Chen Tianqi development rabit, rabit Allreduce and Broadcast API, want to ask someone used it,
Recently wrote programs need to implement the communication between different nodes in the cluster, hope that can get everyone's advice, thank you

CodePudding user response:

The building Lord, how do you do
Don't know your problem solved: no, I also recently in distributed cluster research, using xgboost rabit mainly come from the MPI interface abstraction, in fact, the communication mechanism of MPI, are embedded in the xgboost rabit interface, can be used directly XGB. The rabit, not long ago and see people using slurm cluster scheduling to distributed computing, now I have just met some distributed and cluster knowledge,
If the host's problems have been solved, in the hope that the original poster can write some blogs to share out,

CodePudding user response:

Last year, according to study direction I do, rabit research over a period of time, groping after successful installation, later found out that rabit framework and is not consistent with my research content, then started to write myself a distributed framework is no longer used rabit, rabit installation related process I have written a blog post, it is on my home page, hereafter have problems can communicate together:)

CodePudding user response:

A rare encounter do distributed deep learning of students!

CodePudding user response:

Ha, ha, ha, we can communicate more