My 10 card rig works ok, 13 card rig crashes.
My build is for linux, so it's normal to have a different size.
I get +5 mh for each card (1080ti), too bad that it doesnt work on the 13 cards rig
what's your gcc and CUDA version? Did you add nvcc_ARCH += -gencode=arch=compute_61,code=\"sm_61,compute_61\" to Makefile.am?