Proposal to solve workload problems in the simulator of the Wolf architecture
Autor(a) principal: | |
---|---|
Data de Publicação: | 2000 |
Outros Autores: | , |
Tipo de documento: | Artigo de conferência |
Idioma: | eng |
Título da fonte: | Repositório Institucional da UNESP |
Texto Completo: | http://dx.doi.org/10.1109/HPC.2000.846568 http://hdl.handle.net/11449/219913 |
Resumo: | Presents an approach that has been proposed in order to investigate the workload balance problems in a simulator of the Wolf parallel architecture, known as Saw (Simulator of Architecture Wolf). This approach can be applied to the simulator in order to eliminate overloaded and underloaded units observed during the first experiments on the Wolf architecture. The unexpected results obtained are well-understood if one considers that the architecture simulated is based on a circular pipeline: if there is only one overloaded (or underloaded) unit in the simulator, it can affect the performance of the entire system, adding latency to it. The first attempt to improve the results obtained with the simulator was to calibrate all of the simulator units. The approach used leads to the necessity of dividing the units into two categories: grouped units and isolated units. These two categories are analyzed separately, resulting in modifications to the original architecture. |
id |
UNSP_a8348f826b1f38099861dfb6cd51cd87 |
---|---|
oai_identifier_str |
oai:repositorio.unesp.br:11449/219913 |
network_acronym_str |
UNSP |
network_name_str |
Repositório Institucional da UNESP |
repository_id_str |
2946 |
spelling |
Proposal to solve workload problems in the simulator of the Wolf architecturePresents an approach that has been proposed in order to investigate the workload balance problems in a simulator of the Wolf parallel architecture, known as Saw (Simulator of Architecture Wolf). This approach can be applied to the simulator in order to eliminate overloaded and underloaded units observed during the first experiments on the Wolf architecture. The unexpected results obtained are well-understood if one considers that the architecture simulated is based on a circular pipeline: if there is only one overloaded (or underloaded) unit in the simulator, it can affect the performance of the entire system, adding latency to it. The first attempt to improve the results obtained with the simulator was to calibrate all of the simulator units. The approach used leads to the necessity of dividing the units into two categories: grouped units and isolated units. These two categories are analyzed separately, resulting in modifications to the original architecture.State Univ. of Sao Paulo Dept. Computer Science, Av. Luis CoubeState University of Maringa Dept. of Informatics UEMUniversity of Sao Paulo FFI - USPUniversidade de São Paulo (USP)Universidade Estadual de Maringá (UEM)Cavenaghi, M. A.Martini, J. A.Neto, A. G.2022-04-28T18:58:28Z2022-04-28T18:58:28Z2000-01-01info:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/conferenceObject315-318http://dx.doi.org/10.1109/HPC.2000.846568Proceedings - 4th International Conference/Exhibition on High Performance Computing in the Asia-Pacific Region, HPC-Asia 2000, v. 1, p. 315-318.http://hdl.handle.net/11449/21991310.1109/HPC.2000.8465682-s2.0-84878579226Scopusreponame:Repositório Institucional da UNESPinstname:Universidade Estadual Paulista (UNESP)instacron:UNESPengProceedings - 4th International Conference/Exhibition on High Performance Computing in the Asia-Pacific Region, HPC-Asia 2000info:eu-repo/semantics/openAccess2022-04-28T18:58:28Zoai:repositorio.unesp.br:11449/219913Repositório InstitucionalPUBhttp://repositorio.unesp.br/oai/requestopendoar:29462024-08-05T22:14:21.553560Repositório Institucional da UNESP - Universidade Estadual Paulista (UNESP)false |
dc.title.none.fl_str_mv |
Proposal to solve workload problems in the simulator of the Wolf architecture |
title |
Proposal to solve workload problems in the simulator of the Wolf architecture |
spellingShingle |
Proposal to solve workload problems in the simulator of the Wolf architecture Cavenaghi, M. A. |
title_short |
Proposal to solve workload problems in the simulator of the Wolf architecture |
title_full |
Proposal to solve workload problems in the simulator of the Wolf architecture |
title_fullStr |
Proposal to solve workload problems in the simulator of the Wolf architecture |
title_full_unstemmed |
Proposal to solve workload problems in the simulator of the Wolf architecture |
title_sort |
Proposal to solve workload problems in the simulator of the Wolf architecture |
author |
Cavenaghi, M. A. |
author_facet |
Cavenaghi, M. A. Martini, J. A. Neto, A. G. |
author_role |
author |
author2 |
Martini, J. A. Neto, A. G. |
author2_role |
author author |
dc.contributor.none.fl_str_mv |
Universidade de São Paulo (USP) Universidade Estadual de Maringá (UEM) |
dc.contributor.author.fl_str_mv |
Cavenaghi, M. A. Martini, J. A. Neto, A. G. |
description |
Presents an approach that has been proposed in order to investigate the workload balance problems in a simulator of the Wolf parallel architecture, known as Saw (Simulator of Architecture Wolf). This approach can be applied to the simulator in order to eliminate overloaded and underloaded units observed during the first experiments on the Wolf architecture. The unexpected results obtained are well-understood if one considers that the architecture simulated is based on a circular pipeline: if there is only one overloaded (or underloaded) unit in the simulator, it can affect the performance of the entire system, adding latency to it. The first attempt to improve the results obtained with the simulator was to calibrate all of the simulator units. The approach used leads to the necessity of dividing the units into two categories: grouped units and isolated units. These two categories are analyzed separately, resulting in modifications to the original architecture. |
publishDate |
2000 |
dc.date.none.fl_str_mv |
2000-01-01 2022-04-28T18:58:28Z 2022-04-28T18:58:28Z |
dc.type.status.fl_str_mv |
info:eu-repo/semantics/publishedVersion |
dc.type.driver.fl_str_mv |
info:eu-repo/semantics/conferenceObject |
format |
conferenceObject |
status_str |
publishedVersion |
dc.identifier.uri.fl_str_mv |
http://dx.doi.org/10.1109/HPC.2000.846568 Proceedings - 4th International Conference/Exhibition on High Performance Computing in the Asia-Pacific Region, HPC-Asia 2000, v. 1, p. 315-318. http://hdl.handle.net/11449/219913 10.1109/HPC.2000.846568 2-s2.0-84878579226 |
url |
http://dx.doi.org/10.1109/HPC.2000.846568 http://hdl.handle.net/11449/219913 |
identifier_str_mv |
Proceedings - 4th International Conference/Exhibition on High Performance Computing in the Asia-Pacific Region, HPC-Asia 2000, v. 1, p. 315-318. 10.1109/HPC.2000.846568 2-s2.0-84878579226 |
dc.language.iso.fl_str_mv |
eng |
language |
eng |
dc.relation.none.fl_str_mv |
Proceedings - 4th International Conference/Exhibition on High Performance Computing in the Asia-Pacific Region, HPC-Asia 2000 |
dc.rights.driver.fl_str_mv |
info:eu-repo/semantics/openAccess |
eu_rights_str_mv |
openAccess |
dc.format.none.fl_str_mv |
315-318 |
dc.source.none.fl_str_mv |
Scopus reponame:Repositório Institucional da UNESP instname:Universidade Estadual Paulista (UNESP) instacron:UNESP |
instname_str |
Universidade Estadual Paulista (UNESP) |
instacron_str |
UNESP |
institution |
UNESP |
reponame_str |
Repositório Institucional da UNESP |
collection |
Repositório Institucional da UNESP |
repository.name.fl_str_mv |
Repositório Institucional da UNESP - Universidade Estadual Paulista (UNESP) |
repository.mail.fl_str_mv |
|
_version_ |
1808129407519293440 |