Name
Abstract | ||
|---|---|---|
Multi-/many-core CPU based architectures are seeing widespread adoption due to their unprecedented compute performance in a small power envelope. With the increasingly large number of cores on each node, applications spend a significant portion of their execution time in intra-node communication. While shared memory is commonly used for intra-node communication, it needs to copy each message once at the sender and once at the receiver side. Kernel-assisted mechanisms transfer a message using a single copy but suffer from significant contention with a large number of concurrent accesses. Consequently, naively using Kernel-assisted copy techniques in collectives can lead to severe performance degradation. In this work, we analyze and propose a model to quantify the contention and design collective algorithms to avoid this bottleneck. We evaluate the proposed designs on three different architectures - Xeon, Xeon Phi, and OpenPOWER and compare them against state-of-the-art MPI libraries - MVAPICH2, Intel MPI, and Open MPI. Our designs show up to 50x improvement for One-to-all and All-to-one collectives (Scatter and Gather) and up to 5x improvement for All-to-all collectives (Allgather and Alltoall). |
Year | DOI | Venue |
|---|---|---|
2017 | 10.1109/CLUSTER.2017.106 | 2017 IEEE International Conference on Cluster Computing (CLUSTER) |
Keywords | Field | DocType |
Collective Communication,Kernel-assisted,Cross-Memory Attach,Multi-core,CMA,MPI,HPC | Kernel (linear algebra),Bottleneck,Algorithm design,Shared memory,Xeon Phi,Computer science,Parallel computing,Communication source,Real-time computing,Execution time,Xeon,Distributed computing | Conference |
ISSN | ISBN | Citations |
1552-5244 | 978-1-5386-2327-5 | 3 |
PageRank | References | Authors |
0.39 | 19 | 3 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Sourav Chakraborty | 1 | 381 | 49.27 |
| Hari Subramoni | 2 | 466 | 50.51 |
| Dhabaleswar K. Panda | 3 | 5366 | 446.70 |