This week was a bit of a bummer. Starting Monday I ran 10 programs set, meaning that I ran the genetic algorithm program, the program to test the individuals and one to visualize and help test the individuals. I run these program on 30 different machines and for what ever reason some of the runs stopped, others went on to some point then died. The bulk of the week was hunting down the problem. On Tuesday I thought that I found the problem, but it turned out to be the herald to the actual culprit. In the end on Thursday, my mentor and I saw that when the program used 40-ish or more threads the machines load would reach a max and anyone else who ran a program on that machine would over load my program. It was no ones fault, just a discovery of a limitation. The runs that where stopped still had good data and It would have been a shame if it all had to go to waste, so the second big project for the week was making a plan B, if a run did stop in the middle of running I could call this class and the run would continue with little change. The only thing that would not be available would be the time it takes. The run saves all the important data, and the time it takes is not quite as pertinent. other then that this program would/should pick up right where the last computation left off, making the runs usable.
|
CategoriesLee Curryjournal about my summer |