Communication Analysis for Shared and Distributed Memory Machines
نویسنده
چکیده
Advances in programming languages and parallelizing compilers are making parallel computers easier to use by providing a high-level portable programming model that protects software investment. However, experience has shown that simply finding parallelism is not always sufficient for obtaining good performance from today’s multiprocessors, largely because the cost of interprocessor communication is much greater than computation or local memory accesses. To overcome this problem, I believe compilers need to perform communication analysis to locate and optimize interprocessor communication. I show how communication analysis has been used to improve performance for both shared and distributed memory machines, and describe a new project to apply these techniques to compilers for software distributed-shared-memory (DSM) systems.
منابع مشابه
On the Coexistence of Shared-Memory and Message-Passing in the Programming of Parallel Applications
Interoperability in non-sequential applications requires communication to exchange information using either the shared-memory or message-passing paradigm. In the past, the communication paradigm in use was determined through the architecture of the underlying computing platform. Shared-memory computing systems were programmed to use shared-memory communication, whereas distributed-memory archi-...
متن کاملTo Appear in the First Symposium on Operating Systems Design and Implementation The Design and Evaluation of a Shared Object System for Distributed Memory Machines
This paper describes the design and evaluation of SAM, a shared object system for distributed memory machines. SAM is a portable run-time system that provides a global name space and automatic caching of shared data. SAM incorporates mechanisms to address the problem of high communication overheads on distributed memory machines; these mechanisms include tying synchronization to data access, ch...
متن کاملThe Effects of Latency, Occupancy, and Bandwidth in Distributed Shared Memory Multiprocessors
Distributed shared memory (DSM) machines can be characterized by four parameters, based on a slightly modified version of the logP model. The l (latency) and o (occupancy of the communication controller) parameters are the keys to performance in these machines, and are largely determined by major architectural decisions about the aggressiveness and customization of the node and network. For rec...
متن کاملComparison of MPI Benchmark Programs on Shared Memory and Distributed Memory Machines (Point-to-Point Communication)
There are several benchmark programs available to measure the performance of MPI on parallel computers. The most common use of MPI benchmarks software are SKaMPI, Pallas MPI Benchmark, MPBench, Mpptest and MPIBench. It is interesting to analyze the differences between different benchmark. Presently, there have been few comparisons done between the different benchmarks. Thus, in this paper we di...
متن کاملAutomatic Data Decomposit ion for Message-Passing Machines
1 I n t r o d u c t i o n Distributed-memory message-passing computers are becoming more common these days because they offer significant advantages over shared-memory machines in terms of cost and scalability. However, distr ibuted-memory machines are more difficult to program than shared-memory machines because programmers of distributed-memory machines have to manage low-level tasks like dis...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
دوره شماره
صفحات -
تاریخ انتشار 1995