Nathalie Furmento

Learn More
The increasing numbers of cores, shared caches and memory nodes within machines introduces a complex hardware topology. High-performance computing applications now have to carefully adapt their placement and behavior according to the underlying hierarchy of hardware resources and their software affinities. We introduce the Hardware Locality (hwloc) software(More)
The move towards Service Grids, where services are composed to meet the requirements of a user community within constraints specified by the resource provider, present many challenges to service provision and description. To support our research activities in the autonomous composition of services to form a Semantic Service Grid we describe the adoption(More)
This paper presents the design of an implementation-independent Service-Oriented Architecture (SOA) which is the main basis of the ICENI Grid middleware. Three implementations of this architecture have been provided on top of Jini, JXTA and the Open Grid Services Infrastructure (OGSI). The main goal of this paper is to discuss these different(More)
Communication libraries have dramatically made progress over the fifteen years, pushed by the success of cluster architectures as the preferred platform for high performance distributed computing. However, many potential optimizations are left unexplored in the process of mapping application communication requests onto low level network commands. The(More)
As the number of cores per machine increases, memory architectures are being redesigned to avoid bus contention and sustain higher throughput needs. The emergence of Non-Uniform Memory Access (NUMA) constraints has caused affinities between threads and buffers to become an important decision criterion for schedulers.
With the prevalence of component based and service oriented architectures used to support e-Science activities, we examine different views of application composition supported within these systems, which tend to be spatial composition in the former case, and temporal composition (workflow) in the latter. We consider the advantages of each view; spatial(More)
GPUs clusters are becoming widespread HPC platforms. Exploiting them is however challenging, as this requires two separate paradigms (MPI and CUDA or OpenCL) and careful load balancing due to node heterogeneity. Current paradigms usually either limit themselves to offload part of the computation and leave CPUs idle, or require static CPU/GPU work(More)
Exploiting the full computational power of current hierarchical multiprocessor machines requires a very careful distribution of threads and data among the underlying non-uniform architecture so as to avoid remote memory access penalties. Directive-based programming languages such as OpenMP, can greatly help to perform such a distribution by providing(More)