markhl
09-03-2002, 01:20 PM
On the protein that is now running, my system can do almost 9000 structures in the first hour. Overnight, it averages only 7600 / hour. Over the long weekend (89 hours CPU time) it averaged 7160 / hour.
I saw the same effect with the previous 69-residue protein. Speeds varied from 5800 / hour for a short run to 5200 over a weekend.
Again, this is Windows XP & I am running as a service with the -rt switch. 1.8 GHz system.
This may be related to the memory leak, see that thread. If Howard fixes that issue, perhaps the speed will also hold constant?
Mark
I saw the same effect with the previous 69-residue protein. Speeds varied from 5800 / hour for a short run to 5200 over a weekend.
Again, this is Windows XP & I am running as a service with the -rt switch. 1.8 GHz system.
This may be related to the memory leak, see that thread. If Howard fixes that issue, perhaps the speed will also hold constant?
Mark