Dense linear algebra operations appear very frequently in high-performance computing (HPC) applications, rendering their performance crucial to achieve optimal scalability. As many modern HPC clusters contain multi-GPU nodes, BLAS operations are frequently offloaded on GPUs, necessitating the use of optimized libraries to ensure good performance. Unfortunately, multi-GPU systems are accompanied by two significant optimization challenges: data transfer bottlenecks as well as problem splitting and scheduling in multiple workers (GPUs) with distinct memories. We demonstrate that the current multi-GPU BLAS methods for tackling these challenges target very specific problem and data characteristics, resulting in serious performance degradation for any slightly deviating workload. Additionally, an even more critical decision is omitted because it cannot be addressed using current scheduler-based approaches: the determination of which devices should be used for a certain routine invocation. To address these issues we propose a model-based approach: using performance estimation to provide problem-specific autotuning during runtime. We integrate this autotuning into an end-to-end BLAS framework named PARALiA. This framework couples autotuning with an optimized task scheduler, leading to near-optimal data distribution and performance-aware resource utilization. We evaluate PARALiA in an HPC testbed with 8 NVIDIA-V100 GPUs, improving the average performance of GEMM by 1.7× and energy efficiency by 2.5× over the state-of-the-art in a large and diverse dataset and demonstrating the adaptability of our performance-aware approach to future heterogeneous systems.

Containers improve the efficiency in application deployment and thus have been widely utilised on Cloud and lately in High Performance Computing (HPC) environments. Containers encapsulate complex programs with their dependencies in isolated environments making applications more compatible and portable. Often HPC systems have higher security levels compared to Cloud systems, which restrict users’ ability to customise environments. Therefore, containers on HPC need to include a heavy package of libraries making their size relatively large. These libraries usually are specifically optimised for the hardware, which compromises portability of containers.Per contra, a Cloud container has smaller volume and is more portable. Furthermore, containers would benefit from orchestrators that facilitate deployment and management of containers at a large scale. Cloud systems in practice usually incorporate sophisticated container orchestration mechanisms as opposed to HPC systems. Nevertheless, some solutions to enable container orchestration on HPC systems have been proposed in state of the art. This paper gives a survey and taxonomy of efforts in both containerisation and its orchestration strategies on HPC systems. It highlights differences thereof between Cloud and HPC. Lastly, challenges are discussed and the potentials for research and engineering are envisioned.

Nowadays, machine learning (ML), especially deep learning(DL) methods, provide ever more real-life solutions. However, the lack of training data is often a crucial issue for these learning algorithms, the performance accuracy of which relies on the amount and the quality of the available data. This is particularly true when applying ML/DL based methods for specific areas e.g. material characteristics identification, as it requires huge cost of time and manual power getting observational data from real life. In the mean while, simulations on HPC have already been commonly used in computational science due to the fact that it has the ability of generating sufficient and noise free data, which can be used for training the ML/DL based models. However, in order to achieve accurate simulation results the input parameters usually have to be determined and validated by a large number of tests. Furthermore, the evaluation and validation of such input parameters for the simulation often require a deep understanding of the domain specific knowledge, software and programming skills, which can in turn be solved by ML/DL based methods. In this paper, a novel hybrid workflow combining a multi-task neural network and the simulation on high performance computers(HPC) is proposed, which can address the problem of data sparsity and reduce the demand for expertise, resources, and time in determining the validated parameters for simulation. This work is demonstrated through experiments on determination of material behaviors, and the results prove a promising performance (MSE = 0.0386) through this workflow.

Due to the increasing demand for fish consumption, sustainable fishery become more and more challenging. To prevent from overfishing, massive data in open sea fishing have been collected and analyzed to achieve efficient management of fishery. Still, it is extremely difficult for fishers and fishery managers to exploit available data for accurate prediction, because of their limited data processing capacities, and the overall lack of adequate database systems [1].The goal of this work is therefore to analyze the relationship between data collected from all sensors installed on-board fishing vessels and catch weight, to better support generating a map showing likely fishing effort allocation. To do so, we train neural networks to predict catch weight using all available data from sensors on fishing vessels. The raw data are pre-processed using random sampling techniques to be fed into a neural network for training. A multi-layer perceptron (MLP) neural network is proposed as the baseline. We propose a data augmentation method and a training strategy in order to optimize the prediction accuracy of the model. Our data augmentation method conducts random sampling of the original data multiple times, which reduces the root mean square error (RMSE) by 15.8%, as compared with the results obtained by the model trained without data augmentation. Our training strategy works well to further optimize the prediction accuracy of the model trained with an augmented dataset, which significantly decreased the RMSE by 11. 2%. To the best of our knowledge, this is the first study on the catch weight prediction using neural networks.

Monitoring of High Performance Computing (HPC) platforms is critical to successful operations, can provide insights into performance-impacting conditions, and can inform methodologies for improving science throughput. However, monitoring systems are not generally considered core capabilities in system requirements specifications nor in vendor development strategies. In this paper we present work performed at a number of large-scale HPC sites towards developing monitoring capabilities that fill current gaps in ease of problem identification and root cause discovery. We also present our collective views, based on the experiences presented, on needs and requirements for enabling development by vendors or users of effective sharable end-to-end monitoring capabilities.

We propose a competence partitioning strategy for Web search result presentation: the unmodified head of a ranked result list is combined with a clustering of documents from the result list tail. We identify two principles to which such a clustering must adhere to improve the user's search experience: (1) Avoid the unwanted effect of query aspect repetition, which is called shadowing here. (2) Avoid extreme clusterings, i.e., neither the number of cluster labels nor the number of documents per cluster should exceed the size of the result list head. We present measures to quantify the shadowing effect, and with Faceted Clustering we introduce an algorithm that optimizes the identified principles. The key idea of Faceted Clustering is a dynamic, user-controlled reorganization of a clustering, similar to a faceted navigation system. We report on evaluations using the AMBIENT corpus and demonstrate the potential of our approach by a comparison with two well-known clustering search engines.

The paper addresses the missing user acceptance of web search result clustering. We report on selected analyses and propose new concepts to improve existing result clustering approaches. Our findings in a nutshell are: 1. Don't compete with a search engine's top hits. In response to a query we presume search engines to return an optimal result list in the sense of the probabilistic ranking principle: documents that are expected by the majority of users are placed on top and form the result list head. We argue that, with respect to the top results, it is not beneficial to replace this established form of result presentation. 2. Improve document access in the result list tail. Documents that address the information need of "minorities" appear at some position in the result list tail. Especially for ambiguous and multi-faceted queries we expect this tail to be long, with many users appreciating different documents. In this situation web search result clustering can improve user satisfaction by reorganizing the long tail into topic-specific clusters. 3. Avoid shadowing when constructing cluster labels. We show that most of the cluster labels that are generated by current clustering technology occur within the snippets of the result list head--an effect which we call shadowing. The value of such labels for topic organization and navigating within a clustering of the entire result list is limited. We propose and analyze a filtering approach to significantly alleviate the label shadowing effect.