GPUs render higher computing unit density than contemporary CPUs and thus exhibit much higher power consumption despite its higher power efficiency. The power consumption has become an important issue that impacts CPU...
详细信息
GPUs render higher computing unit density than contemporary CPUs and thus exhibit much higher power consumption despite its higher power efficiency. The power consumption has become an important issue that impacts CPU's applications, thereby necessitating the low power optimization technology for GPUs. Software prefetching is an efficient way to alleviate the memory wall problem which overlaps the computing and memory access latencies. However, software prefetching will cause some power overhead because it increases the number and density of the instructions. Thus, we should consider the balance between the performance income and the power overhead when applying the optimization. To address this problem, in this paper we first analyze the multi-thread execution model of GPU and validate the potential space of software prefetching optimization. Then we give the software prefetching method for GPU programs to improve the performance. Aiming at two different objects: energy optimization under performance constraint and performance optimization under power constraint, we discuss the optimization methods based on software prefetching and dynamic voltage scaling technologies. The experimental results show that our method can efficiently optimize the energy consumption (performance) under the performance (power) constraint.
The development of multi-core processor makes the parallelization of traditional sequential algorithms increasingly important. Meanwhile, transactional memory serves a good parallel programming model. This paper takes...
详细信息
The development of multi-core processor makes the parallelization of traditional sequential algorithms increasingly important. Meanwhile, transactional memory serves a good parallel programming model. This paper takes the advantage of software transactional memory to parallelize the Multi-Exit Asymmetric Adaboost algorithm for face detection. The parallel version is evaluated on three different implementations of software transactional memory. The experiment results show that the transactional memory based parallelization outperforms the traditional lock based approach. A speedup of nearly seven is achieved on a eight-core machine on an eight-core system.
Many challenges in multi-agent coordination can be modeled as distributed Constraint Optimization Problems (DCOPs). Aiming at DCOPs with low constraint density, this paper proposes a distributed algorithm based on the...
详细信息
The reliability issue of Exascale system is extremely serious. Traditional passive fault-tolerant methods, such as rollback-recovery, can not fully guarantee system reliability any more because of their large executin...
详细信息
The reliability issue of Exascale system is extremely serious. Traditional passive fault-tolerant methods, such as rollback-recovery, can not fully guarantee system reliability any more because of their large executing overhead and long recovering duration. Active fault tolerance is expected to become another important fault-tolerant approach for Exascale system. Focusing on system failure prediction, which is one key step of active fault tolerance, we construct online failure prediction model and research on the effective method of system status pretreatment. In order to improve the accuracy and real-time feature of current methods, the proposed Improved Adaptive Semantic Filter (IASF) method processes the latest system logs regularly, filtering useless information out of them according to their semantics. Adopting the main idea of Vector Space Model (VSM), IASF method creates Event Vector corresponding to each log record. By calculating the cosine of vectorial angle, it evaluates the semantics correlation between different log records, and then executes temporal and spatial redundant filter considering the burst feature of log records. IASF method is insensitive to the type of system log and does not introduce any expert system or domain knowledge. The experiment result shows that system can eliminate about 99.6% of useless log records after executing IASF method.
Cloud needs to have rapid and elastic resources supply capability, because of the fluctuant resources demand of end-users. Multi-scale resources elastic binding is an important method to provide cloud services with ra...
详细信息
Cloud needs to have rapid and elastic resources supply capability, because of the fluctuant resources demand of end-users. Multi-scale resources elastic binding is an important method to provide cloud services with rapid and elastic service capability. The most challenging problem in multi-scale resources elastic binding is how to predict the dynamic resource demand of end-users, and then decide when and to what extent multi-scale resources need elastic binding based on the prediction. In this paper, we present the prediction model based on RBF (Radial Basis Function) Network, which is used to predict end-users resource demand in advance. Compared with current prediction methods, it has faster prediction speed and higher prediction accuracy. Then we use traces data (the bandwidth demand of Web type of cloud services) collected from a real-world cloud provider: ChinaCache, as the training and testing data set to validate the method. Finally, we evaluate the predicted results using general prediction accuracy metrics. The results prove that the prediction model based on RBF network is able to resolve the decision problem in multi-scale resources elastic binding.
The Quiet DDoS attack becomes one of the most severely threat to the network safety, because this kind of attack completely adopts legal TCP flow while distributing its destination IP to evade various countermeasu...
详细信息
The Quiet DDoS attack becomes one of the most severely threat to the network safety, because this kind of attack completely adopts legal TCP flow while distributing its destination IP to evade various countermeasures deployed in the network. However, the high distributed degree of the destination IP becomes one characteristics of the attack. However, we think this characteristic make partially of the attack flow not match the behavior habit of network users. Inspired by this viewpoint, we propose a novel method to counter the Quiet DDoS attack based on the NBHU (network behavior habit of users). Furthermore, we carry on simulation of our method using NS2 platform, and the results show that this method can reduce the attack performance.
This paper presents a method that adapting planning description to bring the semantic information into play for service composition through action language C. It shows how service descriptions can be expressed by prec...
详细信息
This paper presents a method that adapting planning description to bring the semantic information into play for service composition through action language C. It shows how service descriptions can be expressed by preconditions and effects and the action language C provides a richer syntax and semantic for complex service descriptions. We also presents the algorithm of Translating semantic Web service described by OWL-S to action language C. Thanks to the structured description and the powerful expression of C, we only consider the initial Situation and the desired goal ignoring details of transition and planning. At last we use satisfiability planning to solve the planning problem by translating the action language into disjunctive logic program.
Increasing Internet business and computing footprint motivate server consolidation in data centers. Through virtualization technology, server consolidation can reduce physical hosts and provide scalable services. Howe...
详细信息
Increasing Internet business and computing footprint motivate server consolidation in data centers. Through virtualization technology, server consolidation can reduce physical hosts and provide scalable services. However, the ineffective memory usage among multiple virtual machines (VMs) becomes the bottleneck in server consolidation environment. Because of inaccurate memory usage estimate and the lack of memory resource managements, there is much service performance degradation in data centers, even though they have occupied a large amount of memory. In order to improve this scenario, we first introduce VM's memory division view and VM's free memory division view. Based on them, we propose a hierarchal memory service mechanism. We have designed and implemented the corresponding memory scheduling algorithm to enhance memory efficiency and achieve service level agreement. The benchmark test results show that our implementation can save 30% physical memory with 1% to 5% performance degradation. Based on Xen virtualization platform and balloon driver technology, our works actually bring dramatic benefits to commercial cloud computing center which is providing more than 2,000 VMs' services to cloud computing users.
暂无评论