The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
O cluster FPGA é uma plataforma promissora para a computação futura, não apenas na nuvem, mas também nas estações base sem fio 5G com fornecimento de energia limitado, aproveitando significativamente a eficiência energética. No entanto, quase nenhuma análise de potência com sistemas reais foi relatada. Este trabalho relata as análises detalhadas do consumo de energia de dois clusters FPGA, nomeadamente clusters FiC e M-KUBOS, com a introdução de ferramentas de medição de energia e execução de aplicações reais. A partir das análises detalhadas, constatamos que o número de links ativados determina principalmente o consumo total de energia dos sistemas, independentemente de serem utilizados ou não. Para melhorar o desempenho das aplicações e ao mesmo tempo reduzir o consumo de energia, devemos aumentar a frequência do clock das aplicações, usar o número mínimo de links e aplicar agregação de links. Também propomos o modelo de potência para ambos os clusters a partir dos resultados das análises e este modelo pode estimar o consumo total de energia de ambos os clusters FPGA na etapa de projeto com erros de 15% no máximo.
Kensuke IIZUKA
Keio University
Haruna TAKAGI
Keio University
Aika KAMEI
Keio University
Kazuei HIRONAKA
Keio University
Hideharu AMANO
Keio University
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copiar
Kensuke IIZUKA, Haruna TAKAGI, Aika KAMEI, Kazuei HIRONAKA, Hideharu AMANO, "Power Analysis and Power Modeling of Directly-Connected FPGA Clusters" in IEICE TRANSACTIONS on Information,
vol. E106-D, no. 12, pp. 1997-2005, December 2023, doi: 10.1587/transinf.2023PAP0009.
Abstract: FPGA cluster is a promising platform for future computing not only in the cloud but in the 5G wireless base stations with limited power supply by taking significant advantage of power efficiency. However, almost no power analyses with real systems have been reported. This work reports the detailed power consumption analyses of two FPGA clusters, namely FiC and M-KUBOS clusters with introducing power measurement tools and running the real applications. From the detailed analyses, we find that the number of activated links mainly determines the total power consumption of the systems regardless they are used or not. To improve the performance of applications while reducing power consumption, we should increase the clock frequency of the applications, use the minimum number of links and apply link aggregation. We also propose the power model for both clusters from the results of the analyses and this model can estimate the total power consumption of both FPGA clusters at the design step with 15% errors at maximum.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2023PAP0009/_p
Copiar
@ARTICLE{e106-d_12_1997,
author={Kensuke IIZUKA, Haruna TAKAGI, Aika KAMEI, Kazuei HIRONAKA, Hideharu AMANO, },
journal={IEICE TRANSACTIONS on Information},
title={Power Analysis and Power Modeling of Directly-Connected FPGA Clusters},
year={2023},
volume={E106-D},
number={12},
pages={1997-2005},
abstract={FPGA cluster is a promising platform for future computing not only in the cloud but in the 5G wireless base stations with limited power supply by taking significant advantage of power efficiency. However, almost no power analyses with real systems have been reported. This work reports the detailed power consumption analyses of two FPGA clusters, namely FiC and M-KUBOS clusters with introducing power measurement tools and running the real applications. From the detailed analyses, we find that the number of activated links mainly determines the total power consumption of the systems regardless they are used or not. To improve the performance of applications while reducing power consumption, we should increase the clock frequency of the applications, use the minimum number of links and apply link aggregation. We also propose the power model for both clusters from the results of the analyses and this model can estimate the total power consumption of both FPGA clusters at the design step with 15% errors at maximum.},
keywords={},
doi={10.1587/transinf.2023PAP0009},
ISSN={1745-1361},
month={December},}
Copiar
TY - JOUR
TI - Power Analysis and Power Modeling of Directly-Connected FPGA Clusters
T2 - IEICE TRANSACTIONS on Information
SP - 1997
EP - 2005
AU - Kensuke IIZUKA
AU - Haruna TAKAGI
AU - Aika KAMEI
AU - Kazuei HIRONAKA
AU - Hideharu AMANO
PY - 2023
DO - 10.1587/transinf.2023PAP0009
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E106-D
IS - 12
JA - IEICE TRANSACTIONS on Information
Y1 - December 2023
AB - FPGA cluster is a promising platform for future computing not only in the cloud but in the 5G wireless base stations with limited power supply by taking significant advantage of power efficiency. However, almost no power analyses with real systems have been reported. This work reports the detailed power consumption analyses of two FPGA clusters, namely FiC and M-KUBOS clusters with introducing power measurement tools and running the real applications. From the detailed analyses, we find that the number of activated links mainly determines the total power consumption of the systems regardless they are used or not. To improve the performance of applications while reducing power consumption, we should increase the clock frequency of the applications, use the minimum number of links and apply link aggregation. We also propose the power model for both clusters from the results of the analyses and this model can estimate the total power consumption of both FPGA clusters at the design step with 15% errors at maximum.
ER -