Resource-Based Communication Placement Analysis
نویسندگان
چکیده
Communication placement analysis is an important step in the compilation of data-parallel programs. However, to simplify the placement analysis, previous techniques ignored most machine-dependent resource constraints. This paper demonstrates the necessity of incorporating resource constraints in ensuring the correctness of the communication placement. It presents a new placement analysis technique that minimizes frequency of communication, eliminates redundant communication , and maximizes communication latency hiding while taking limited resources into account. The paper illustrates resource-based placement analysis in the context of placement of distributed-memory communication primitives and limited buuer size resource constraint. In addition, it shows the use of stripmining transformation in improving the eecacy of resource-based communication placement.
منابع مشابه
Quasi Random Deployment Strategy for Reliable Communication Backbones in Wireless Sensor Networks
Topology construction and topology maintenance are significant sub-problems of topology control. Spanning tree based algorithms for topology control are basically transmission range based type construction algorithms. The construction of an effective backbone, however, is indirectly related to the placement of nodes. Also, the dependence of network reliability on the communication path undertak...
متن کاملCommunication-Aware Traffic Stream Optimization for Virtual Machine Placement in Cloud Datacenters with VL2 Topology
By pervasiveness of cloud computing, a colossal amount of applications from gigantic organizations increasingly tend to rely on cloud services. These demands caused a great number of applications in form of couple of virtual machines (VMs) requests to be executed on data centers’ servers. Some of applications are as big as not possible to be processed upon a single VM. Also, there exists severa...
متن کاملAdaptive Dynamic Data Placement Algorithm for Hadoop in Heterogeneous Environments
Hadoop MapReduce framework is an important distributed processing model for large-scale data intensive applications. The current Hadoop and the existing Hadoop distributed file system’s rack-aware data placement strategy in MapReduce in the homogeneous Hadoop cluster assume that each node in a cluster has the same computing capacity and a same workload is assigned to each node. Default Hadoop d...
متن کاملThe Impact of Doing Problem-Solving Tasks on Willingness to Communicate, Self-perceived Communication Competence, and Communication Apprehension among Iranian EFL Learners
Willingness to communicate (WTC) plays a central role in learning a second/foreign language. The present study was conducted to investigate the effect of doing problem-solving task on learners’ WTC in English. To fulfill the purpose of the study, 61 pre-intermediate learners were chosen by means of administrating a placement test. Learners were pretested through three scales of WTC and its two ...
متن کاملTrade-Off Study of Localizing Communication and Balancing Network Traffic on a Dragonfly System
Dragonfly networks are being widely adopted in high-performance computing systems. On these networks, however, interference caused by resource sharing can lead to significant network congestion and performance variability. We present a comparative analysis exploring the trade-off between localizing communication and balancing network traffic. We conduct trace-based simulations for applications ...
متن کامل