Segmentation fault about glow HOT 2 CLOSED

[server:13964] *** Process received signal ***
[server:13964] Signal: Segmentation fault (11)
[server:13964] Signal code: Address not mapped (1)
[server:13964] Failing at address: (nil)
[server:13964] [ 0] /lib/x86_64-linux-gnu/libpthread.so.0(+0x11390)[0x7f21205bc390]
[server:13964] [ 1] /home/he/anaconda3/lib/python3.6/site-packages/horovod/common/mpi_lib.cpython-36m-x86_64-linux-gnu.so(+0x22841)[0x7f211e872841]
[server:13964] [ 2] /home/he/anaconda3/lib/python3.6/site-packages/horovod/common/mpi_lib.cpython-36m-x86_64-linux-gnu.so(+0x24c2e)[0x7f211e874c2e]
[server:13964] [ 3] /home/he/anaconda3/lib/libstdc++.so.6(+0xafc5c)[0x7f211e0d4c5c]
[server:13964] [ 4] /lib/x86_64-linux-gnu/libpthread.so.0(+0x76ba)[0x7f21205b26ba]
[server:13964] [ 5] /lib/x86_64-linux-gnu/libc.so.6(clone+0x6d)[0x7f21202e841d]
[server:13964] *** End of error message ***

I met the same issue. My server has 4x 1080 Ti's and 64GB RAM. I don't know which part of the memory cause the error......

from glow.

This is a Horovod/MPI issue so I suggest you look at that forum.

from glow.