Abstract
Motivated by implementing collective communication operations on workstation clusters, a problem of scheduling a dance party is formulated. The problem is solved by two algorithms based on searching and divide-and-conquer that generate suboptimal schedules and an algorithm based on graph factorization that generates optimal schedules. It is shown how to use dance schedules to implement collective communication operations such as all—gather. Experiment data on a single ethernet segment of SUN SPARC-10 workstations and on a switch connected network, the IBM SP2, show that the all—gather implementation that simply uses the optimal schedule performs better for long messages than the implementations of system MPL from IBM and public-domain systems LAM and MPICH/p4.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
