The Experts below are selected from a list of 222 Experts worldwide ranked by ideXlab platform
Tjerk P Straatsma - One of the best experts on this subject based on the ideXlab platform.
-
a multiprotocol communication support for the global address space programming model on the ibm sp
European Conference on Parallel Processing, 2000Co-Authors: Jarek Nieplocha, Jialin Ju, Tjerk P StraatsmaAbstract:The paper describes an efficient communication support for the global address space programming model on the IBM SP, a commercial example of the SMP (symmetric multi-processor) clusters. Our approach integrates shared memory with active messages, threads and remote memory copy between nodes. The shared memory operations offer substantial performance improvement over LAPI, IBM one-sided communication library, within an SMP node. Based on the experiments with the SPLASH-2 LU benchmark and a molecular dynamics simulation, our multiprotocol support for the global address space is found to improve performance and scalability of applications. This approach could also be used in optimizing the MPI-2 one-sided communication on the SMP clusters.
-
Euro-Par - A Multiprotocol Communication Support for the Global Address Space Programming Model on the IBM SP
Euro-Par 2000 Parallel Processing, 2000Co-Authors: Jarek Nieplocha, Jialin Ju, Tjerk P StraatsmaAbstract:The paper describes an efficient communication support for the global address space programming model on the IBM SP, a commercial example of the SMP (symmetric multi-processor) clusters. Our approach integrates shared memory with active messages, threads and remote memory copy between nodes. The shared memory operations offer substantial performance improvement over LAPI, IBM one-sided communication library, within an SMP node. Based on the experiments with the SPLASH-2 LU benchmark and a molecular dynamics simulation, our multiprotocol support for the global address space is found to improve performance and scalability of applications. This approach could also be used in optimizing the MPI-2 one-sided communication on the SMP clusters.
Jarek Nieplocha - One of the best experts on this subject based on the ideXlab platform.
-
a multiprotocol communication support for the global address space programming model on the ibm sp
European Conference on Parallel Processing, 2000Co-Authors: Jarek Nieplocha, Jialin Ju, Tjerk P StraatsmaAbstract:The paper describes an efficient communication support for the global address space programming model on the IBM SP, a commercial example of the SMP (symmetric multi-processor) clusters. Our approach integrates shared memory with active messages, threads and remote memory copy between nodes. The shared memory operations offer substantial performance improvement over LAPI, IBM one-sided communication library, within an SMP node. Based on the experiments with the SPLASH-2 LU benchmark and a molecular dynamics simulation, our multiprotocol support for the global address space is found to improve performance and scalability of applications. This approach could also be used in optimizing the MPI-2 one-sided communication on the SMP clusters.
-
Euro-Par - A Multiprotocol Communication Support for the Global Address Space Programming Model on the IBM SP
Euro-Par 2000 Parallel Processing, 2000Co-Authors: Jarek Nieplocha, Jialin Ju, Tjerk P StraatsmaAbstract:The paper describes an efficient communication support for the global address space programming model on the IBM SP, a commercial example of the SMP (symmetric multi-processor) clusters. Our approach integrates shared memory with active messages, threads and remote memory copy between nodes. The shared memory operations offer substantial performance improvement over LAPI, IBM one-sided communication library, within an SMP node. Based on the experiments with the SPLASH-2 LU benchmark and a molecular dynamics simulation, our multiprotocol support for the global address space is found to improve performance and scalability of applications. This approach could also be used in optimizing the MPI-2 one-sided communication on the SMP clusters.
Albert John Camilleri - One of the best experts on this subject based on the ideXlab platform.
-
TPHOLs - A Hybrid Approach to Verifying Liveness in a symmetric multi-processor
Lecture Notes in Computer Science, 1997Co-Authors: Albert John CamilleriAbstract:The verification of increasingly complex multiprocessor systems has become a formidable challenge. Given the complexity of each of the multi-processor components and the complex interactions between them, simulation may no longer provide the desired level of coverage to detect possible starvation and deadlocks. The Formal Verification Group at Hewlett-Packard has used a combination of model checking and theorem proving techniques with great effect to analyse the correctness of the Runway bus arbitration protocols in models representing several members of the HP 9000 K-CLASS servers and J-CLASS workstations. This paper presents some basic ideas behind this verification methodology, highlights its strengths and shortcomings, and discusses how it can be used to enhance the existing system-verification methodology.
Huang Hao - One of the best experts on this subject based on the ideXlab platform.
-
Improving the performance of NIDS using symmetric multi-processor
Computer Engineering, 2005Co-Authors: Lai Hai-guang, Huang HaoAbstract:Network-based intrusion detection system (NIDS) detects attacks by capturing and analyzing network packets. As network band increases, NIDS can hardly keep up with the speed of networks. A method of improving NIDS' process ability using symmetric multi-processor (SMP) was proposed in the paper. Several CPUs of the system were used to process network packets in parallel to improve the performance. After analyzing NIDS' process procedure, an effective parallel processing structure was devised, which guaranteed threads on different CPUs running in parallel. Moreover, the synchronization method of threads proposed avoided the mutually exclusive access to the shared resource, which further increased the parallelity of threads, and guaranteed the correctness of the functionality of the program. Experiments show that the NIDS implemented on a SMP system with dual CPUs is almost 80% faster than the one based on a system with unique CPU.
Jialin Ju - One of the best experts on this subject based on the ideXlab platform.
-
a multiprotocol communication support for the global address space programming model on the ibm sp
European Conference on Parallel Processing, 2000Co-Authors: Jarek Nieplocha, Jialin Ju, Tjerk P StraatsmaAbstract:The paper describes an efficient communication support for the global address space programming model on the IBM SP, a commercial example of the SMP (symmetric multi-processor) clusters. Our approach integrates shared memory with active messages, threads and remote memory copy between nodes. The shared memory operations offer substantial performance improvement over LAPI, IBM one-sided communication library, within an SMP node. Based on the experiments with the SPLASH-2 LU benchmark and a molecular dynamics simulation, our multiprotocol support for the global address space is found to improve performance and scalability of applications. This approach could also be used in optimizing the MPI-2 one-sided communication on the SMP clusters.
-
Euro-Par - A Multiprotocol Communication Support for the Global Address Space Programming Model on the IBM SP
Euro-Par 2000 Parallel Processing, 2000Co-Authors: Jarek Nieplocha, Jialin Ju, Tjerk P StraatsmaAbstract:The paper describes an efficient communication support for the global address space programming model on the IBM SP, a commercial example of the SMP (symmetric multi-processor) clusters. Our approach integrates shared memory with active messages, threads and remote memory copy between nodes. The shared memory operations offer substantial performance improvement over LAPI, IBM one-sided communication library, within an SMP node. Based on the experiments with the SPLASH-2 LU benchmark and a molecular dynamics simulation, our multiprotocol support for the global address space is found to improve performance and scalability of applications. This approach could also be used in optimizing the MPI-2 one-sided communication on the SMP clusters.
