I have successfully written an RNN algorithm and used the galib to evolve it (420 weights). When I use 15 robots, the process is killed in the 7th generation. But when I use 25 robots, the process is killed in the 4th generation. And I try to know why my process was killed by
Code: Select all
dmesg | egrep -i -B100 'killed process'
Code: Select all
[232556.977741] oom-kill:constraint=CONSTRAINT_NONE,nodemask=(null),cpuset=/,mems_allowed=0,global_oom,task_memcg=/,task=rnn_pathformati,pid=61766,uid=1000
[232556.977751] Out of memory: Killed process 61766 (rnn_pathformati) total-vm:5096852kB, anon-rss:4877928kB, file-rss:0kB, shmem-rss:0kB
Do you have any suggestions? The cause of this phenomenon is insufficient memory, so should I use another computer with a higher memory to run it, or is there some deficiency in my code?