Supporting > 64K ranks in CH4/UCX netmod
8 Apr
2021
8 Apr
'21
12:29 a.m.
Hi, I am using MPICH on CH4/UCX netmod, the version is mpich-3.4.1. I noticed that there is a configure parameter "--with-ch4-rank-bits" which can set the value of CH4_RANK_BITS, but seems CH4_RANK_BITS is not used in the code. And I also find in the netmod/ucx/ucx_impl.h, _UCX_init_tag()/_UCX_recv_tag() use only 16 bits to set MPI rank in the ucp_tag, but this cannot differentiate correct ucp_tag when MPI ranks exceed 64K. In Open MPI, 20 bits is used in pml/ucx module to set rank in ucp_tag, 20 bits for context, 24 bits for MPI tag, thus the maximum ranks in Open MPI can be 1M. Is there any plan to support > 64K ranks in MPICH/CH4/UCX? Thanks. Min
1930
Age (days ago)
1930
Last active (days ago)
0 comments
1 participants
participants (1)
-
M Xie