High-performance RMA-based broadcast on the intel SCC

Abstract

Many-core chips with more than 1000 cores are expected by the end of the decade. To overcome scalability issues related to cache coherence at such a scale, one of the main research directions is to leverage the message-passing programming model. The Intel Single-Chip Cloud Computer (SCC) is a prototype of a message-passing many-core chip. It offers the ability to move data between on-chip Message Passing Buffers (MPB) using Remote Memory Access (RMA). Performance of message-passing applications is directly affected by efficiency of collective operations, such as broadcast. In this paper, we study how to make use of the MPBs to implement an efficient broadcast algorithm for the SCC. We propose OC-Bcast (On-Chip Broadcast), a pipelined k-ary tree algorithm tailored to exploit the parallelism provided by on-chip RMA. Using a LogP-based model, we present an analytical evaluation that compares our algorithm to the state-of-the-art broadcast algorithms implemented for the SCC. As predicted by the model, experimental results show that OC-Bcast attains almost three times better throughput, and improves latency by at least 27%. Furthermore, the analytical evaluation highlights the benefits of our approach: OC-Bcast takes direct advantage of RMA, unlike the other considered broadcast algorithms, which are based on a higher-level send/receive interface. This leads us to the conclusion that RMA-based collective operations are needed to take full advantage of hardware features of future message-passing many-core architectures.

High-Throughput Maps on Message-Passing Manycore Architectures: Partitioning versus Replication

“…Intel SCC [17]. The main reason is that inter-core communication in these architectures involves some synchronization costs [21] which are not included in our model.…”

“…A few recent studies have proposed performance models for other manycore architectures [21,24]. Our approach is similar to the one used in these papers.…”

“…They all cover the same communication scenarios as the LogP model [11] (or its extensions) that is commonly used in message-passing systems. The main difference is that the underlying communication system considered in these studies are different from the one of this chapter: [21] models RMA-based communication and targets the Intel SCC processor; [24] models point-to-point communication on top of cache-coherent shared memory and targets the Intel Xeon Phi processor.…”

Performance analysis of broadcasting algorithms on the Intel Single-Chip Cloud Computer

“…OC-BCast [17] is another efficient broadcasting algorithm; this algorithm is similar to ModMPB as it also has the receiving cores copy the message from the sending core's MPB to their own MPBs. However, there are a couple distinct differences between the two broadcasts.…”

Fault tolerance for remote memory access programming models

“…Partitioned Global Address Space (PGAS), and the wider class of Remote Memory Access (RMA) programming models enable high-performance communications that often outperform Message Passing [19,34]. RMA utilizes remote direct memory access (RDMA) hardware features to access Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page.…”