The effect of seance communication on multiprocessing systems

Abstract

This paper introduces the seance communication phenomenon and analyzes its effect on a multiprocessing environment. Seance communication is an unnecessary coherency-related activity that is associated with dead cache information. Dead information may reside in the cache for various reasons: task migration, context switches, or working-set changes. Dead information does not have a significant performance impact on a single-processor system; however, it can dominate the performance of multicache environment. In order to evaluate the overhead of seance communication, we develop an analytical model that is based on the fractal behavior of the memory references. So far, all previous works that used the same modeling approach extracted the fractal parameters of a program manually. This paper provides an additional important contribution by demonstrating how these parameters can be automatically extracted from the program trace. Our analysis indicates that Seance communication may severely reduce the overall system performance when using write-update or writeinvalidate cache coherency protocols. In addition, we find that the performance of writeupdate protocols is affected more severely than write-invalidate protocols. The results that are provided by our model are important for better understanding of the coherency-related overhead in multicache systems and for better development of parallel applications and operating systems.

Lock Cohorting

“…That is, we can move the thread to the cache containing the data instead of viceversa. Thread migration is generally a costly operation, however, and leaves the thread vulnerable to so-called seance communication [Mendelson and Gabbay 2001].…”

Compiler-Guided Throughput Scheduling for Many-core Machines

“…The need for better management of these resources in other computing paradigms is addressed by load balancing mechanisms employed in parallel programming [11,14] and at the cluster level in distributed scheduling [3,4,10]. However, throughput-oriented, system-level schedulers for many-core machines are rare because of the difficulties in effective OS-level load balancing [13,18,21].…”

Tumbler

“…Computing is progressing toward the use of machines that employ multiple multicore CPUs (or Sockets) to provide a large number of cores [Wentzlaff and Agarwal 2009]. While such systems present an opportunity to achieve high performance for multithreaded applications, achieving high performance on a multiCPU multicore system with a large number of cores is difficult due to the challenge of performing effective OSlevel load balancing [Joao et al 2012;Mendelson and Gabbay 2001;Peter et al 2010].…”