consulting about libtorch memory leak about libdeepvac HOT 3 CLOSED

Please be noted that memory Increase != memory leak, instead, you can report below RSS info here:
1, the libtorch process RSS usage after 100 times inference (i.e. the memory usage is stable now);
2, the libtorch process RSS usage after 200 times inference;
If memory usage increase dramatically during 100 and 200 times inference, then you can test your code on MLab HomePod 2.0 pro. If memory still leak on HomePod environment, we can help to debug this issue.

from libdeepvac.

Please be noted that memory Increase != memory leak, instead, you can report below RSS info here:
1, the libtorch process RSS usage after 100 times inference (i.e. the memory usage is stable now);
2, the libtorch process RSS usage after 200 times inference;
If memory usage increase dramatically during 100 and 200 times inference, then you can test your code on MLab HomePod 2.0 pro. If memory still leak on HomePod environment, we can help to debug this issue.

Thank you for your reply. Yes though valgrind shows possible lost, but the RSS is relatively stable, increase a little but not dramatically after some inference. So, I think maybe there is no problem at all.

from libdeepvac.

不客气。

from libdeepvac.