Photographer Zé Carlos de Andrade

SBAC-PAD 2023

35th IEEE International Symposium on Computer Architecture and High Performance Computing


October 17-20, 2023 — Porto Alegre, Brazil



Program


Program Overview



Program Details



Tuesday, October 17


08:00

Foyer

Registration


09:00

Auditorium

HPC/DL

09:00 - 09:15

Opening Session

09:15 - 10:30

Developing Efficient Scientific Gateways for Supercomputer Environments Supported by Deep Learning Models

Speaker: Prof. Carla Osthoff (LNCC)

The scientific gateway BioinfoPortal for bioinformatics applications is hosted in the National Laboratory for Scientific Computing (LNCC) and is coupled to the Santos Dumont (SDumont) supercomputer environment. BioinfoPortal offers a catalog of bioinformatics computational software that benefits from the parallel and distributed architecture offered by LNCC. Task submissions consume SDumont nodes shared by other supercomputer users; then, it is required to set the best configuration, defined by the best choice of the number of threads/nodes, to be allocated for every task submission. This talk presents research analysis using Deep Neural Networks to estimate the computational time required to execute bioinformatics software in several scenarios using a pre-configured number of nodes and threads. We aim to demonstrate the computational behavior of software in Bioinfoportal and which computational scenario can be chosen to execute software in SDumont efficiently. Results support that the neural networks can predict the most representative variable and identify the configuration with the lowest computational time. This way, BioinforPortal consuming time can lead to an efficient and green gateway, increasing Santos Dumont Supercomputing execution job throughput and decreasing job execution queue waiting time.


09:00

Room 501

WAMCA

09:00 - 09:50

Session 1 - Compilation and Performance

Chair: Claude Tadonki (MINES ParisTech - PSL)

DASS: Dynamic Adaptive Sub-Target Specialization PDF

Tyler Gobran (University of Alberta, Canada), João P. L. de Carvalho (University of Alberta, Canada), Christopher Barton (IBM Corporation, Canada), Quinn Pham (University of Alberta, Canada), José Nelson Amaral (University of Alberta, Canada), Nemanja Ivanovic (IBM Corporation, Canada)

Analyzing C++ Stream Parallelism in Shared-Memory when Porting to Flink and Storm PDF

Renato B. Hoffmann (Pontifical Catholic University of Rio Grande do Sul), Leonardo G. Faé (Pontifical Catholic University of Rio Grande do Sul), Isabel H. Manssour (Pontifical Catholic University of Rio Grande do Sul), Dalvan Griebler (Pontifical Catholic University of Rio Grande do Sul)

09:50 - 10:40

Session 2 - Shared Memory Management

Chair: Claude Tadonki (MINES ParisTech - PSL)

A Source-to-Source NUMA Profiling Approach PDF

Leticia S.F. Machado (Universidade Federal de São Carlos), Claude Tadonki (Centre de Recherche en Informatique (CRI) Mines ParisTech - PSL, Fontainebleau, France), Hermes Senger (Universidade Federal de São Carlos)

Energy Consumption Analysis of Instruction Cache Prefetching Methods PDF

Morteza Baradaran (University of Virginia, USA), Ali Ansari (Sharif University of Technology, Iran), Mohammad Sadrosadati (Sharif University of Technology, Iran), Hamid Sarbazi-Azad (Sharif University of Technology, Iran)


09:00

WSCAD Tutorials (1, 2, and 3) Event in Portuguese

09:00 - 10:30

Processamento e Análise de Big Data e Aplicação de Algoritmos de Machine Learning através da utilização da Plataforma HPCC Systems PDF

Authors: Mauro Marques (LexisNexis Risk Solutions), Alysson Oliveira (LexisNexis Risk Solutions)

Room 502

Ao longo do minicurso, os participantes terão a oportunidade de conhecer os conceitos essenciais de processamento e análise de volumes massivos de dados (Big Data) e o processo de desenvolvimento de um serviço de consulta fazendo uso da plataforma opensource composta por um Cluster Computacional de Alto Desempenho (HPCC Systems) e, também, a utilização de algoritmos de Aprendizado de Máquina, bem como terão a possibilidade de aplicar os conhecimentos adquiridos em um ambiente de treinamento disponibilizado em sala de aula.

09:00 - 10:30

Getting up and Running with the OpenMP Cluster Programming Model PDF

Authors: Emilio Francesquini (Universidade Federal do ABC), Herve Yviquel (UNICAMP) Marcio Pereira (UNICAMP), Sandro Rigo (UNICAMP), Guido Araujo (UNICAMP)

Room 514

The authors will present the new OpenMP Cluster (OMPC) distributed programming model in this tutorial. The OMPC runtime allows the programmer to annotate their code using OpenMP target offloading directives and run the application in a distributed environment seamlessly using a task-based programming model. OMPC is responsible for scheduling tasks to available nodes, transferring input/output data between nodes, and triggering remote execution all the while handling fault tolerance. The runtime leverages the LLVM infrastructure and is implemented using the well-known MPI library.

09:00 - 10:30

Além do Básico: Otimizando Aplicações Paralelas em Arquiteturas Modernas PDF

Authors: Arthur Lorenzon (UFRGS)

Room 515

Este minicurso ensina as ferramentas e técnicas fundamentais para otimizar o desempenho e consumo de energia de aplicações escritas em linguagens C/C++, explorando a combinação de diferentes tipos de processadores, como CPUs e GPUs. O minicurso não exige nenhum conhecimento prévio em interfaces de programação paralela, mas requer alguns conhecimentos básicos em C/C++, sistemas operacionais e arquitetura de computadores.


10:30

Foyer

Coffee-break


11:00

Auditorium

HPC/DL

11:00 - 12:00

Paper Session 1

Chair: Flavia Bernardini (UFF) / Edson Borin (UNICAMP)

Assessing the Performance of an Architecture-Aware Optimization Tool for Neural Networks PDF

Raúl Marichal (Universidad de la República, Uruguay), Ernesto Dufrechou (Universidad de la República, Uruguay), Pablo Ezzatti (Universidad de la República, Uruguay)

An Exploratory Study of Deep Learning for Predicting Computational Tasks Behavior in HPC Systems PDF

Alexandre Henrique Lopes Porto (National Laboratory for Scientific Computing), Micaella Coelho (National Laboratory for Scientific Computing), Kary Ocaña (National Laboratory for Scientific Computing), Carla Osthoff (National Laboratory for Scientific Computing), Francieli Boito (University of Bordeaux, CNRS, Bordeaux INP, INRIA, LaBRI), Douglas O. Cardoso (Smart Cities Research Center, Polytechnic Institute of Tomar)


11:00

Room 501

WAMCA

11:00 - 11:40

Session 3 - Accelerators and Checkpointing

Chair: Claude Tadonki (MINES ParisTech - PSL)

Multi-GPU accelerating strategies of Ant Colony Optimization algorithms using Rank Based and Strong Elitist versions PDF

Juan Aedo (Universidad de La Frontera), Mariela González-Flores (Universidad de Aysén), Andrés Ávila (Universidad de La Frontera)


11:00

WSCAD Tutorials (1, 2, and 3) Event in Portuguese

11:00 - 12:00

Processamento e Análise de Big Data e Aplicação de Algoritmos de Machine Learning através da utilização da Plataforma HPCC Systems PDF

Authors: Mauro Marques (LexisNexis Risk Solutions), Alysson Oliveira (LexisNexis Risk Solutions)

Room 502

Ao longo do minicurso, os participantes terão a oportunidade de conhecer os conceitos essenciais de processamento e análise de volumes massivos de dados (Big Data) e o processo de desenvolvimento de um serviço de consulta fazendo uso da plataforma opensource composta por um Cluster Computacional de Alto Desempenho (HPCC Systems) e, também, a utilização de algoritmos de Aprendizado de Máquina, bem como terão a possibilidade de aplicar os conhecimentos adquiridos em um ambiente de treinamento disponibilizado em sala de aula.

11:00 - 12:00

Getting up and Running with the OpenMP Cluster Programming Model PDF

Authors: Emilio Francesquini (Universidade Federal do ABC), Herve Yviquel (UNICAMP) Marcio Pereira (UNICAMP), Sandro Rigo (UNICAMP), Guido Araujo (UNICAMP)

Room 514

The authors will present the new OpenMP Cluster (OMPC) distributed programming model in this tutorial. The OMPC runtime allows the programmer to annotate their code using OpenMP target offloading directives and run the application in a distributed environment seamlessly using a task-based programming model. OMPC is responsible for scheduling tasks to available nodes, transferring input/output data between nodes, and triggering remote execution all the while handling fault tolerance. The runtime leverages the LLVM infrastructure and is implemented using the well-known MPI library.

11:00 - 12:00

Além do Básico: Otimizando Aplicações Paralelas em Arquiteturas Modernas PDF

Authors: Arthur Lorenzon (UFRGS)

Room 515

Este minicurso ensina as ferramentas e técnicas fundamentais para otimizar o desempenho e consumo de energia de aplicações escritas em linguagens C/C++, explorando a combinação de diferentes tipos de processadores, como CPUs e GPUs. O minicurso não exige nenhum conhecimento prévio em interfaces de programação paralela, mas requer alguns conhecimentos básicos em C/C++, sistemas operacionais e arquitetura de computadores.


12:00

Lunch (Not included)


13:30

Auditorium

HPC/DL

13:30 - 15:00

Paper Session 2

Chair: Flavia Bernardini (UFF) / Edson Borin (UNICAMP)

PINNProv: Provenance for Physics-informed Neural Networks PDF

Lyncoln S. de Oliveira (Federal University of Rio de Janeiro, Brazil; Universidade Federal Fluminense, Brazil), Liliane Kunstmann (Federal University of Rio de Janeiro, Brazil), Débora Pina (Federal University of Rio de Janeiro, Brazil), Daniel de Oliveira (Universidade Federal Fluminense, Brazil), Marta Mattoso (Federal University of Rio de Janeiro, Brazil)

Exploring Federated Learning to Trace Depression in Social Media with Language Models PDF

Arthur Vasconcelos (Universidade Federal Fluminense, Brazil), Lúcia Drummond (Universidade Federal Fluminense, Brazil), Rafaela Brum (Universidade Federal Fluminense, Brazil), Aline Paes (Universidade Federal Fluminense, Brazil)

Computing Seismic Attributes with Deep-Learning Models PDF

Nícolas Hecker (Universidade Estadual de Campinas (UNICAMP), Brazil), Otávio O. Napoli (Universidade Estadual de Campinas (UNICAMP), Brazil), Carlos A. Astudillo (Universidade Estadual de Campinas (UNICAMP), Brazil), João Paulo Navarro (NVIDIA, Brazil), Alan Souza (Petróleo Brasileiro S.A. (PETROBRAS), Brazil), Daniel Miranda (Petróleo Brasileiro S.A. (PETROBRAS), Brazil), Leandro A. Villas (Universidade Estadual de Campinas (UNICAMP), Brazil), Edson Borin (Universidade Estadual de Campinas (UNICAMP), Brazil)


13:30

Room 501

WCC

13:30 - 13:40

Opening Session

13:40 - 14:20

Bridging the Resource Chasm: Sustainable Structured Parallelism for HPC and Cloud Applications

Speaker: Horacio González-Vélez

In the ever-evolving landscape of cloud computing and high-performance applications, this talk delves into the critical juncture of HPC and cloud, with special emphasis on key challenges such as Power Consumption, Speed and Energy of Data Movement, Fault Tolerance, and Billion-Way Parallelism. This keynote explores the intersection of cloud computing and high-performance computing from a programmability perspective, shedding light on how they can harmonise to propel us into the next era of sustainability. The discussion will delve into the challenges and opportunities that arise when pushing the boundaries of programmability and performance in the cloud, particularly in the context of Power, Data Movement, Fault Tolerance, and Billion-Way Parallelism. It will address strategies for optimising resources to meet the demanding requirements of high-performance applications, including those in fields such as scientific research and data-intensive workloads, emphasising the importance of programmability, sustainability, energy efficiency, and cost-effectiveness.

14:20 - 15:00

Technical Session 1

Chair: Márcio Castro (UFSC) / Philippe O. A. Navaux (UFRGS)

Harnessing Cloud Computing for Geophysical Exploration PDF

Rodrigo C. Machado (UFRGS), Cristiano A. Kunas (UFRGS), Arthur F. Lorenzon (UFRGS), Philippe O. A. Navaux (UFRGS)

Exploring the Serverless First Strategy in Cloud Application Development PDF

Adriano Prado Cavalheiro (UNIPAMPA), Claudio Schepke (UNIPAMPA)


13:30

WSCAD Tutorials (4, 5, and 6) Event in Portuguese

13:30 - 15:00

Are you Root? Reproducible Experiments in User Space PDF

Authors: Vinicius Garcia Pinto (UFRGS), Lucas Nesi (UFRGS), Lucas Mello Schnorr (UFRGS)

Room 502

Plataformas computacionais para processamento de alto desempenho usualmente são compartilhadas por usuários com demandas variadas. Com frequência aplicações legadas ou aquelas que possuem requisitos e dependências muito específicas sobrecarregam os administradores do sistema ou simplesmente tem sua execução inviabilizada. Uma solução passa a ser configurar, compilar e instalar a aplicação e as respectivas dependências inteiramente em espaço de usuário. Entretanto, tal tarefa é custosa e propensa a erros, o que motiva a adoção de soluções automatizadas. Neste minicurso, propomos a apresentação de duas soluções que permitem a instalação de pacotes inteiramente em espaço de usuário de maneira reprodutível e compartilhável.

13:30 - 15:00

Abordagem Gamificada para Introdução em Arquitetura de Hardware Evolutivo

Authors: Bernardo Cunha (PUC Minas), Carlos Augusto Martins (PUC Minas)

Room 514

Este minicurso consiste em 3 partes, de aproximadamente 50 minutos. Inicialmente, haverá exposição dos conceitos básicos para compreensão do tema, utilizando abordagem gamificada. Na segunda parte, os participantes colocarão estas novas informações em prática, "hackeando" o sistema evolutivo, experimentando variações nos parâmetros de entrada e de configuração. Por último, será escolhida a melhor abordagem a ser implementada em um próximo encontro (virtual), é demonstrado como o sistema selecionado pode ser implementado utilizando um dispositivo de computação reconfigurável (FPGA).

13:30 - 15:00

Armazenamento - Melhores Práticas - Construindo Sistemas de Armazenamento com a Performance Adequada PDF

Authors: Guilherme Friol

Room 515

Este minicurso abordará as principais formas de conexões existentes, tais como SATA, SCSI, SAS, iSCSI, Fibre Channel, entre outras. Os participantes poderão compreender as vantagens e desvantagens de cada uma dessas opções, bem como entender a importância da escolha correta na hora de definir a solução de armazenamento adequada para cada situação. Serão discutidos os diversos tipos de dispositivos de armazenamento existentes no mercado, tais como unidade de disco rígido (HDD) , unidades de disco sólido (SSD), unidades de fita (tape drive), cartões de memória, entre outros. Os participantes poderão entender as características de cada um desses dispositivos, bem como saber qual a melhor opção.


15:00

Foyer

Coffee-break


15:30

Auditorium

HPC/DL

15:30 - 17:00

Panel: Towards developing HPC through ML and ML through HPC - the Chicken Egg dilemma

Panelists: Profa. Lucia Drummond (UFF), Prof. Antônio Tadeu Azevedo Gomes (LNCC), Profa. Mariza Ferro (UFF), Dr. Fabio Alves de Oliveira (NVidia)

Moderator: Profa. Flavia Bernardini (UFF)

Machine Learning (ML) area have evolved in the last years due to the development of High Performance Computing (HPC). On the other hand, we also have seen many works for evolving HPC through ML. In this context, we could ask: If the community needs HPC to make huge ML-based models more effective, but this same community also needs ML to empower HPC, which area starts this run? Are they exclusive or can we consider both initiatives? This panel aims to discuss this chicken-egg dilemma.


15:30

Room 501

WCC

15:30 - 17:00

Technical Session 2

Chair: Márcio Castro (UFSC) / Philippe O. A. Navaux (UFRGS)

A Preliminary Review of Function as a Service Platform Running with AWS Spot Instances PDF

Luciana Marques (USP), Alfredo Goldman (USP)

Evaluation Model and Performance Analysis of NIC Aggregations in Containerized Private Clouds PDF

Anderson Mattheus Maliszewski (UFRGS), Dalvan Griebler (PUCRS), Eduardo Roloff (UFRGS), Rodrigo Righi (UNISINOS), Philippe O. A. Navaux (UFRGS)

A Performance Comparison of HPC Workloads on Traditional and Cloud-based HPC Clusters PDF

Vanderlei Munhoz (UFSC), Antoine Bonfils (Polytech Grenoble), Márcio Castro (UFSC), Odorico Mendizabal (UFSC)

Optimizing Microservices Performance and Resource Utilization through Containerized Grouping: An Experimental Study PDF

Fernando Buzato (USP), Alfredo Goldman (USP)

Conceptual and comparative analysis of application metrics in microservices PDF

Lucas Eduardo Pulcinelli (USP), Diego Pedroso (USP), Sarita Bruschi (USP)


15:30

WSCAD Tutorials (4, 5, and 6) Event in Portuguese

15:30 - 17:00

Are you Root? Reproducible Experiments in User Space PDF

Authors: Vinicius Garcia Pinto (UFRGS), Lucas Nesi (UFRGS), Lucas Mello Schnorr (UFRGS)

Room 502

Plataformas computacionais para processamento de alto desempenho usualmente são compartilhadas por usuários com demandas variadas. Com frequência aplicações legadas ou aquelas que possuem requisitos e dependências muito específicas sobrecarregam os administradores do sistema ou simplesmente tem sua execução inviabilizada. Uma solução passa a ser configurar, compilar e instalar a aplicação e as respectivas dependências inteiramente em espaço de usuário. Entretanto, tal tarefa é custosa e propensa a erros, o que motiva a adoção de soluções automatizadas. Neste minicurso, propomos a apresentação de duas soluções que permitem a instalação de pacotes inteiramente em espaço de usuário de maneira reprodutível e compartilhável.

15:30 - 17:00

Abordagem Gamificada para Introdução em Arquitetura de Hardware Evolutivo

Authors: Bernardo Cunha (PUC Minas), Carlos Augusto Martins (PUC Minas)

Room 514

Este minicurso consiste em 3 partes, de aproximadamente 50 minutos. Inicialmente, haverá exposição dos conceitos básicos para compreensão do tema, utilizando abordagem gamificada. Na segunda parte, os participantes colocarão estas novas informações em prática, "hackeando" o sistema evolutivo, experimentando variações nos parâmetros de entrada e de configuração. Por último, será escolhida a melhor abordagem a ser implementada em um próximo encontro (virtual), é demonstrado como o sistema selecionado pode ser implementado utilizando um dispositivo de computação reconfigurável (FPGA).

15:30 - 17:00

Armazenamento - Melhores Práticas - Construindo Sistemas de Armazenamento com a Performance Adequada PDF

Authors: Guilherme Friol

Room 515

Este minicurso abordará as principais formas de conexões existentes, tais como SATA, SCSI, SAS, iSCSI, Fibre Channel, entre outras. Os participantes poderão compreender as vantagens e desvantagens de cada uma dessas opções, bem como entender a importância da escolha correta na hora de definir a solução de armazenamento adequada para cada situação. Serão discutidos os diversos tipos de dispositivos de armazenamento existentes no mercado, tais como unidade de disco rígido (HDD) , unidades de disco sólido (SSD), unidades de fita (tape drive), cartões de memória, entre outros. Os participantes poderão entender as características de cada um desses dispositivos, bem como saber qual a melhor opção.


17:00

Theater

Opening Session


18:00

Theater

Philippe O. A. Navaux - Academic Trajectory (Honorary Researcher SBAC-PAD 2022)

Speaker: Philippe O. A. Navaux - 2021 CEACPAD Award on HPC

Those who received the CEACPAD award are invited to present their trajectory. We will present an overview of the main points of my career as a researcher and educator with more than 100 master's and PhD students and more than 400 published papers. From the first steps in electronic engineering projects in physical labs to current works on projects with hospitals and oil and gas companies.


19:30

Museum

Cocktail




Wednesday, October 18


08:00

Foyer

Registration


09:00

Theater

SBAC-PAD 1 | Best Papers I

Chair: Jay Lofstead (Sandia National Laboratories, USA)

Improved Computation of Database Operators via Vector Processing Near-Data PDF

Sairo Santos, Tiago Rodrigo Kepe and Marco Antonio Zanata Alves

Analysing Mechanisms for Virtual Channel Management in Low-Diameter Networks PDF

Alejandro Cano Cos, Cristóbal Camarero Coterillo, Carmen Martínez Fernández and Ramón Beivide Palacio

Dynasor: A Dynamic Memory Layout for Accelerating Sparse MTTKRP for Tensor Decomposition on Multi-core CPU PDF

Sasindu Wijeratne, Rajgopal Kannan and Viktor Prasanna


09:00

Auditorium

WSCAD 1 | Arquitetura e Sistema Operacional Event in Portuguese

Chair: Arthur Lorenzon (UFRGS)

New Kids on the Unblocking: Strategies to Overcome Blocking Networks PDF

Caio Morais (UFV), Jeronimo Penha (CEFET/MG), José Augusto Nacif (UFV), Ricardo Ferreira (UFV)

Virtualização e Migração de Processos em um Sistema Operacional Distribuído para Lightweight Manycores PDF

Nicolas Vanz (UFSC), João Souto (UFSC), Márcio Castro (UFSC)

KCGRA- Uma Arquitetura Reconfigurável de Dominio Específico para K-means PDF

Matheus da Silva Alves (UFV), Lucas Bragança da Silva (UFV), Jeronimo Penha (CEFET/MG), Ricardo Ferreira (UFV), José Augusto Nacif (UFV)

Instruction-Level Loop Perforation PDF

Daniela Catelan (UFMS), Ricardo Santos (UFMS), Liana Duenha (UFMS), Lucas Wanner (UNICAMP)


09:00

Room 501

WSCAD WIC 1 | Simulação e Compilação Event in Portuguese

Chair: Guilherme Galante (Unioeste)

Simulador do Algoritmo de Tomasulo com Conjunto de Instruções RISC-V PDF

Thiago Nolasco (PUC Minas), Danielle Vieira (PUC Minas), João Augusto dos Santos Silva (PUC Minas), Henrique Cota de Freitas (PUC Minas)

Abordagem para Aprendizado do Simulador gem5 para Pesquisadores Iniciantes PDF

Pedro Rigotto (PUC Minas), Henrique Cota de Freitas (PUC Minas)

Performance Evaluation of Intel and AMD Memory Hierarchies Using a Simulation-driven Approach With Gem5 PDF

João Victor Vieira (PUC Minas), Matheus Souza (PUC Minas), Henrique Cota de Freitas (PUC Minas)

A Systematic Literature Review on Optimization Techniques for Quantum Computing Compilers PDF

Camilla Da Rocha (PUC Minas), Ana Luiza Santos (PUC Minas), Matheus Souza (PUC Minas)


10:30

Foyer

Coffee-break


11:00

Theater

Benchmarking of Hybrid Computational Continuum

Speaker: Ivona Brandic

Chair: César De Rose

Real life applications can be executed on devices that range in size from smartphones to warehouse size data centers. Meanwhile, architectures are becoming heterogeneous including various accelerators and even non von Neumann computers into the computational continuum. In this talk we discuss challenges when benchmarking different types of applications on a hybrid computational continuum. First, we present the methods for decomposition and execution of HPC applications on hybrid Classic/Quantum systems. Second, we discuss the benefits but also problems and challenges when executing applications of hybrid systems. Third, we revisit the current state of the art testbeds for the execution and benchmarking of hybrid classic/quantum.


12:00

Lunch (Not included)


13:30

Theater

SBAC-PAD 2 | Architecture

Chair: Krishna Kavi (University of North Texas, USA)

Using Logging-on-Write to Improve Non-Volatile Memory Checkpoints via Processing-in-Memory PDF

Kleber Kruger, Ricardo Pannain and Rodolfo Azevedo

A Practical Approach For Workload-Aware Data Movement in Disaggregated Memory Systems PDF

Amit Puri, Kartheek Bellamkonda, Kailash Narreddy, John Jose, Tamarapalli Venkatesh

Performance Modeling and Estimation of a Configurable Output Stationary Neural Network Accelerator PDF

Ali Oudrhiri, Emilien Taly, Nathan Bain, Alix Munier, Roberto Guizzetti, and Pascal Urard


13:30

Auditorium

WSCAD 2 | Aprendizado de Máquina e Computação em Borda Event in Portuguese

Chair: Márcio Castro (UFSC)

Prediction of Reservoir Simulation Jobs Times Using a Real-World SLURM Log PDF

Alan Nunes (UFF), Felipe Portella (PETROBRAS), Paulo Estrela (PETROBRAS), Renzo Malini (PETROBRAS), Bruno Lopes (UFF), Arthur Vasconcelos (UFF), Gabriel Leite (UFF), Lúcia Drummond (UFF)

Análise da Execução de Algoritmos de Aprendizado de Máquina em Dispositivos Embarcados PDF

Lucas Alf (PUCRS), Renato Barreto Hoffmann Filho (PUCRS), Caetano Müler (PUCRS), Dalvan Griebler (PUCRS)

Estratégia de Posicionamento de Aplicações Sensíveis à Privacidade e Latência em Bordas Federadas PDF

Marcos Konzen (IFFar/UFPEL), Paulo De Souza (PUCRS), Fabio Rossi (IFFar), Julio Mattos (UFPEL)

Dynamic Provisioning of Container Registries in Edge Computing Infrastructures PDF

Lucas Roges (PUCRS), Tiago Ferreto (PUCRS)


13:30

Room 501

WSCAD WIC 2 | Avaliação e Otimização de Desempenho Event in Portuguese

Chair: Arthur Lorenzon (UFRGS)

Case Study on the Use of Winograd-Based Convolution for CNN Inference in FPGA PDF

Artur Pereira (UnB), Marcus Lamar (UnB)

Análise do Desempenho Computacional de Algoritmos Paralelizados com OpenMP e MPI Executados em Raspberry Pi PDF

Augusto Linhares Junqueira Ignácio (IFRO), Wanderson Roger Azevedo Dias (IFRO)

Avaliando eficiência energética em padrões de algoritmos para computação científica e de alto desempenho PDF

Paulo Anjos (UNIFESP), Alvaro Fazenda (UNIFESP)

Estudo de Desempenho de Duas Estratégias Paralelas Aplicadas ao Ajuste de Parâmetros de um Modelo Matemático da Esclerose Múltipla PDF

Gustavo Gonçalves Silva (UFJF), Matheus Avila Moreira de Paula (UFJF), Barbara de Melo Quintela (UFJF), Marcelo Lobosco (UFJF)

Implementando Elasticidade no Nível do Sistema Operacional PDF

Valquíria Belusso (Unioeste), Guilherme Galante (Unioeste)


15:00

Foyer

Coffee-break


15:30

Theater

SBAC-PAD 3 | Distributed Systems

Chair: Viktor Prasanna (University of Southern California, USA)

Improving Fault Tolerance in Blockchain Sharding using One-to-Many Block-to-Shard Mapping PDF

Tirathraj Ramburn and Dhrubajyoti Goswami

A Distributed Algorithm for Identifying Strongly Connected Components on Incremental Graphs PDF

Sudharshan Srinivasan, Arindam Khanda, Sriram Srinivasan, Aashish Pandey, Sajal Das, Sanjukta Bhowmick and Boyana Norris

Performance Modeling of MARE2DEM's Adaptive Mesh Refinement for Makespan Estimation PDF

Bruno da Silva Alves and Lucas Mello Schnorr


15:30

Auditorium

WSCAD 3 | GPU e Blockchain Event in Portuguese

Chair: Antonio C. S. Beck Filho (UFRGS)

Comparison of scalable distributed algorithms for assessing the kNNG in multi-GPU PDF

Gabriel Orlando (UFSCar), Hermes Senger (UFSCar), Murilo Naldi (UFSCar)

Data mapping strategies for multi-GPU implementation of a seismic application PDF

Yuri Freire (UFSCar), Edson Gomi (UFSCar), Hermes Senger (UFSCar)

BWJOIN: A Blockwise GPU-based Algorithm for Set Similarity Joins PDF

Rafael Quirino (UFG), Andre Moreira Quirino (Visiona Tecnologia Espacial), Leonardo Andrade Ribeiro (UFG), Wellington Martins (UFG)

The Impact of Hyperledger Fabric Setup on Blockchain Performance when Using Large Volumes of Heterogeneous Medical Data PDF

Ana Spengler (ICMC-USP), Paulo Lopes de Souza (ICMC-USP)


15:30

Room 501

WSCAD WEAC | Educação em Arquitetura de Computadores Event in Portuguese

Chair: Carlos Augusto Martins (PUC Minas) e Ricardo Menotti (UFSCar)

Aprendendo Hierarquia de Memória e a Exploração das Localidades Espacial e Temporal com o Simulador Amnesia PDF

Danielle Vieira (PUC Minas), Thiago Nolasco (PUC Minas), João Augusto dos Santos Silva (PUC Minas), Cecília Bouchardet (PUC Minas), Henrique Cota de Freitas (PUC Minas)

Enhancing Supercomputing Education through a Low-Cost Cluster: A Case Study at Insper PDF

Lícia Lima (Insper), Tiago Demay (Insper), Leonardo Rosa (Insper), André Batista (Insper), Luciano Silva (Insper)

Ensino de Software Pipelining e Escalonamento em GPUs com Python PDF

Ricardo Ferreira (UFV), José Augusto Nacif (UFV)

ODS como Temática para a Prática em Circuitos Digitais: relato de experiência PDF

Denise Stringhini (UNIFESP), Thaina Tosta (UNIFESP), Tiago Oliveira (UNIFESP)

Experiências com o Ensino de Arquiteturas e Programação Paralela: da Construção à Avaliação PDF

Guilherme Galante (Unioeste)


17:00

Theater

High Performance Computing Solutions with Dell technologies

Speaker: Rogério Chola (Dell)

Chair: Dalvan Griebler


17:30

Theater

Panel: Brazil HPC Investment needs for the coming years Event in Portuguese

Panelists: Dr. Luiz Monnerat, Pedro Leite da Silva Dias, Antônio Tadeu A. Gomes

Moderator: Philippe Olivier Alexandre Navaux

In the last TOP500, Brazil appears with 9 computers on the list. There are currently several investments in machines being prepared, including a call for 100 million reais for machines. The Panel will discuss upcoming investments in supercomputing centers and the country's needs in this area.




Thursday, October 19


08:00

Foyer

Registration


09:00

Theater

SBAC-PAD 4 | Best Papers II

Chair: Vinod Rebello (Universidade Federal Fluminense, Brazil)

Exploiting the Potential of Flexible Processing Units PDF

Mateo Vazquez, Muhammad Waqar Azhar and Pedro Trancoso

Trading Performance, Power, and Area on Low-Precision Posit MAC Units for CNN Training PDF

Luís Crespo, Pedro Tomás, Nuno Roma and Nuno Neves

BEASY: Making EASY backfilling renewable-only PDF

Igor Fontana de Nardin, Patricia Stolf and Stephane Caux


09:00

Auditorium

WSCAD 4 | Dimensionamento, Balanceamento, Escalabilidade e Produtividade Event in Portuguese

Chair: Wellington Martins (UFG)

GraspCC-LB: Dimensionamento de Recursos para Execução de Workflows em Ambientes de Computação de Alto Desempenho PDF

Luis Alvarenga (CEFET/RJ), Yuri Frota (UFF), Daniel de Oliveira (UFF), Rafaelli Coutinho (CEFET/RJ)

Analisando a Escalabilidade e a Acurácia de Implementações Paralelas e Distribuídas para a Detecção de Comunidades em Grafos PDF

Gabriel Santos (PUCRS), Cesar De Rose (PUCRS), Kartik Lakhotia (Intel Labs)

Balanceamento de Carga Dinâmico em Ambientes Kubernetes com o Kubernetes Scheduling Extension (KSE) PDF

Pedro Moritz de Carvalho Neto (UFSC), Márcio Castro (UFSC), Frank Siqueira (UFSC)

Extending the Planning Poker Method to Estimate the Development Effort of Parallel Applications PDF

Gabriella Andrade (PUCRS), Dalvan Griebler (PUCRS), Rodrigo Santos (UNIRIO), Luiz Gustavo Fernandes (PUCRS)


09:00

Room 501

WSCAD CTD 1 | Mestrado Event in Portuguese

Chair: Alba Melo (UnB) e Rodrigo Righi (Unisinos)

Impacts of Parallel Programming on Limited-Resource Hardware

Renato Barreto Hoffmann Filho (PUCRS)

Conversão dinâmica de instruções para processamento vetorial em memória

Rodrigo Sokulski (UFPR)

Thermal Optimization for OpenMP Applications through Thread Throttling and Turbo Boosting

Sandro Matheus Marques (Unipampa)


10:30

Foyer

Coffee-break


11:00

Theater

Frontier: The World’s Most Powerful Computer for Science

Speaker: Bronson Messer

Chair: Tiago Ferreto

The first exascale computer, called Frontier, has been delivered to Oak Ridge National Laboratory this past year. This unique scientific instrument is the culmination of more than a decade of concerted effort. I will relate a bit of the history of hybrid-node computing at the Oak Ridge Leadership Computing Facility (OLCF) and how Frontier represents the latest iteration of that approach. Some details of Frontier’s architecture will be discussed, including an overview of the new AMD GPUs that provide the bulk of the computational power for Frontier. Finally, we will take a look at some problems that will benefit from the increased capability at exascale and I will convey some lesson learned from trying to get the applications attacking these problems running on this machine.


12:00

Lunch (Not included)


13:30

Theater

SBAC-PAD 5 | Performance Engineering

Chair: Edson Borin (UNICAMP)

Performance Tuning for GPU-Embedded Systems: Machine-Learning-based and Analytical Model-driven Tuning Methodologies PDF

Adrián Perez Diéguez and Margarita Amor López

Comparing Performance and Portability between CUDA and SYCL for Protein Database Search on NVIDIA, AMD, and Intel GPUs PDF

Manuel Costanzo, Enzo Rucci, Carlos García Sánchez, Marcelo Naiouf and Manuel Prieto-Mátias

Achieving Enhanced Performance Combining Checkpointing and Dynamic State Partitioning PDF

Henrique dos Santos Goulart, João Gabriel Trombeta, Álvaro Junio Pereira Franco, and Odorico Machado Mendizabal


13:30

Auditorium

WSCAD 5 | Computação em Nuvem Event in Portuguese

Chair: Eduardo Camilo Inacio (UniSenai)

Functionalities as a Service - An Approach to Conciliate Interoperability and Data Reduction in E-Health PDF

João Pedro da Costa (UFJF), Mario Dantas (UFJF), José Maria David (UFJF), Fernando Silva (UFJF)

Evaluating the Parallel Simulation of Dynamics of Electrons in Molecules on AWS Spot Instances PDF

Vanderlei Pereira Filho (UFSC), Márcio Castro (UFSC), Luis Guilherme Carvalho Rego (UFSC)

Análise de Custo e Desempenho de um Sistema de Modelagem Atmosférica Tolerante a Falhas no AWS ParallelCluster PDF

Mateus Melo (UFF), Roberto Pinto Souto (LNCC), Lúcia Drummond (UFF)

Explorando a Variabilidade de Processo para Otimizar a Eficiência Energética em Servidores de Nuvem PDF

Thiago dos Santos Gonçalves (UFRGS), Antonio Carlos Schneider Beck (UFRGS), Arthur Lorenzon (UFRGS)

How FaaS with DBaaS performs in different regions: an evaluation by the Orama Framework PDF

Leonardo Rebouças de Carvalho (UnB), Bruno Kamienski (UnB), Aleteia Araujo (UnB)


13:30

Room 501

WSCAD CTD 2 | Doutorado Event in Portuguese

Chair: Alba Melo (UnB) e Rodrigo Righi (Unisinos)

Advanced Computational Strategies for Reverse Time Migration

Carlos Barbosa (COPPE/UFRJ)

Suporte Arquitetural para Computação Aproximada

Isaias Felzmann (UNICAMP)

Easing the Benchmarking of Parallel Stream Processing on Multi-cores

Adriano Garcia (PUCRS)


15:00

Foyer

Coffee-break


15:30

Theater

Lenovo Neptune - learn about disruptive liquid cooling solutions for servers

Speaker: Joaquim Merino (Lenovo)

Chair: Dalvan Griebler

Discover Lenovo's HPC solutions, #1 in supplying supercomputers in the Top500. Be surprised by Lenovo Neptune, liquid server cooling technology that guarantees better performance, savings and enables companies and organizations to achieve their ESG objectives.


16:30

Theater

SBAC-PAD 6 | Systems Optimization

Chair: Jay Lofstead (Sandia National Laboratories, USA)

Toward Open Repository of Performance Portability of Applications, Benchmarks and Models PDF

Ami Marowka

NeurOPar, A Neural Network-driven EDP Optimization Strategy for Parallel Workloads PDF

Cristiano Alex Kunas, Fábio Rossi, Marcelo Caggiani Luizelli, Rodrigo Calheiros, Philippe Navaux and Arthur Lorenzon

Forecasting file lifecycles for intelligent data placement in hierarchical storage PDF

Adrian Khelili, Sophie Robert Hayek and Soraya Zertal


16:30

Auditorium

WSCAD 6 | Computação em GPU Event in Portuguese

Chair: Matheus Souza (PUC Minas)

KNN paralelo em GPU para grandes volumes de dados com agregação de consultas PDF

Michel Brasil Cordeiro (UFPR), Wagner Zola (UFPR)

A GPU-based DP algorithm for solving multiple instances of the knapsack problem PDF

Dayllon Xavier (UFG), Humberto José Longo (UFG), Leslie Richard Foulds (UFG), Wellington Martins (UFG)

Avaliação de estilos de código para árvore de decisão em GPU com Microbenchmarks PDF

Jeronimo Penha (CEFET/MG), Alysson Kelvim Caetano da Silva (UFV), Olavo Silva (UFV), Icaro Moreira (UFV), José Augusto Nacif (UFV), Ricardo Ferreira (UFV)

A constructive parallel heuristic approach for the MWPSP PDF

Rafael de Castro Silva (UFG), Vinicius Coelho (UFG), Wellington Martins (UFG), Leslie Richard Foulds (UFG), Humberto José Longo (UFG)


16:30

Room 501

CEACPAD Meeting Event in Portuguese


19:30

Veleiros do Sul

Dinner




Friday, October 20


08:00

Foyer

Registration


09:00

Theater

SBAC-PAD 7 | Parallel Applications

Chair: J. Nelson Amaral (University of Alberta, Canada)

Reverse Time Migration with Lossy and Lossless Wavefield Compression PDF

Carlos Henrique dos Santos Barbosa and Alvaro Luiz Gayoso de Azeredo Coutinho

Improving the discovery and clustering of three-dimensional protein patterns with OpenMP PDF

Alejandro Valdés-Jiménez, Miguel Reyes-Parada, Gabriel Nuñez-Vivanco, Fabio Durán-Verdugo and Daniel Jiménez-González

A Shared Memory SMC Sampler for Decision Trees PDF

Efthyvoulos Drousiotis, Alessandro Varsi, Paul Spirakis, and Simon Maskell


09:00

Auditorium

WSCAD 7 | Programação e Gerenciamento de Recursos Event in Portuguese

Chair: Ricardo Ferreira (UFV)

Structured platform-aware programming PDF

Francisco Carvalho-Junior (UFC), Julio Mendes (Arpeggeo Technologies), Allberson Oliveira (Unilab), Tiago Carneiro (IMEC, Belgium), Claro Sales (UFC), Pedro Sales (UFC)

Conversão do NAS Parallel Benchmarks para C++ Standard PDF

Arthur Bianchessi (PUCRS), Leonardo Mallmann (PUCRS), Renato Hoffman (PUCRS), Dalvan Griebler (PUCRS)

PM.NET: Uma biblioteca de desenvolvimento para memória persistente com C# PDF

Henrique Guirelli (UFABC), Emilio Francesquini (UFABC), Alexandro Baldassin (UNESP)

Impact of a dynamic Allocation Policy for Resource and Job Management Systems in deadline-oriented Scenarios PDF

Barry Linnert (TU-Berlin), Cesar De Rose (PUCRS), Hans-Ulrich Heiss (TU-Berlin)

Fragmentando o DNA de Ferramentas de Alinhamento Progressivo: uma Metaferramenta Eficiente PDF

Mario João JR (UERJ/UFF), Alexandre Sena (UERJ), Vinod Rebello (UFF)


10:30

Foyer

Coffee-break


11:00

Theater

Pairwise, Structural and Multiple Biological Sequence Comparison in HPC Platforms: the Quest for the Optimal Solution

Speaker: Alba Cristina Magalhaes Alves de Melo

Chair: Arthur Lorenzon

Biological sequence comparison is an important problem in Bioinformatics and its goal is to define how similar the sequences are, producing a score, and highlighting their similarities, producing an alignment. There are many ways to compare biological sequences and all of them require high performance computing solutions, when the optimal solution is needed. In this talk, we will deal with three types of sequence comparison. First, we discuss pairwise sequence comparison, which is often solved with dynamic programming using variants of the Smith-Waterman algorithm, producing the optimal solution with time complexity O(n^2), where n is the length of the sequences. We present our MASA tools, which can be used in CPU or GPU to pairwise compare long DNA sequences. The last version of MASA for GPUs (MASA-CUDAlign-MultiBP) attained the best performance in the literature in 2021. Then, we examine the structural RNA alignment problem, that is also solved with dynamic programming, using the Sankoff algorithm, with time complexity O(n^6). We present our CUDA-Sankoff tool and show that it has very good speedups. Next, we discuss the multiple sequence alignment (MSA) problem, which is proven NP-Complete. We present our PA-Star CPU-based tool, which executes a variant of the A-Star algorithm to compute the optimal MSA. We show that PA-Star is able to compare multiple sequences in reduced time, when compared to the literature. At the end of the talk, we will do a covid-19 case study, showing how our three tools are used to compare SARS-CoV-2 sequences.


12:00

Lunch (Not included)


13:30

Theater

SBAC-PAD 8 | Cloud Computing

Chair: Lucas Mello Schnorr (Federal University of Rio Grande do Sul, Brazil)

Kub: Elastic HPC Workloads on Containerized Environments PDF

Daniel Medeiros, Jacob Wahlgren, Gabin Schieffer and Ivy Peng

WCSim: A Cloud Computing Simulator with Support for Bag of Tasks Workflows PDF

Maicon Santos, Gabriel Grabher, Matheus Kovaleski, Cláudio Geyer, and Gerson Cavalheiro


14:30

Theater

Closing Session