A message passing interface (MPI) collective operation such as broadcast involves multiple processes. The process arrival pattern denotes the timing when each process arrives at a collective operation. It can have a profound impact on the performance since it decides the time when each process can start participating in the operation. In this paper, we investigate the broadcast operation with different process arrival patterns. We analyze commonly used broadcast algorithms and show that they cannot guarantee high performance for different process arrival patterns. We develop two process arrival pattern aware algorithms for broadcasting large messages. The performance of proposed algorithms is theoretically within a constant factor of the optimal for any given process arrival pattern. Our experimental evaluation confirms the analytical results: existing broadcast algorithms cannot achieve high performance for many process arrival patterns while the proposed algorithms are robust and efficient across different process arrival patterns.

Process arrival pattern, which denotes the timing when different processes arrive at an MPI collective operation, can have a significant impact on the performance of the operation. In this work, we characterize the process arrival patterns in a set of MPI programs on two common cluster platforms, use a micro-benchmark to study the process arrival patterns in MPI programs with balanced loads, and investigate the impacts of the process arrival pattern on collective algorithms. Our results show that (1) the differences between the times when different processes arrive at a collective operation are usually sufficiently large to affect the performance; (2) application developers in general cannot effectively control the process arrival patterns in their MPI programs in cluster environments: balancing loads at the application level does not balance the process arrival patterns; and (3) the performance of the collective communication algorithms is sensitive to process arrival patterns. These results indicate that the process arrival pattern is an important factor that must be taken into consideration in developing and optimizing MPI collective routines. We propose a scheme that achieves high performance with different process arrival patterns, and demonstrate that by explicitly considering process arrival pattern, more efficient MPI collective routines than the current ones can be obtained.

We consider efficient implementations of the all-reduce operation with large data sizes on tree topologies. We prove a tight lower bound of the amount of data that must be transmitted to carry out the all-reduce operation and use it to derive the lower bound for the communication time of this operation. We develop a topology specific algorithm that is bandwidth efficient in that (1) the amount of data sent/received by each process is minimum for this operation; and (2) the communications do not incur network contention on the tree topology. With the proposed algorithm, the all-reduce operation can be realized on the tree topology as efficiently as on any other topology when the data size is sufficiently large. The proposed algorithm can be applied to several contemporary cluster environments, including high-end clusters of workstations with SMP and/or multi-core nodes and low-end Ethernet switched clusters. We evaluate the algorithm on various clusters of workstations, including a Myrinet cluster with dual-processor SMP nodes, an InfiniBand cluster with two dual-core processors SMP nodes, and an Ethernet switched cluster with single processor nodes. The results show that the routines implemented based on the proposed algorithm significantly outperform the native MPI_Allreduce and other recently developed algorithms for high-end SMP clusters when the data size is sufficiently large.

We develop a message scheduling scheme for efficiently realizing all-to-all personalized communication (AAPC) on Ethernet switched clusters with one or more switches. To avoid network contention and achieve high performance, the message scheduling scheme partitions AAPC into phases such that 1) there is no network contention within each phase and 2) the number of phases is minimum. Thus, realizing AAPC with the contention-free phases computed by the message scheduling algorithm can potentially achieve the minimum communication completion time. In practice, phased AAPC schemes must introduce synchronizations to separate messages in different phases. We investigate various synchronization mechanisms and various methods for incorporating synchronizations into the AAPC phases. Experimental results show that the message scheduling-based AAPC implementations with proper synchronization consistently achieve high performance on clusters with many different network topologies when the message size is large

We consider unicast-based pipelined broadcast schemes for clusters connected by multiple Ethernet switches. By splitting a large broadcast message into segments and broadcasting the segments in a pipelined fashion, pipelined broadcast may achieve very high performance. We develop algorithms for computing various contention-free broadcast trees on Ethernet switched clusters that are suitable for pipelined broadcast, and evaluate the schemes through experimentation. The conclusions drawn from our theoretical and experimental study include the following. First, pipelined broadcast can be more effective than other common broadcast schemes including the ones used in the latest versions of MPICH and LAM/MPI when the message size is sufficiently large. Second, contention-free broadcast trees are essential for pipelined broadcast to achieve high performance. Finally, while it is difficult to determine the optimal message segment size for pipelined broadcast, finding one size that gives good performance is relatively easy.

An MPI collective operation involves multiple processes that may arrive at the operation at different times. We call such arrival timing the process arrival pattern. The process arrival pattern can significantly affect the performance of a collective operation since it decides the time when each process starts participating in the operation. We characterize the process arrival pattern in a set of MPI programs on two common cluster platforms, study it in a micro-benchmark with balanced loads, and evaluate its impacts on some collective communication algorithms. The results indicate that application developers cannot effectively control the process arrival patterns in cluster environments. Furthermore, the performance of communication algorithms is sensitive to process arrival patterns. Based on these observations, we conclude that, to develop MPI collective routines that can achieve high performance for practical applications in cluster environments, MPI developers must take the process arrival pattern into consideration.

We develop an all-to-all broadcast scheme that achieves maximum bandwidth efficiency for clusters with tree topologies. Using our scheme for clusters with cut-through switches, any tree topology can support all-to-all broadcast as efficiently as a single switch connecting all machines when the message size is sufficiently large. Since a tree topology can be embedded in almost any connected network, it follows that efficient all-to-all broadcast can be achieved in almost all topologies, regular or irregular. To perform all-to-all broadcast efficiently on clusters with store-and-forward switches, the algorithm must minimize the communication path lengths in addition to maximizing bandwidth efficiency. This turns out to be a harder algorithmic problem. We develop schemes that give solutions to common cases for such systems. The performance of our algorithms is evaluated on Ethernet switched clusters with different topologies. The results confirm our theoretical finding. Furthermore, depending on the topology, our algorithms sometimes out-perform the topology-unaware algorithms used in MPI libraries, including MPICH and LAM/MPI, to a very large degree