Quo Vadis MPI RMA? Towards a More Efficient Use of MPI One-Sided Communication

TL;DR

This paper analyzes limitations in MPI RMA's current implementation and proposes extensions like window duplication and memory handles to enhance performance and flexibility in one-sided communication.

Contribution

It introduces novel MPI RMA extensions, including window duplication and memory handles, to improve efficiency and usability of one-sided communication.

Findings

Improved accumulate latency performance.

Reduced overhead in multi-threaded flushes.

Enabled zero-overhead dynamic memory windows.

Abstract

The MPI standard has long included one-sided communication abstractions through the MPI Remote Memory Access (RMA) interface. Unfortunately, the MPI RMA chapter in the 4.0 version of the MPI standard still contains both well-known and lesser known short-comings for both implementations and users, which lead to potentially non-optimal usage patterns. In this paper, we identify a set of issues and propose ways for applications to better express anticipated usage of RMA routines, allowing the MPI implementation to better adapt to the application's needs. In order to increase the flexibility of the RMA interface, we add the capability to duplicate windows, allowing access to the same resources encapsulated by a window using different configurations. In the same vein, we introduce the concept of MPI memory handles, meant to provide life-time guarantees on memory attached to dynamic windows, removing the overhead currently present in using dynamically exposed memory. We will show that our extensions provide improved accumulate latencies, reduced overheads for multi-threaded flushes, and allow for zero overhead dynamic memory window usage.