Arm usage has substantially grown in the High-Performance Computing (HPC) community. Japanese supercomputer Fugaku, powered by Arm-based A64FX processors, held the top position on the Top500 list between June 2020 and June 2022, currently sitting in the fourth position. The recently released 7th generation of Amazon EC2 instances for compute-intensive workloads (C7 g) is also powered by Arm Graviton3 processors. Projects like European Mont-Blanc and U.S. DOE/NNSA Astra are further examples of Arm irruption in HPC. In parallel, over the last decade, the rapid improvement of genomic sequencing technologies and the exponential growth of sequencing data has placed a significant bottleneck on the computational side. While most genomics applications have been thoroughly tested and optimized for x86 systems, just a few are prepared to perform efficiently on Arm machines. Moreover, these applications do not exploit the newly introduced Scalable Vector Extensions (SVE).This paper presents GenArchBench, the first genome analysis benchmark suite targeting Arm architectures. We have selected computationally demanding kernels from the most widely used tools in genome data analysis and ported them to Arm-based A64FX and Graviton3 processors. Overall, the GenArch benchmark suite comprises 13 multi-core kernels from critical stages of widely-used genome analysis pipelines, including base-calling, read mapping, variant calling, and genome assembly. Our benchmark suite includes different input data sets per kernel (small and large), each with a corresponding regression test to verify the correctness of each execution automatically. Moreover, the porting features the usage of the novel Arm SVE instructions, algorithmic and code optimizations, and the exploitation of Arm-optimized libraries. We present the optimizations implemented in each kernel and a detailed performance evaluation and comparison of their performance on four different HPC machines (i.e., A64FX, Graviton3, Intel Xeon Skylake Platinum, and AMD EPYC Rome). Overall, the experimental evaluation shows that Graviton3 outperforms other machines on average. Moreover, we observed that the performance of the A64FX is significantly constrained by its small memory hierarchy and latencies. Additionally, as proof of concept, we study the performance of a production-ready tool that exploits two of the ported and optimized genomic kernels.

The FM-index is a data structure used in genomics for exact search of input sequences over large reference genomes. Algorithms based on the FM-index show an irregular memory access pattern, resulting in a memory bound problem. We analyze a recent implementation of the FM-index and highlight existing throughput-memory trade-offs, showing that memory requirements limit implementation of large k-steps. We propose COFI, a COmpressed FM-Index for large K-steps. COFI enables a 15-step FM-index using less than 16 GB for a human genome reference of 3 giga base pairs. An algorithm based on this new layout is evaluated on both a Knights Landing (KNL) and an Skylake-based system (SKX). We achieve average speed-ups of 1.46× and 1.39×, respectively, with respect to an state-of-the-art FM-index implementation that is already well optimized.

The open-source and community-supported gem5 simulator is one of the most popular tools for computer architecture research. This simulation infrastructure allows researchers to model modern computer hardware at the cycle level, and it has enough fidelity to boot unmodified Linux-based operating systems and run full applications for multiple architectures including x86, Arm, and RISC-V. The gem5 simulator has been under active development over the last nine years since the original gem5 release. In this time, there have been over 7500 commits to the codebase from over 250 unique contributors which have improved the simulator by adding new features, fixing bugs, and increasing the code quality. In this paper, we give and overview of gem5's usage and features, describe the current state of the gem5 simulator, and enumerate the major changes since the initial release of gem5. We also discuss how the gem5 simulator has transitioned to a formal governance model to enable continued improvement and community support for the next 20 years of computer architecture research.

SVE (Scalable Vector Extension) is Arm's new vector instruction extension targeting high performance workloads. SVE offers many opportunities to optimise compute intensive workloads but, with the availability of SVE-enabled hardware still years away, we have to rely on simulation techniques in order to evaluate our implementations. Working with simulators can be tricky and it comes with many limitations but, used properly, a simulator like Gem5 is a valuable tool that can provide an opportunity to explore the possibilities opened by this new extension. As a use case, we focus our attention on the field of genomics, where the recent advent of high-throughput sequencing machines producing big amounts of genomic data has boosted the interest in efficient approximate string matching and alignment techniques. Genomics algorithms are the key computational building blocks for the downstream data analysis on resequencing projects where hundreds of GBytes of sequenced data are analysed against a reference genome in order to filter sequencing errors and detect potential genomic variation events. The computational requirements and sheer size of the input data used by these applications make them a challenging problem. Fortunately, they also exhibit a high degree of data parallelism, making them good candidates for vectorisation techniques. In this work we explore the unique opportunities that SVE provides in order to exploit the parallelism present in genomics algorithms. We discuss preliminary results, our simulation strategy, some of the obstacles and limitations we faced, and how to work around them in order to obtain meaningful results.

The Modular Microserver DataCentre (M2DC) project investigates, develops and demonstrates a modular, highly-efficient, cost-optimized server architecture composed of heterogeneous microserver computing resources. The resulting server architecture will be able to be tailored to meet requirements from a wide range of application domains. M2DC is built on three main pillars: a flexible server architecture that can be easily customised, maintained and updated; advanced management strategies and system efficiency enhancements (SEE); well-defined interfaces to the surrounding software data centre ecosystem. In this paper, we focus in particular on the thermal management strategies and on the initial benchmarking of the Aarch64 ARM architecture.

Drug repositioning, using known drugs for treating conditions different from those the drug was originally designed to treat, is an important drug discovery tool that allows for a faster and cheaper development process by using drugs that are already approved or in an advanced trial stage for another purpose. This is especially relevant for orphan diseases because they affect too few people to make drug research de novo economically viable. In this paper we present NFFinder, a bioinformatics tool for identifying potential useful drugs in the context of orphan diseases. NFFinder uses transcriptomic data to find relationships between drugs, diseases and a phenotype of interest, as well as identifying experts having published on that domain. The application shows in a dashboard a series of graphics and tables designed to help researchers formulate repositioning hypotheses and identify potential biological relationships between drugs and diseases. NFFinder is freely available at http://nffinder.cnb.csic.es.

UNLABELLEDFrom conformational studies to atomistic descriptions of enzymatic reactions, potential and free energy landscapes can be used to describe biomolecular systems in detail. However, extracting the relevant data of complex 3D energy surfaces can sometimes be laborious. In this article, we present MEPSA (Minimum Energy Path Surface Analysis), a cross-platform user friendly tool for the analysis of energy landscapes from a transition state theory perspective. Some of its most relevant features are: identification of all the barriers and minima of the landscape at once, description of maxima edge profiles, detection of the lowest energy path connecting two minima and generation of transition state theory diagrams along these paths. In addition to a built-in plotting system, MEPSA can save most of the generated data into easily parseable text files, allowing more versatile uses of MEPSA's output such as the generation of molecular dynamics restraints from a calculated path.AVAILABILITY AND IMPLEMENTATIONMEPSA is freely available (under GPLv3 license) at: http://bioweb.cbm.uam.es/software/MEPSA/ CONTACT: pagomez@cbm.csic.es.SUPPLEMENTARY INFORMATIONSupplementary data are available at Bioinformatics online.

BACKGROUND:In the last few years, the Non-negative Matrix Factorization ( NMF ) technique has gained a great interest among the Bioinformatics community, since it is able to extract interpretable parts from high-dimensional datasets. However, the computing time required to process large data matrices may become impractical, even for a parallel application running on a multiprocessors cluster. In this paper, we present NMF-mGPU, an efficient and easy-to-use implementation of the NMF algorithm that takes advantage of the high computing performance delivered by Graphics-Processing Units ( GPUs ). Driven by the ever-growing demands from the video-games industry, graphics cards usually provided in PCs and laptops have evolved from simple graphics-drawing platforms into high-performance programmable systems that can be used as coprocessors for linear-algebra operations. However, these devices may have a limited amount of on-board memory, which is not considered by other NMF implementations on GPU.RESULTS:NMF-mGPU is based on CUDA ( Compute Unified Device Architecture ), the NVIDIA's framework for GPU computing. On devices with low memory available, large input matrices are blockwise transferred from the system's main memory to the GPU's memory, and processed accordingly. In addition, NMF-mGPU has been explicitly optimized for the different CUDA architectures. Finally, platforms with multiple GPUs can be synchronized through MPI ( Message Passing Interface ). In a four-GPU system, this implementation is about 120 times faster than a single conventional processor, and more than four times faster than a single GPU device (i.e., a super-linear speedup).CONCLUSIONS:Applications of GPUs in Bioinformatics are getting more and more attention due to their outstanding performance when compared to traditional processors. In addition, their relatively low price represents a highly cost-effective alternative to conventional clusters. In life sciences, this results in an excellent opportunity to facilitate the daily work of bioinformaticians that are trying to extract biological meaning out of hundreds of gigabytes of experimental information. NMF-mGPU can be used "out of the box" by researchers with little or no expertise in GPU programming in a variety of platforms, such as PCs, laptops, or high-end GPU clusters. NMF-mGPU is freely available at https://github.com/bioinfo-cnb/bionmf-gpu .

We performed a retrospective chart review of the last consecutive 40 patients admitted in our Epilepsy Unit for presurgical evaluation to find out if they met criteria for drug resistant epilepsy according to the recently published consensus definition. 276 drug trials had been performed in the 40 patients. In total, 196 trials were considered “uninformative” versus 80 informative and adequate trials. Finally, a firm diagnosis of drug resistant epilepsy could be made only in 13/40 patients (32.5%, 90% confidence interval for proportion 21.7–45.5%), due to insufficient information regarding previous drug trials. The definition should be spread among general neurologists for earlier and more complete referrals.

Different approaches have been proposed over the last few years for improving holistic methods for face recognition. Some of them include color processing, different face representations and image processing techniques to increase robustness against illumination changes. One of the most successful strategies has shown to be the use of Gabor representation of the images. There has been also some research about the combination of different recognition methods, both at the feature and score levels. In this paper, we propose an effective combination scheme that is able to improve a single holistic method by fusing the recognition scores obtained from both natural face images and their Gabor representations. We have evaluated this scheme using some of the best known holistic approaches in the context of the Face Recognition Grand Challenge (FRGC). Results show at least 10% improvements in all cases. Moreover, this scheme also works when the scores are obtained from two different methods whenever one of them uses natural images and the other their Gabor representation. These results suggest that some complementariness exists between both representations, which can be easily exploited by fusion at the score level.

We have explored in this paper a framework to test in a quantitative manner the stability of different endmember extraction and spectral unmixing algorithms based on the concept of Consensus Clustering. The idea is to investigate if the sensibility of those algorithms to the number of endmembers can be used to estimate this parameter itself. Preliminary results on synthetic data reveal that the proposed scheme, which can be implemented efficiently in parallel, can compete with state-of-the-art schemes.

UNLABELLED:The aim of this study was to investigate if sequential analysis of BOLD signal changes induced by seizures is useful for preoperative identification of the site of seizure onset in patients with pharmaco-resistant focal epilepsy. METHOD:We analyzed BOLD raw data from 5 patients with focal medically refractory epilepsy who experienced partial seizures during fMRI as part of a preoperative evaluation. To sequence the changes in BOLD signal seizure-induced, each seizure epoch was divided into groups of five consecutive images (ten-second blocks). t-maps were calculated continuously from 120 s before the onset of clinical/EEG seizure onwards by comparing two consecutive groups of five images. Time lag between each comparison was 2 s. Relative changes in BOLD signal between two consecutive groups of five images along the seizure epoch were determined. Results were compared with those of subtraction ictal SPECT coregistered with MRI (SISCOM) and intracranial EEG (2 patients). RESULTS:A typical seizure was registered in each patient. After sequential analysis, a well-localized and statistically significant (t: 7-14) area of signal increase was consistently found at seizure initiation in each patient. This area invariably preceded the onset of clinical/electrical seizure by several seconds (6-52 s); was concordant with SISCOM results in all but one patient; and overlapped with the ictal onset zone determined by intracranial EEG in those 2 patients who underwent invasive-EEG recordings. Complete resection of this initial area of signal increase resulted in seizure remission. Three out of four patients who underwent epilepsy surgery remained seizure-free. CONCLUSION:Sequential analysis of ictal-fMRI data may be useful to precisely and non-invasively delineate the ictal onset zone within the brain; and provide insights into the cerebral substrates involved in the generation and propagation of seizures.

SummaryPurpose: The aim of this study was to introduce a new approach for analysis of functional magnetic resonance imaging (fMRI) data in order to illustrate the temporal development of the blood oxygenation level–dependent (BOLD) signal changes induced by epileptic seizures.Method: In order to sequentially analyze the fMRI images acquired during epileptic seizures, a continuous series of echo planar imaging (EPI) scans covering the complete period of a seizure was acquired. Data were segmented into 10‐s blocks. Each block, representing a unique experimental condition, was contrasted with a neutral (no seizure) baseline condition. Visual comparison of the activations from one block to the next highlighted the course of activations and deactivations during the seizure event. This analysis was applied to three independent seizures of one patient with peri‐rolandic epilepsy secondary to chronic encephalitis: one seizure before epilepsy surgery and two after unsuccessful tailored resection. Observations were compared to results from invasive subdural electroencephalography (EEG) monitoring, single‐photon emission computed tomography (SPECT) coregistered to MRI (SISCOM), and independent component analysis (ICA), a model‐free method of BOLD‐signal analysis.Results: The initial increase in BOLD signal occurred 10–40 s before clinical onset in the same location compared to the seizure‐onset zone determined by invasive subdural evaluation and SISCOM. Sequential involvement of cortical and subcortical structures was in agreement with SISCOM, intracranial EEG recordings, and ICA results.Discussion: In selected patients, sequential analysis of changes in BOLD signal induced by epileptic seizures might represent a useful approach for investigating the temporal development of brain activity during epileptic seizures, thereby allowing imaging of those cerebral structures involved in seizure generation and propagation.

Monocyte-derived dendritic cells (DCs) loaded with heat-inactivated HIV are used in therapeutic immunizations. It is not known whether they migrate in vivo to lymph nodes. We used an In-111-oxine-labeled DC (ILDC) method to visualize the migration of DCs. The activity, time and incubation medium were investigated to obtain the highest cellular viability and radiolabeling yield. A trypan-blue exclusion test was used to determine the cellular viability. In five patients, 2 x 10(6) ILDCs were injected subcutaneously in the arm. An initial dynamic study was performed during the first 5 min after injection. This was followed by static acquisitions at several time points, using a high-resolution (general electric) gamma-camera and quantifying the activity at regions of interest drawn on the injection point. The sensitivity of the gamma-camera was evaluated. The highest number of viable DCs (>83%) and the best radiolabeling yield (>70%) were obtained with 1.11 MBq In-111-oxine, after 10 min of incubation at 37 degrees C in sodium chloride solution 0.9%. We did not observe migration of ILDCs to local lymph nodes in any patient. However, focal uptake at the place of injection continued during the study period. We observed a higher than expected loss of activity from the injection point (median A(t)/A(0) = 0.60 at day 2), which correlated with an increase in total cytotoxic T lymphocytes (CD8(+) and granzyme B+ cells) in the lypmphoid tissue observed after immunization (R-2 = 0.92, p = 0.03). If more than 20,000 ILDCs had migrated, they could have been detected. In future trials, a higher number of DCs or alternative methods should be used to assess the migration of DCs to lymph nodes.

Graphics Processing Units (GPUs) have become highly parallel and programmable systems used as commodity data-parallel coprocessors. Moreover manufacturers have developed new software interfaces that facilitate their use. Thus, new compilation strategies that enable automatic mapping of sequential code would very likely arise in the near future. To open this path, we need to define some performance metrics or heuristics that steer the mapping, as well as to develop new tools that support such metrics or heuristics. In this paper we sketch a new compiler intermediate representation built on top of the hierarchy of computational kernels provided by the XARK compiler framework. In addition, we outline several compilation strategies based on such intermediate representation and present several experiments aimed at analyzing the relative importance of the two principal factors that NVIDIA reports to influence the most on performance of their GPUs: occupancy and memory hierarchy usage. Our results show that taking decisions based only on the optimization of the memory system provide the best performance, even if some resources are underutilized.

This chapter focuses on mapping hyperspectral imaging algorithms to graphics processing units (CPU). The performance and parallel processing capabilities of these units, coupled with their compact size and relative low cost, make them appealing for onboard data processing. We begin by giving a short review of GPU architectures. We then outline a methodology for mapping image processing algorithms to these architectures, and illustrate the key code transformation and algorithm trade-offs involved in this process. To make this methodology precise, we conclude with an example in which we map a hyperspectral endmember extraction algorithm to a modern GPU.

This paper outlines the Nallatech Accelerator Layer (NAL) and its relationship to Intel’s Accelerator Abstraction Layer. The NAL is looked at in its academicion Layer. The NAL is looked at in its academic context. Hardware platforms that support the NAL are discussed: the Nallatech H101, the Intel FSB-FPGA Module and the BenOne PCIe. The Intel QuickAssist Technology initiative and its associated Accelerator Abstraction Layer (AAL) are introduced.ion Layer (AAL) are introduced. To demonstrate the NAL system, two closed-form expressions are implemented. These functions are singleprecision floating-point, and make use of arithmetic operations and elementary functions. The functions selected were the probability density function (PDF) and the Black-Scholes-Merton options pricing formula (BSM). These functions were implemented on a dual-core Opteron, a Nallatech H101 card using the NAL, a Nallatech BenOne PCIe card, an NVIDIA G80 using CUDA and a Cell BE system. The GPU system showed the best silicon performance for the implementation of these kernels. Including data transfer times, the BenOne PCIe had the highest performance.

Postictal psychosis (PIP) represents 25% of the psychoses seen in epileptic patients. A high frequency of bilateral independent epileptiform activity has been observed in patients with PIP. The objective of this study was to determine the frequency of PIP in patients with temporal lobe epilepsy (TLE) who underwent video-EEG monitoring and to investigate possible differences between PIP and control patients.Methods: Clinical, electroencephalographic and neuroimaging data of 5 PIP patients with TLE were compared with data of 50 patients with TLE without psychotic antecedents. Patients with a past history of interictal psychosis were excluded.Results: From 55 patients, 5 were patients with PIP and 50 controls. 31 (62%) were men, 9 (16.4%) had a previous history of encephalitis and 6 (10.9%) Of Status epilepticus. The mean age was 42.2 years (S.D. 12.93). Mean age at epilepsy onset was 16.95 years (S.D. 12.93) and mean seizure frequency 5 seizures/month (S.D. 1.87). The frequency of PIP was 5/55 (9.1%). Previous history Of Status epilepticus was more frequent in PIP patients than in controls (p: 0.019). PIP patients more frequently had a non-lateralizing ictal EEG than controls (p: 0.001). Bitemporal lobe dysfunction revealed by neuropsychological studies was greater than expected by the observed lesion on MRI Studies in patients with PIP. Moreover, the presurgical Study was less conclusive in PIP than in control patients (p: 0.049).Conclusions: PIP is observed in up to 9% Of patients with TLE undergoing video-EEG monitoring and most often develops in patients with bitemporal lobe dysfunction. (c) 2008 British Epilepsy Association. Published by Elsevier Ltd. All rights reserved.

Hyperspectral analysis algorithms exhibit inherent parallelism at multiple levels, and map nicely on high performance systems such as massively parallel clusters and networks of computers. Unfortunately, these systems are generally expensive and difficult to adapt to onboard data processing scenarios, in which low-weight and low-power integrated components are desirable to reduce mission pay-load. An exciting new development in this field is the emergence of programmable graphics hardware. Driven by the ever-growing demands of game industry, graphics processing units (GPUs) have evolved from expensive, application-specific units into highly parallel and programmable systems which can satisfy extremely high computational requirements at low cost. In this paper, we investigate GPU-based implementations of a morphological endmember extraction algorithm, which is used as a representative case study of joint spatial/spectral techniques for hyperspectral analysis. The proposed implementations are quantitatively compared and assessed in terms of both endmember extraction accuracy and parallel efficiency. Combined, these parts offer a thoughtful perspective on the potential and emerging challenges of implementing hyperspectral imaging algorithms on commodity graphics hardware.