Next to traditional communication systems like servers or routers at the backbone and personal computers or laptops acting as clients, the Future Internet will additionally comprise many so-called smart objects. Most ...
详细信息
Machine learning (ML) has demonstrated considerable promise and superiority in the prediction of microwave dielectric ceramic (MWDC) properties. Nonetheless, applying traditional ML models often requires processing nu...
详细信息
The increasing demand for performance has stimulated the wide adoption of many-core accelerators like IntelR Xeon PhiTMCoprocessor, which is based on Intel’s Many Integrated Core architecture. While many HPC applicat...
详细信息
The increasing demand for performance has stimulated the wide adoption of many-core accelerators like IntelR Xeon PhiTMCoprocessor, which is based on Intel’s Many Integrated Core architecture. While many HPC applications running in native mode have been tuned to run efficiently on Xeon Phi, it is still unclear how a managed runtime like JVM performs on such an architecture. In this paper, we present the first measurement study of a set of Java HPC applications on Xeon Phi under JVM. One key obstacle to the study is that there is currently little support of Java for Xeon Phi. This paper presents the result based on the first porting of Open JDK platform to Xeon Phi, in which the Hot Spot virtual machine acts as the kernel execution engine. The main difficulty includes the incompatibility between Xeon Phi ISA and the assembly library of Hotspot *** evaluating the multithreaded Java Grande benchmark suite and our ported Java Phoenix benchmarks, we quantitatively study the performance and scalability issues of JVM on Xeon Phi and draw several conclusions from the study. To fully utilize the vector computing capability and hide the significant memory access latency on the coprocessor, we present a semi-automatic vectorization scheme and software prefetching model in Hot *** with 60 physical cores and tuning, our optimized JVM achieves averagely 2.7 x and 3.5 x speedup compared to Xeon CPU processor by using vectorization and prefetching accordingly. Our study also indicates that it is viable and potentially performance-beneficial to run applications written for such a managed runtime like JVM on Xeon Phi.
Simulations of mobile networks require the specification of the node movement. The movement is either computed based on mathematical models or given by previously recorded mobility traces. The Mobile Node Trace Genera...
Routing is a general task, yet the implementation of routing protocols requires specific operating system related knowledge. The developer has to deal with particular kernel internals that might have severe side effec...
详细信息
Dynamic Source Routing (DSR) was one of the first proposed protocols for routing in Mobile Ad-hoc Networks and is among the few that were ratified by the IETF. Despite its source routing overhead, the protocol exhibit...
详细信息
Gossip routing is an approach to limit the overhead of flooding in wireless networks. Each node, that receives a packet that would normally be flooded, applies a probabilistic approach. The packet is either forwarded ...
详细信息
With computing systems undergone a fundamen- tal transformation from single-processor devices at the turn of the century to the ubiquitous and networked devices and the warehouse-scale computing via the cloud, the par...
详细信息
With computing systems undergone a fundamen- tal transformation from single-processor devices at the turn of the century to the ubiquitous and networked devices and the warehouse-scale computing via the cloud, the parallelism has become ubiquitous at many levels. At micro level, par- allelisms are being explored from the underlying circuits, to pipelining and instruction level parallelism on multi-cores or many cores on a chip as well as in a machine. From macro level, parallelisms are being promoted from multiple ma- chines on a rack, many racks in a data center, to the glob- ally shared infrastructure of the Internet. With the push of big data, we are entering a new era of parallel computing driven by novel and ground breaking research innovation on elas- tic parallelism and scalability. In this paper, we will give an overview of computing infrastructure for big data processing, focusing on architectural, storage and networking challenges of supporting big data paper. We will briefly discuss emerging computing infrastructure and technologies that are promising for improving data parallelism, task parallelism and encour- aging vertical and horizontal computation parallelism.
Interference is one of the major causes for performance degradation in wireless networks. Channel assignment algorithms have been proven successful to decrease the networkwide interference by using non-overlapping cha...
详细信息
ISBN:
(纸本)9781612081717
Interference is one of the major causes for performance degradation in wireless networks. Channel assignment algorithms have been proven successful to decrease the networkwide interference by using non-overlapping channels for otherwise interfering links. However, external co-located networks and devices are usually not considered in the channel assignment procedure, since they are not under the control of the network operator and their activity is therefore hard to capture. In our work we fill this gap by additionally considering the interference resulting from external devices. The novelty of this approach is that not only co-located IEEE 802.11 networks are captured, but also other sources of interference that utilize the same frequency band. We present the spectrum sensing component DES-Sense, a software solution for 802.11a/b/g that detects congested channels and does not require any changes to the *** present a first algorithm for external interference-aware channel assignment and show proof-of-concept results from the DES-Testbed, a wireless mesh network with 120 multi-radio nodes.
暂无评论