Javier Cuenca

This work proposes a hierarchical approach to reduce the training time of task-based routines by reusing previously obtained autotuning information. This approach has been integrated into a working prototype of Chameleon, a dense linear algebra software whose tile-based routines are executed on the available computational resources by means of a runtime system. The results show that this approach provides a high degree of scalability to the entire self-optimization process, achieving a reduction in training time of up to 80% and an appropriate selection of values for the adjustable parameters.

In this work, an automatic optimisation approach for parallel routines on multi-GPU systems is presented. Several inter-GPU communication libraries (such as CUDA-Aware MPI or NCCL) are used with a set of routines to perform the numerical operations among the GPUs located on the compute nodes. The main objective is the selection of the most appropriate communication library, the number of GPUs to be used and the workload to be distributed among them in order to reduce the cost of data movements, which represent a large percentage of the total execution time. To this end, a hierarchical modelling of the execution time of each routine to be optimised is proposed, combining experimental and theoretical approaches. The results show that near-optimal decisions are taken in all the scenarios analysed.

PARCSIM is a parallel software simulator that allows a user to capture, through a graphical interface, matrix algorithm schemes that solve scientific problems. With this tool, the user can analyse the execution times that would be obtained by using different spatio-temporal mapping of computational tasks on available computational units, parallelism parameters and computational libraries. Furthermore, for complex problem models, the self-optimization engine incorporated in this tool analyses the huge tree of possible calculations grouping and mapping strategies in search of the choice that makes the best use of the available hardware resources. This tool also offers polyalgorithmic resolution by making automatically the best decision between different software approaches to solve a given problem on the hardware system available. This work shows the usefulness of this simulator to efficiently solve hierarchical problems constructed from previously modelled subproblems. This task is performed by reusing, in a scalable way, the optimization information of these subproblems to establish the best execution configuration for the composite problem.

In the last years, the huge amount of data available in many disciplines makes the mathematical modeling, and, more concretely, econometric models, a very important technique to explain those data. One of the most used of those econometric techniques is the Vector Autoregression Models (VAR) which are multi-equation models that linearly describe the interactions and behavior of a group of variables by using their past. Traditionally, Ordinary Least Squares and Maximum likelihood estimators have been used in the estimation of VAR models. These techniques are consistent and asymptotically efficient under ideal conditions of the data and the identification problem. Otherwise, these techniques would yield inconsistent parameter estimations. This paper considers the estimation of a VAR model by minimizing the difference between the dependent variables in a certain time, and the expression of their own past and the exogenous variables of the model (in this case denoted as VARX model). The solution of this optimization problem is approached through hybrid metaheuristics. The high computational cost due to the huge amount of data makes it necessary to exploit High-Performance Computing for the acceleration of methods to obtain the models. The parameterized, parallel implementation of the metaheuristics and the matrix formulation ease the simultaneous exploitation of parallelism for groups of hybrid metaheuristics. Multilevel and heterogeneous parallelism are exploited in multicore CPU plus multiGPU nodes, with the optimum combination of the different parallelism parameters depending on the particular metaheuristic and the problem it is applied to.

Aims: Hypertrophic cardiomyopathy (HCM) is often accompanied by increased trabeculated myocardium (TM)—which clinical relevance is unknown. We aim to measure the left ventricular (LV) mass and proportion of trabeculation in an HCM population and to analyze its clinical implication. Methods and Results: We evaluated 211 patients with HCM (mean age 47.8 ± 16.3 years, 73.0% males) with cardiac magnetic resonance (CMR) studies. LV trabecular and compacted mass were measured using dedicated software for automatic delineation of borders. Mean compacted myocardium (CM) was 160.0 ± 62.0 g and trabecular myocardium (TM) 55.5 ± 18.7 g. The percentage of trabeculated myocardium (TM%) was 26.7% ± 6.4%. Females had significantly increased TM% compared to males (29.7 ± 7.2 vs. 25.6 ± 5.8, p < 0.0001). Patients with LVEF < 50% had significantly higher values of TM% (30.2% ± 6.0% vs. 26.6% ± 6.4%, p = 0.02). Multivariable analysis showed that female gender and neutral pattern of hypertrophy were directly associated with TM%, while dynamic obstruction, maximal wall thickness and LVEF% were inversely associated with TM%. There was no association between TM% with arterial hypertension, physical activity, or symptoms. Atrial fibrillation and severity of hypertrophy were the only variables associated with cardiovascular death. Multivariable analysis failed to demonstrate any correlation between TM% and arrhythmias. Conclusions: Approximately 25% of myocardium appears non-compacted and can automatically be measured in HCM series. Proportion of non-compacted myocardium is increased in female, non-obstructives, and in those with lower contractility. The amount of trabeculation might help to identify HCM patients prone to systolic heart failure.

VAR models are a type of multi-equation model that have been widely applied in econometrics. With the arrival of Big Data, huge amounts of data are being collected in numerous fields, making feasible the application of these kind of statistical models. Tools exist to tackle this problem, but the large amount of data, along with the availability of computational techniques and high performance systems, advise an in-depth analysis of the computational aspects of VAR, so large models can be solved efficiently with today's computational systems. This work aims to solve a VAR model by obtaining the coefficients through heuristic and metaheuristic algorithms, minimizing one parameter criterion, and also to compare with those coefficients obtained by OLS. Furthermore, we consider different approaches to reduce the time required to find the model like using matrix decompositions (QR or LQ), exploiting matrix structure, using high performance linear algebra subroutines (BLAS and LAPACK) or parallel metaheuristics.