Analytical speedup modeling for parallel applications with variable memoryaccess delay in symmetric architectures
Several analytical models created since Amdahl’s pioneering work have explored aspects such as variation in the size of the problem, memory size, communication overhead, and synchronization overhead. However, delays in memory access are considered constant. Such delays can vary, for example, accord...
সংরক্ষণ করুন:
প্রধান লেখক: | |
---|---|
অন্যান্য লেখক: | |
বিন্যাস: | doctoralThesis |
ভাষা: | pt_BR |
প্রকাশিত: |
Universidade Federal do Rio Grande do Norte
|
বিষয়গুলি: | |
অনলাইন ব্যবহার করুন: | https://repositorio.ufrn.br/handle/123456789/32749 |
ট্যাগগুলো: |
ট্যাগ যুক্ত করুন
কোনো ট্যাগ নেই, প্রথমজন হিসাবে ট্যাগ করুন!
|
id |
ri-123456789-32749 |
---|---|
record_format |
dspace |
institution |
Repositório Institucional |
collection |
RI - UFRN |
language |
pt_BR |
topic |
Analytical modeling Speedup modeling Parallel systems Data access delay Speedup Memory wall |
spellingShingle |
Analytical modeling Speedup modeling Parallel systems Data access delay Speedup Memory wall Furtunato, Alex Fabiano de Araújo Analytical speedup modeling for parallel applications with variable memoryaccess delay in symmetric architectures |
description |
Several analytical models created since Amdahl’s pioneering work have explored aspects such as variation in the size of the problem, memory size, communication overhead, and synchronization overhead. However, delays in memory access are considered
constant. Such delays can vary, for example, according to the number of cores, the relationship between the processor and memory frequencies, and the problem size. Given
different problem sizes and many possible configurations of operational frequency and
number of cores that current architectures can offer, speedup models suitable for describing such differences are quite desirable for either offline or online scheduling decisions.
This thesis presents a novel analytical speedup model that considers variations in the
average data-access delay to describe the limiting effect of the memory wall in parallel
applications associated with homogeneous shared memory architectures. The experimental results indicate that the proposed model incorporates the behavior of the application
adequately. The approach presented in this work shows that considering parameters that
reflect the intrinsic characteristics of applications has advantages over statistical models
such as those based on machine learning. The experiments also show that the conventional
machine learning modeling may require measurements with one order of magnitude above
to achieve the same accuracy level when compared with the proposed model. |
author2 |
Souza, Samuel Xavier de |
author_facet |
Souza, Samuel Xavier de Furtunato, Alex Fabiano de Araújo |
format |
doctoralThesis |
author |
Furtunato, Alex Fabiano de Araújo |
author_sort |
Furtunato, Alex Fabiano de Araújo |
title |
Analytical speedup modeling for parallel applications with variable memoryaccess delay in symmetric architectures |
title_short |
Analytical speedup modeling for parallel applications with variable memoryaccess delay in symmetric architectures |
title_full |
Analytical speedup modeling for parallel applications with variable memoryaccess delay in symmetric architectures |
title_fullStr |
Analytical speedup modeling for parallel applications with variable memoryaccess delay in symmetric architectures |
title_full_unstemmed |
Analytical speedup modeling for parallel applications with variable memoryaccess delay in symmetric architectures |
title_sort |
analytical speedup modeling for parallel applications with variable memoryaccess delay in symmetric architectures |
publisher |
Universidade Federal do Rio Grande do Norte |
publishDate |
2021 |
url |
https://repositorio.ufrn.br/handle/123456789/32749 |
work_keys_str_mv |
AT furtunatoalexfabianodearaujo analyticalspeedupmodelingforparallelapplicationswithvariablememoryaccessdelayinsymmetricarchitectures |
_version_ |
1773958341404393472 |
spelling |
ri-123456789-327492021-06-21T23:04:01Z Analytical speedup modeling for parallel applications with variable memoryaccess delay in symmetric architectures Furtunato, Alex Fabiano de Araújo Souza, Samuel Xavier de http://lattes.cnpq.br/9659861253244985 http://lattes.cnpq.br/9892239670106361 Silveira, Luiz Felipe de Queiroz http://lattes.cnpq.br/4139452169580807 Silva, Ivanovitch Medeiros Dantas da http://lattes.cnpq.br/3608440944832201 Lorenzon, Arthur Francisco http://lattes.cnpq.br/2890260984567329 Borin, Edson http://lattes.cnpq.br/4176915322233893 Analytical modeling Speedup modeling Parallel systems Data access delay Speedup Memory wall Several analytical models created since Amdahl’s pioneering work have explored aspects such as variation in the size of the problem, memory size, communication overhead, and synchronization overhead. However, delays in memory access are considered constant. Such delays can vary, for example, according to the number of cores, the relationship between the processor and memory frequencies, and the problem size. Given different problem sizes and many possible configurations of operational frequency and number of cores that current architectures can offer, speedup models suitable for describing such differences are quite desirable for either offline or online scheduling decisions. This thesis presents a novel analytical speedup model that considers variations in the average data-access delay to describe the limiting effect of the memory wall in parallel applications associated with homogeneous shared memory architectures. The experimental results indicate that the proposed model incorporates the behavior of the application adequately. The approach presented in this work shows that considering parameters that reflect the intrinsic characteristics of applications has advantages over statistical models such as those based on machine learning. The experiments also show that the conventional machine learning modeling may require measurements with one order of magnitude above to achieve the same accuracy level when compared with the proposed model. Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES Vários modelos analíticos criados desde o trabalho pioneiro de Amdahl exploraram aspectos tais como variação no tamanho do problema, tamanho da memória, sobrecarga de comunicação e sobrecarga de sincronização. Contudo, atrasos no acesso à memória são considerados constantes. Esses atrasos podem variar, por exemplo, de acordo com o número de núcleos usados, a relação entre as frequências do processador e da memória e o tamanho do problema. Dado os diferentes tamanhos de problemas e o grande número de configurações possíveis de frequência operacional e número de núcleos que as arquiteturas atuais podem oferecer, modelos de speedup adequados para descrever tais variações entre essas configurações são bastante desejáveis para decisões de escalonamento offline ou online. Esta tese apresenta um novo modelo analítico de speedup que considera variações no atraso médio de acesso à memória para descrever o efeito limitador da barreira de memória em aplicações paralelas em arquiteturas homogêneas de memória compartilhada. Os resultados experimentais indicam que a modelagem proposta captura bem o comportamento da aplicação. A proposta apresentada nesse trabalho mostra que incorporar parâmetros que refletem as características intrínsecas das aplicações tem vantagens sobre modelos estatísticos como os baseados em aprendizagem de máquina. Os experimentos também mostram que a modelagem de aprendizagem de máquina convencional pode precisar de uma ordem de magnitude a mais de medições para atingir o mesmo nível de acurácia em comparação com o modelo proposto. 2021-06-21T23:03:10Z 2021-06-21T23:03:10Z 2021-02-09 doctoralThesis FURTUNATO, Alex Fabiano de Araújo. Analytical speedup modeling for parallel applications with variable memoryaccess delay in symmetric architectures. 2021. 125f. Tese (Doutorado em Engenharia Elétrica e de Computação) - Centro de Tecnologia, Universidade Federal do Rio Grande do Norte, Natal, 2021. https://repositorio.ufrn.br/handle/123456789/32749 pt_BR Acesso Aberto application/pdf Universidade Federal do Rio Grande do Norte Brasil UFRN PROGRAMA DE PÓS-GRADUAÇÃO EM ENGENHARIA ELÉTRICA E DE COMPUTAÇÃO |