Abstract
Early power modeling and analysis using electronic system-level methodology enables designers to explore energy saving opportunities more efficiently at a higher abstraction level. However, power modeling for third party IPs are challenging due to the limited observability and unknown architecture details. To model the data dependency for blackbox IPs, several works rely on adopting Hamming distance of input data to approximate the switching activity, which might be not enough for modeling complex IPs such as image signal processors (ISP). This work introduces a content-aware line-based power modeling method for ISP by training an associated energy table. To effectively estimate ISP energy consumption which involves many two-dimensional data processing, this work presents a direct energy-mapping strategy using pixel luminance and gradient. Moreover, an iterative box-constrained least-squares estimation and its associated constraint refinement scheme is proposed to increase the robustness of the trained energy table even with limited training data. Simulation results show that the proposed method can reduce at least 11.54% of average error and 55.52% of max error as compared to the existing content-blind power model.
| Original language | English |
|---|---|
| Title of host publication | Proceedings - 30th IEEE International System on Chip Conference, SOCC 2017 |
| Editors | Jurgen Becker, Ramalingam Sridhar, Hai Li, Ulf Schlichtmann, Massimo Alioto |
| Publisher | IEEE Computer Society |
| Pages | 346-350 |
| Number of pages | 5 |
| ISBN (Electronic) | 9781538640333 |
| DOIs | |
| Publication status | Published - 2017 Dec 18 |
| Event | 30th IEEE International System on Chip Conference, SOCC 2017 - Munich, Germany Duration: 2017 Sep 5 → 2017 Sep 8 |
Publication series
| Name | International System on Chip Conference |
|---|---|
| Volume | 2017-September |
| ISSN (Print) | 2164-1676 |
| ISSN (Electronic) | 2164-1706 |
Other
| Other | 30th IEEE International System on Chip Conference, SOCC 2017 |
|---|---|
| Country | Germany |
| City | Munich |
| Period | 17-09-05 → 17-09-08 |
Fingerprint
All Science Journal Classification (ASJC) codes
- Hardware and Architecture
- Control and Systems Engineering
- Electrical and Electronic Engineering
Cite this
}
Content-aware line-based power modeling methodology for image signal processor. / Chen, Chun Wei; Shieh, Ming-Der; Lu, Juin Ming; Huang, Hsun Lun; Chen, Yao Hua.
Proceedings - 30th IEEE International System on Chip Conference, SOCC 2017. ed. / Jurgen Becker; Ramalingam Sridhar; Hai Li; Ulf Schlichtmann; Massimo Alioto. IEEE Computer Society, 2017. p. 346-350 (International System on Chip Conference; Vol. 2017-September).Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
TY - GEN
T1 - Content-aware line-based power modeling methodology for image signal processor
AU - Chen, Chun Wei
AU - Shieh, Ming-Der
AU - Lu, Juin Ming
AU - Huang, Hsun Lun
AU - Chen, Yao Hua
PY - 2017/12/18
Y1 - 2017/12/18
N2 - Early power modeling and analysis using electronic system-level methodology enables designers to explore energy saving opportunities more efficiently at a higher abstraction level. However, power modeling for third party IPs are challenging due to the limited observability and unknown architecture details. To model the data dependency for blackbox IPs, several works rely on adopting Hamming distance of input data to approximate the switching activity, which might be not enough for modeling complex IPs such as image signal processors (ISP). This work introduces a content-aware line-based power modeling method for ISP by training an associated energy table. To effectively estimate ISP energy consumption which involves many two-dimensional data processing, this work presents a direct energy-mapping strategy using pixel luminance and gradient. Moreover, an iterative box-constrained least-squares estimation and its associated constraint refinement scheme is proposed to increase the robustness of the trained energy table even with limited training data. Simulation results show that the proposed method can reduce at least 11.54% of average error and 55.52% of max error as compared to the existing content-blind power model.
AB - Early power modeling and analysis using electronic system-level methodology enables designers to explore energy saving opportunities more efficiently at a higher abstraction level. However, power modeling for third party IPs are challenging due to the limited observability and unknown architecture details. To model the data dependency for blackbox IPs, several works rely on adopting Hamming distance of input data to approximate the switching activity, which might be not enough for modeling complex IPs such as image signal processors (ISP). This work introduces a content-aware line-based power modeling method for ISP by training an associated energy table. To effectively estimate ISP energy consumption which involves many two-dimensional data processing, this work presents a direct energy-mapping strategy using pixel luminance and gradient. Moreover, an iterative box-constrained least-squares estimation and its associated constraint refinement scheme is proposed to increase the robustness of the trained energy table even with limited training data. Simulation results show that the proposed method can reduce at least 11.54% of average error and 55.52% of max error as compared to the existing content-blind power model.
UR - http://www.scopus.com/inward/record.url?scp=85044329281&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85044329281&partnerID=8YFLogxK
U2 - 10.1109/SOCC.2017.8226075
DO - 10.1109/SOCC.2017.8226075
M3 - Conference contribution
T3 - International System on Chip Conference
SP - 346
EP - 350
BT - Proceedings - 30th IEEE International System on Chip Conference, SOCC 2017
A2 - Becker, Jurgen
A2 - Sridhar, Ramalingam
A2 - Li, Hai
A2 - Schlichtmann, Ulf
A2 - Alioto, Massimo
PB - IEEE Computer Society
ER -