this, all proteins with the same name had the same number of frames. EM can now read this info properly from your csv file.
I'm just beginning to write the code that will "predict" frame times based on what the "averages" say they should be. Currently it is "very" crude. Don't expect it to be correct or even close. 1.9.5 release will hopefully be better. It can never be perfect as each protein is different even if they have the same name, and some more than others.
There were a few more minor tracking problems that rarely showed up, but I'm slowly getting this tracking code back in order. Since the introduction of the Gromacs core, it has been difficult to "reuse" the tracking code that has been good for a long time because there are some differences, and some are radical. In this version I haven't seen any problems over the last day or so. Hopefully it is back to being fairly stable.
Get it here: http://www.em-dc.com
http://www.em-dc.com/images/latestem.jpg