<meta http-equiv="Content-Type" content="text/html; charset=utf-8"><div dir="ltr">Hi,<div><br></div><div>I learnt that MPI does broadcast with some tree-based hierarchical communication, which result in O(log p) latency, where p is the number of processes. However, I'm thinking that if I do the broadcast naively(send to all receivers simultaneously), seems that conceptually the latency can be reduced to O(1).</div><div><br></div><div>Does the above make sense? Thanks!</div><div><br></div><div>Best Regards,</div><div>Fan</div></div>