Search

Bloom filters are data structures that can efficiently represent a set of elements providing operations of insertion and membership testing. Nevertheless, these filters may yield false positive results when testing for elements that have not been previously inserted. In general, higher false positive rates are expected for sets with larger cardinality with constant filter size. This paper shows that for sets where a distance metric can be defined, reducing the false positive rate is possible if elements to be inserted exhibit locality according to this metric. In this way, a hardware alternative to Bloom filters able to extract spatial locality features is proposed and analyzed. [ABSTRACT FROM AUTHOR]

Modern Graphics Processing Units (GPU) are very powerful computational systems on a chip. For this reason there is a growing interest in using these units as general purpose hardware accelerators (GPGPU). To facilitate the programming of general purpose applications, NVIDIA introduced the CUDA programming environment. CUDA provides a simplified abstraction of the underlying complex GPU architecture, so as a number of critical optimizations must be applied to the code in order to get maximum performance. In this paper we discuss our experience in porting an application kernel to the GPU, and all classes of design decisions we adopted in order to obtain maximum performance. [ABSTRACT FROM AUTHOR]

This article presents a system which is based on a distributed network architecture. The system defines a "double" Client-Server structure which permits to incorporate new client systems in real-time using the Internet. We have used this distributed architecture system to develop a tool to measure the computer performance characteristics of several computer architecture systems. The computer performance is tested by a free library called PAPI (Performance API), which allows us to access to internal status registers of several CPU families in several Operating Systems. As this testing has to be done in real CPUs, we have to create a new system to provide this functionality. The structure proposed has only one entry point to the whole system, the Master Server, and several different Architecture Client Servers. We present the network system description and the usage of PAPI for performance testing. [ABSTRACT FROM AUTHOR]

During the past few years, mainstream computing has rapidly embraced multicore architectures and there is now considerable research into the design of such systems and into all aspects of their utilization. In particular, much work is needed in order to improve their programmability. Limitations on the amount of memory, bandwidth constraints, multiple layers of parallelism and heterogeneous components all introduce new challenges for application development and execution. In recognition of the importance of this architectural shift, multi-core programming was added to the collection of Euro-Par topics this year. The focus of the topic covered general-purpose multi-core programming techniques, models and languages, as well as those for multi-core embedded systems, and included related work on compilers, run-time systems, and performance and scalability studies. [ABSTRACT FROM AUTHOR]

Welcome to Euro-Par's Topic 4, which provides a forum for the presentation of the latest research results and practical experience in Compilers for High Performance. Topic 4 deals with compilation for high performance architectures. Papers were invited targeting both general-purpose platforms and specialised hardware designs such as graphic coprocessors or low-power embedded systems. We were also keen to solicit papers that explore more general issues, such as program analysis, languages, run-time systems, and feedback-directed optimisation. [ABSTRACT FROM AUTHOR]