Martha Torres | Universidade Estadual de Santa Cruz (original) (raw)
Papers by Martha Torres
The aim of this work is to measure the barrier synchronization influence on the overall performan... more The aim of this work is to measure the barrier synchronization influence on the overall performance of several application programs. In order to do that, we use the MPICH implementation version 1.1 of the MPI library. Moreover, we choose two barrier synchronization solutions: the MPICH software solution, i.e. the MPI
Resumo O Plasma é o estado mais comum da matéria no universo, mas é relativamente raro na Terra. ... more Resumo O Plasma é o estado mais comum da matéria no universo, mas é relativamente raro na Terra. O desafio para obtenção de energia através da Fusão Termonuclear Controlada (para muitos a fonte de energia do futuro) depende do aquecimento de ...
Hierarchical clustering algorithms are usually applied to bioinformatic procedures such as groupi... more Hierarchical clustering algorithms are usually applied to bioinformatic procedures such as grouping of genes and proteins with similar structure, reconstruction of evolutionary trees, gene expression analysis, etc [2] [1]. Such applications usually require high memory size and CPU time, therefore, the parallel implementation ot these algorithms is a good option [4].
IEEE Latin America Transactions, 2011
2008 11th IEEE International Conference on Computational Science and Engineering, 2008
Plasma (ionized gas) is the most common state of the matter in the universe but relatively rare o... more Plasma (ionized gas) is the most common state of the matter in the universe but relatively rare on Earth. The quest for the Thermonuclear Controlled Fusion (as an energy source) depends on heating plasmas to temperatures around 10,000,000 Kelvin. The optimization of energy deposition in plasmas by heating them through RF waves in machines called tokamaks has been subject of thorough research (with intense numerical modeling). This work is embedded in this subject with main focus in the run time reduction for the ICANT code, which provides self-consistent calculation of RF current values in a heating antenna model. Typical assessment of an antenna model is made with several runs, each taking several hours, turning this task into day's job. The parallelization was obtained and yielded the same results of the serial version with fair scalability (up to 10 nodes).
lbd.dcc.ufmg.br
Resumo. Este artigo descreve a implementaçao paralela e sequencial do algoritmo de clustering hie... more Resumo. Este artigo descreve a implementaçao paralela e sequencial do algoritmo de clustering hierárquico CURE. Além disso, este artigo mostra a validaç ao da implementaç ao realizada ea avaliaç ao de desempenho da versao paralela.
Genetics and Molecular Biology, 2004
GCLASS is an algorithm which explores small samples of two distinct biological states for finding... more GCLASS is an algorithm which explores small samples of two distinct biological states for finding small sets of genes, which form a feature vector that is enough to separate these two states. A typical sample is a set of 60 microarrays, 30 for each biological state, with several thousand genes. The technique consists of the following: a spreading model defined in the space of small sets of genes studied and centered in each feature vector considered; the designing of optimal linear classifiers under this spreading model; and ranking the designed classifiers, based on their error and robustness relative to the spreading. The feature vectors used in the best classifiers are considered the best feature vectors. Due to the great number of potential feature sets, a parallel implementation is a good option for reducing the procedure execution time. This paper presents a parallel solution of GCLASS and shows some performance results. The experimental results show that the proposed solution provides quasi linear speedup if compared to the sequential implementation. For example, using 60 genes as the complete feature space and 6 genes as the small feature space, our parallel version with 11 processors is approximately 10.98 times faster than the sequential version.
The aim of this work is to measure the barrier synchronization influence on the overall pevormanc... more The aim of this work is to measure the barrier synchronization influence on the overall pevormance of several application programs. In order to do that, we use the MPICH implementation [ I ] version 1.1 of the MPI library. Moreover, we choose two barrier synchronization solutions: the MPICH software solution, i.e. the MPI-Barrier routine, and the TTL-PAPERS hardware solution [ 2 ]. The last one was implemented as a collective operation in MPICH we called MPI-Papers. Finally, we compare the execution time of MPI programs using the MPI-Barrier and MPI-Papers routines. Our testing environment consists of a cluster of 8 Dual Pentium Pro processors interconnected via both IOMb/s ethemet network and a TTL-PAPERS prototype. We use two kinds of MPI programs: test programs that p e~o r m few messages into the network and local calculations, and scientific calculation programs that carry out date exchange and local calculations.
In this paper we propose an efficient parallel algorithm with simple static and dynamic schedulin... more In this paper we propose an efficient parallel algorithm with simple static and dynamic scheduling for generating combinations. It can use any number of processors (NP ≤ n − m + 1) in order to generate the set of all combinations of C(n, m). The main characteristic of this algorithm is to require no integer larger than n during the whole computation. The performance results show that even without a perfect load balance, this algorithm has very good performance, mainly when n is large. Besides, the dynamic algorithm presents a good performance on heterogeneous parallel platforms. 1
The aim of this work is to measure the barrier synchronization influence on the overall performan... more The aim of this work is to measure the barrier synchronization influence on the overall performance of several application programs. In order to do that, we use the MPICH implementation version 1.1 of the MPI library. Moreover, we choose two barrier synchronization solutions: the MPICH software solution, i.e. the MPI
Resumo O Plasma é o estado mais comum da matéria no universo, mas é relativamente raro na Terra. ... more Resumo O Plasma é o estado mais comum da matéria no universo, mas é relativamente raro na Terra. O desafio para obtenção de energia através da Fusão Termonuclear Controlada (para muitos a fonte de energia do futuro) depende do aquecimento de ...
Hierarchical clustering algorithms are usually applied to bioinformatic procedures such as groupi... more Hierarchical clustering algorithms are usually applied to bioinformatic procedures such as grouping of genes and proteins with similar structure, reconstruction of evolutionary trees, gene expression analysis, etc [2] [1]. Such applications usually require high memory size and CPU time, therefore, the parallel implementation ot these algorithms is a good option [4].
IEEE Latin America Transactions, 2011
2008 11th IEEE International Conference on Computational Science and Engineering, 2008
Plasma (ionized gas) is the most common state of the matter in the universe but relatively rare o... more Plasma (ionized gas) is the most common state of the matter in the universe but relatively rare on Earth. The quest for the Thermonuclear Controlled Fusion (as an energy source) depends on heating plasmas to temperatures around 10,000,000 Kelvin. The optimization of energy deposition in plasmas by heating them through RF waves in machines called tokamaks has been subject of thorough research (with intense numerical modeling). This work is embedded in this subject with main focus in the run time reduction for the ICANT code, which provides self-consistent calculation of RF current values in a heating antenna model. Typical assessment of an antenna model is made with several runs, each taking several hours, turning this task into day's job. The parallelization was obtained and yielded the same results of the serial version with fair scalability (up to 10 nodes).
lbd.dcc.ufmg.br
Resumo. Este artigo descreve a implementaçao paralela e sequencial do algoritmo de clustering hie... more Resumo. Este artigo descreve a implementaçao paralela e sequencial do algoritmo de clustering hierárquico CURE. Além disso, este artigo mostra a validaç ao da implementaç ao realizada ea avaliaç ao de desempenho da versao paralela.
Genetics and Molecular Biology, 2004
GCLASS is an algorithm which explores small samples of two distinct biological states for finding... more GCLASS is an algorithm which explores small samples of two distinct biological states for finding small sets of genes, which form a feature vector that is enough to separate these two states. A typical sample is a set of 60 microarrays, 30 for each biological state, with several thousand genes. The technique consists of the following: a spreading model defined in the space of small sets of genes studied and centered in each feature vector considered; the designing of optimal linear classifiers under this spreading model; and ranking the designed classifiers, based on their error and robustness relative to the spreading. The feature vectors used in the best classifiers are considered the best feature vectors. Due to the great number of potential feature sets, a parallel implementation is a good option for reducing the procedure execution time. This paper presents a parallel solution of GCLASS and shows some performance results. The experimental results show that the proposed solution provides quasi linear speedup if compared to the sequential implementation. For example, using 60 genes as the complete feature space and 6 genes as the small feature space, our parallel version with 11 processors is approximately 10.98 times faster than the sequential version.
The aim of this work is to measure the barrier synchronization influence on the overall pevormanc... more The aim of this work is to measure the barrier synchronization influence on the overall pevormance of several application programs. In order to do that, we use the MPICH implementation [ I ] version 1.1 of the MPI library. Moreover, we choose two barrier synchronization solutions: the MPICH software solution, i.e. the MPI-Barrier routine, and the TTL-PAPERS hardware solution [ 2 ]. The last one was implemented as a collective operation in MPICH we called MPI-Papers. Finally, we compare the execution time of MPI programs using the MPI-Barrier and MPI-Papers routines. Our testing environment consists of a cluster of 8 Dual Pentium Pro processors interconnected via both IOMb/s ethemet network and a TTL-PAPERS prototype. We use two kinds of MPI programs: test programs that p e~o r m few messages into the network and local calculations, and scientific calculation programs that carry out date exchange and local calculations.
In this paper we propose an efficient parallel algorithm with simple static and dynamic schedulin... more In this paper we propose an efficient parallel algorithm with simple static and dynamic scheduling for generating combinations. It can use any number of processors (NP ≤ n − m + 1) in order to generate the set of all combinations of C(n, m). The main characteristic of this algorithm is to require no integer larger than n during the whole computation. The performance results show that even without a perfect load balance, this algorithm has very good performance, mainly when n is large. Besides, the dynamic algorithm presents a good performance on heterogeneous parallel platforms. 1