This paper presents an overview of our work, concerning a complete end-to-end framework for automatically generating message passing parallel code for tiled nested for-loops. It considers general parallelepiped tiling...
详细信息
ISBN:
(纸本)9781581138122
This paper presents an overview of our work, concerning a complete end-to-end framework for automatically generating message passing parallel code for tiled nested for-loops. It considers general parallelepiped tiling transformations and general convex iteration spaces. We address all problems regarding both the generation of sequential tiled code and its parallelization. We have implemented our techniques in a tool which automatically generates MPI parallel code and conducted several series of experiments, concerning the compilation time of our tool, the efficiency of the generated code and the speedup attained on a cluster of PCs. Apart from confirming the value of our techniques, our experimental results show the merit of general parallelepiped tiling transformations and verify previous theoretical work on scheduling-optimal tile shapes.
A data warehouse (DW) provides an information for analytical processing, decision making, and data mining tools. On the one hand, the structure and content of a data warehouse reflects a real world, i.e. data stored i...
详细信息
A data warehouse (DW) provides an information for analytical processing, decision making, and data mining tools. On the one hand, the structure and content of a data warehouse reflects a real world, i.e. data stored in a DW come from real production systems. On the other hand, a DW and its tools may be used for predicting trends and simulating a virtual business scenarios. This activity is often called the what-if analysis. Traditional DW systems have static structure of their schemas and relationships between data, and therefore they are not able to support any dynamics in their structure and content. For these purposes, multiversion data warehouses seem to be very promising. In this paper we present a concept and an ongoing implementation of a multiversion data warehouse that is capable of handling changes in the structure of its schema as well as simulating alternative business scenarios.
This work presents the idea and a prototype of workflow systems whose definition is based on constraints. The flexibility is reached through the less rigid definition of workflow definitions - the workflow is defined ...
详细信息
This work presents the idea and a prototype of workflow systems whose definition is based on constraints. The flexibility is reached through the less rigid definition of workflow definitions - the workflow is defined as a set of pre and post conditions of activities, which are selected dynamically as the process instance unfolds. The workflow system besides dispatching activities that have all their preconditions fulfilled to be executed, also helps users to decide which activity to chose through what if scenarios. The system also includes an access control model which not only represents which users have the authority to chose and execute the activities but also the authority to override the constraints. In particular, overriding constraints is itself an activity and thus may have pre and post conditions defined in other constraints. The paper present Tucupi, a prototype of such constraint based WFMS.
We present an efficient and practical lock-free implementation of a concurrent dictionary that is suitable for both fully concurrent (large multi-processor) systems as well as pre-emptive (multi-process) systems. Many...
详细信息
ISBN:
(纸本)9781581138122
We present an efficient and practical lock-free implementation of a concurrent dictionary that is suitable for both fully concurrent (large multi-processor) systems as well as pre-emptive (multi-process) systems. Many algorithms for concurrent dictionaries are based on mutual exclusion. However, mutual exclusion causes blocking which has several drawbacks and degrades the system's overall performance. Non-blocking algorithms avoid blocking, and are either lock-free or wait-free. Our algorithm is based on the randomized sequential list structure called Skiplist, and implements the full set of operations on a dictionary that is suitable for practical settings. In our performance evaluation we compare our algorithm with the most efficient non-blocking implementation of dictionaries known. The experimental results clearly show that our algorithm outperforms the other lock-free algorithm for dictionaries with realistic sizes, both on fully concurrent as well as pre-emptive systems.
Protein threading programs align a probe amino acid sequence onto a library of representative folds of known protein structure to identify a structural homology. A scoring function is usually formulated in terms of th...
详细信息
ISBN:
(纸本)9781581138122
Protein threading programs align a probe amino acid sequence onto a library of representative folds of known protein structure to identify a structural homology. A scoring function is usually formulated in terms of the threading energy to evaluate protein sequence-structure fitness. The structure that yields the lowest total energy is considered the leading template of the probe protein. An alternative approach is to predict the probabilities of observing amino acid side-chains in structural environment without considering the energy of contacts. In this paper, a model named TES is proposed on building a new environment-specific protein sequence-structure mapping with artificial neural network. The decoy sets obtained from the web are used to test the proposed TES method on discrimination of native and decoy protein three-dimensional structure. The verified approach shows that the performance of the proposed method is comparable to those of knowledge-based potential energy function.
Central to grid processing is the scheduling of application tasks to resources. Schedulers need to consider heterogeneous computational and communication resources, producing the shortest possible schedule under time ...
详细信息
ISBN:
(纸本)1595934804;9781595934802
Central to grid processing is the scheduling of application tasks to resources. Schedulers need to consider heterogeneous computational and communication resources, producing the shortest possible schedule under time constraints dictated by both the application needs and the frequency of fluctuation of resource availability. This paper introduces a set of schedulers with such characteristics. Copyright 2007 acm.
Self-healing is one of the main challenges to growing autonomic pervasive computing. Fault detection and recovery are the main steps of self-healing. Due to the characteristics of pervasive computing the self-healing ...
详细信息
ISBN:
(纸本)1595934804;9781595934802
Self-healing is one of the main challenges to growing autonomic pervasive computing. Fault detection and recovery are the main steps of self-healing. Due to the characteristics of pervasive computing the self-healing becomes difficult. In this paper, the challenges of self-healing have been addressed and an approach to develop a self-healing service for autonomic pervasive computing is presented. The self-healing service has been developed and integrated into the middleware named MARKS+ (Middleware Adaptability for Resource discovery, Knowledge usability, and Self-healing). The self-healing approach is being evaluated on a test bed of PDAs. An application is being developed by using the proposed service. Copyright 2007 acm.
In this paper, we propose a representation of controllers that uses high-level sensors and possesses a general and intuitive structure that offers several types of parameters, which can be modified by the animator or ...
详细信息
ISBN:
(纸本)1595934804;9781595934802
In this paper, we propose a representation of controllers that uses high-level sensors and possesses a general and intuitive structure that offers several types of parameters, which can be modified by the animator or automatically. This structure, with the feedback signals provided by its sensors, allows several state machines to act simultaneously on the model, or in a subset of its actuators. The sensors can be optimized, facilitating their definition and use. The representation also permits the animator to define procedures with more general instructions in order to be automatically executed by the controller during the dynamics simulation. Copyright 2007 acm.
Updating legacy applications to let them properly run in a Grid environment generally requires a noticeable development effort. The approach proposed in this work is intended to reduce this effort through the automate...
详细信息
ISBN:
(纸本)1595934804;9781595934802
Updating legacy applications to let them properly run in a Grid environment generally requires a noticeable development effort. The approach proposed in this work is intended to reduce this effort through the automated generation of aspects that, once incorporated into legacy applications, ensure integration with the Globus Grid environment. The generated aspects encapsulate concerns handling the interaction with Globus services, thus keeping the relevant code separated from application classes. Copyright 2007 acm.
We propose a remote file system for use in desktop grid applications. The proposed system provides programs with the ability to query and update files on remote machines without requiring any changes to the program co...
详细信息
ISBN:
(纸本)1595934804;9781595934802
We propose a remote file system for use in desktop grid applications. The proposed system provides programs with the ability to query and update files on remote machines without requiring any changes to the program code. The system (a) generates a profile of the program file operations and (b) uses the profile to implement collective and pre-fetched I/O. Copyright 2007 acm.
暂无评论