*** For details on how to be removed from this list visit the *** *** CCP4 home page http://www.ccp4.ac.uk ***
On Nov 1 2006, Eleanor Dodson wrote:
How can I stop PHASER grabbing ALL the memory and destroying my machine!! Eleanor
There's a section on our FAQ page (http://www-structmed.cimr.cam.ac.uk/phaser/faq.html) that discusses how to reduce the memory requirements. One point that isn't yet addressed is that keeping track of a large number of partial solutions in the automated search also consumes memory. We advise that, if the first run using default parameters doesn't work, you might want to run another job relaxing the criteria for accepting a partial solution -- but I've seen some users go overboard. In particular, it's never a good idea to allow too many clashes.
Usually when Phaser runs out of memory, it fails with a memory allocation error without making your machine grind to a halt. But I'll add something to our FAQ relaying the suggestions about limits, for anyone who is worried that Phaser may interfere with other processes.
There are tradeoffs at work. To allow the automation features that are one of the advantages of Phaser, more information has to be stored. By using C++, we can use dynamic memory allocation so your job will run if there's enough memory on your computer, without having to recompile or rerun after setting some mysterious environment variables. But then a Fortran program is unlikely to expand to fill the memory of your computer.
Nonetheless, we run a lot of test jobs, and we rarely run into memory issues. Eleanor, what exactly is the job you're running that is giving you such grief?
Randy Read
