Closed ildyria closed 9 years ago
Proof of memory leak : 1 possible move, tree that takes 2 Go...
Build : x64, int depth = 20, // depth of the maximum search int timelimit = 10000, // time limit for the simulations int simulR = 1000000, // number of root simulations int simulL = 1, // number of leaf simulations int numVisitExplo = 5 // number visit required before explorating a node
Try and display text in constructors and destructors of your objects, to see which ones are not deleted. Or use shared pointers, that should work too
----- Mail original ----- De: "ildyria" notifications@github.com À: "ildyria/Arimaa" Arimaa@noreply.github.com Envoyé: Vendredi 26 Décembre 2014 20:14:05 Objet: Re: [Arimaa] Possible memory leak... (#3)
Proof of memory leak : capture d ecran 304 1 possible move, tree that takes 2 Go...
Build : x64, int depth = 20, // depth of the maximum search int timelimit = 10000, // time limit for the simulations int simulR = 1000000, // number of root simulations int simulL = 1, // number of leaf simulations int numVisitExplo = 5 // number visit required before explorating a node
— Reply to this email directly or view it on GitHub .
Memory leak should now be solved on the single core version. \o/
Memory leak should be mainly solved now for the multicore version with the help of a critical section. Performance loss around 10% in term of number of simulations.
Still, some bitboards are missing the destruction for unknown reasons.
Possible crash at 2Go... Explain ?
Compiled for win 32 => "32bit is 2gb (for a process)" http://stackoverflow.com/questions/200348/is-there-a-memory-limit-for-a-single-net-process
http://msdn.microsoft.com/en-us/library/9yb4317s.aspx