Regarding MPI_Allgather collective calls in MPI https://conferences.xsede.org/c/message_boards/find_message?p_l_id=&messageId=416794 2012-11-16T16:06:19Z 2012-11-16T16:06:19Z

The implementation of MPI_Allgather uses different algorithms internally based on the message size and number of processes. For example bruck for short messages, ring for larger messages. Why do they use different implementations? Please explain.

2012-11-16T16:06:19Z