Overlapping Communication with Computation Symposium

This Symposium will cover current strategies for computation/communication scalable overlap, including, but not limited to

  • overdecomposition
  • non-blocking communication
  • speculation strategies
  • hardware offloading


The symposium will address current parallel computation implementations use of these and other strategies. Experts in parallel computing strategies will present current and future computation/communication techniques. This will be a single all-day symposium including talks, panel discussions and break out session.


Speaker Affiliation Time Title
Organizing committee   9:00 am Introduction
Dhabaleswar K. (DK) Panda OSU 9:10 am Exploiting Computation and Communication Overlap in MVAPICH2 and MVAPICH2-GDR MPI Libraries
- - 10:30 am COFFEE BREAK
Julien Jaeger CEA - MPC 11:00 am Towards an Efficient Communication Overlap through Hardware Offloading
- - 12:00 noon LUNCH (included with the registration)
Sayantan Sur Intel 1:15 pm Communication Overlap using Open Fabric Interfaces
- - 2:00 pm COFFEE BREAK
Pavel Shamis ARM 2:30 pm HPC network stack on Arm
Howard Pritchard Jr. LANL 3:15 pm MPI RMA – The State of the Art in Open MPI
Jonathan Beard ARM 4:00 pm Reducing dark bandwidth through data reduction near memory
  4:45 pm BREAK
all 5:00 pm Panel Discussion
all 6:30 pm Pay-for-your-own Dinner at Boulder Beer (open to all attendees)


Organizing committee