Sonic Millip3De: A massively parallel 3D-stacked accelerator for 3D ultrasound

Richard Sampson, Ming Yang, Siyuan Wei, Chaitali Chakrabarti, Thomas F. Wenisch

Research output: Chapter in Book/Report/Conference proceedingConference contribution

36 Citations (Scopus)

Abstract

Three-dimensional (3D) ultrasound is becoming common for non-invasive medical imaging because of its high accuracy, safety, and ease of use. Unlike other modalities, ultrasound transducers require little power, which makes hand-held imaging platforms possible, and several low-resolution 2D devices are commercially available today. However, the extreme computational requirements (and associated power requirements) of 3D ultrasound image formation has, to date, precluded hand-held 3D capable devices. We describe the Sonic Millip3De, a new system architecture and accelerator for 3D ultrasound beamformationthe most computationally intensive aspect of image formation. Our three-layer die-stacked design features a custom beamsum accelerator that employs massive data parallelism and a streaming transform-select-reduce pipeline architecture enabled by our new iterative beamsum delay calculation algorithm. Based on RTL-level design and floorplanning for an industrial 45nm process, we show Sonic Millip3De can enable 3D ultrasound with a fully sampled 128×96 transducer array within a 16W full-system power budget (400× less than a conventional DSP solution) and will meet a 5W safe power target by the 11nm node.

Original languageEnglish (US)
Title of host publicationProceedings - International Symposium on High-Performance Computer Architecture
Pages318-329
Number of pages12
DOIs
StatePublished - 2013
Event19th IEEE International Symposium on High Performance Computer Architecture, HPCA 2013 - Shenzhen, China
Duration: Feb 23 2013Feb 27 2013

Other

Other19th IEEE International Symposium on High Performance Computer Architecture, HPCA 2013
CountryChina
CityShenzhen
Period2/23/132/27/13

Fingerprint

Particle accelerators
Ultrasonics
Transducers
Image processing
Medical imaging
Pipelines
Mathematical transformations
Imaging techniques

ASJC Scopus subject areas

  • Hardware and Architecture

Cite this

Sampson, R., Yang, M., Wei, S., Chakrabarti, C., & Wenisch, T. F. (2013). Sonic Millip3De: A massively parallel 3D-stacked accelerator for 3D ultrasound. In Proceedings - International Symposium on High-Performance Computer Architecture (pp. 318-329). [6522329] https://doi.org/10.1109/HPCA.2013.6522329

Sonic Millip3De : A massively parallel 3D-stacked accelerator for 3D ultrasound. / Sampson, Richard; Yang, Ming; Wei, Siyuan; Chakrabarti, Chaitali; Wenisch, Thomas F.

Proceedings - International Symposium on High-Performance Computer Architecture. 2013. p. 318-329 6522329.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Sampson, R, Yang, M, Wei, S, Chakrabarti, C & Wenisch, TF 2013, Sonic Millip3De: A massively parallel 3D-stacked accelerator for 3D ultrasound. in Proceedings - International Symposium on High-Performance Computer Architecture., 6522329, pp. 318-329, 19th IEEE International Symposium on High Performance Computer Architecture, HPCA 2013, Shenzhen, China, 2/23/13. https://doi.org/10.1109/HPCA.2013.6522329
Sampson R, Yang M, Wei S, Chakrabarti C, Wenisch TF. Sonic Millip3De: A massively parallel 3D-stacked accelerator for 3D ultrasound. In Proceedings - International Symposium on High-Performance Computer Architecture. 2013. p. 318-329. 6522329 https://doi.org/10.1109/HPCA.2013.6522329
Sampson, Richard ; Yang, Ming ; Wei, Siyuan ; Chakrabarti, Chaitali ; Wenisch, Thomas F. / Sonic Millip3De : A massively parallel 3D-stacked accelerator for 3D ultrasound. Proceedings - International Symposium on High-Performance Computer Architecture. 2013. pp. 318-329
@inproceedings{814880a5c7a44e45bf8a9ac577635960,
title = "Sonic Millip3De: A massively parallel 3D-stacked accelerator for 3D ultrasound",
abstract = "Three-dimensional (3D) ultrasound is becoming common for non-invasive medical imaging because of its high accuracy, safety, and ease of use. Unlike other modalities, ultrasound transducers require little power, which makes hand-held imaging platforms possible, and several low-resolution 2D devices are commercially available today. However, the extreme computational requirements (and associated power requirements) of 3D ultrasound image formation has, to date, precluded hand-held 3D capable devices. We describe the Sonic Millip3De, a new system architecture and accelerator for 3D ultrasound beamformationthe most computationally intensive aspect of image formation. Our three-layer die-stacked design features a custom beamsum accelerator that employs massive data parallelism and a streaming transform-select-reduce pipeline architecture enabled by our new iterative beamsum delay calculation algorithm. Based on RTL-level design and floorplanning for an industrial 45nm process, we show Sonic Millip3De can enable 3D ultrasound with a fully sampled 128×96 transducer array within a 16W full-system power budget (400× less than a conventional DSP solution) and will meet a 5W safe power target by the 11nm node.",
author = "Richard Sampson and Ming Yang and Siyuan Wei and Chaitali Chakrabarti and Wenisch, {Thomas F.}",
year = "2013",
doi = "10.1109/HPCA.2013.6522329",
language = "English (US)",
isbn = "9781467355858",
pages = "318--329",
booktitle = "Proceedings - International Symposium on High-Performance Computer Architecture",

}

TY - GEN

T1 - Sonic Millip3De

T2 - A massively parallel 3D-stacked accelerator for 3D ultrasound

AU - Sampson, Richard

AU - Yang, Ming

AU - Wei, Siyuan

AU - Chakrabarti, Chaitali

AU - Wenisch, Thomas F.

PY - 2013

Y1 - 2013

N2 - Three-dimensional (3D) ultrasound is becoming common for non-invasive medical imaging because of its high accuracy, safety, and ease of use. Unlike other modalities, ultrasound transducers require little power, which makes hand-held imaging platforms possible, and several low-resolution 2D devices are commercially available today. However, the extreme computational requirements (and associated power requirements) of 3D ultrasound image formation has, to date, precluded hand-held 3D capable devices. We describe the Sonic Millip3De, a new system architecture and accelerator for 3D ultrasound beamformationthe most computationally intensive aspect of image formation. Our three-layer die-stacked design features a custom beamsum accelerator that employs massive data parallelism and a streaming transform-select-reduce pipeline architecture enabled by our new iterative beamsum delay calculation algorithm. Based on RTL-level design and floorplanning for an industrial 45nm process, we show Sonic Millip3De can enable 3D ultrasound with a fully sampled 128×96 transducer array within a 16W full-system power budget (400× less than a conventional DSP solution) and will meet a 5W safe power target by the 11nm node.

AB - Three-dimensional (3D) ultrasound is becoming common for non-invasive medical imaging because of its high accuracy, safety, and ease of use. Unlike other modalities, ultrasound transducers require little power, which makes hand-held imaging platforms possible, and several low-resolution 2D devices are commercially available today. However, the extreme computational requirements (and associated power requirements) of 3D ultrasound image formation has, to date, precluded hand-held 3D capable devices. We describe the Sonic Millip3De, a new system architecture and accelerator for 3D ultrasound beamformationthe most computationally intensive aspect of image formation. Our three-layer die-stacked design features a custom beamsum accelerator that employs massive data parallelism and a streaming transform-select-reduce pipeline architecture enabled by our new iterative beamsum delay calculation algorithm. Based on RTL-level design and floorplanning for an industrial 45nm process, we show Sonic Millip3De can enable 3D ultrasound with a fully sampled 128×96 transducer array within a 16W full-system power budget (400× less than a conventional DSP solution) and will meet a 5W safe power target by the 11nm node.

UR - http://www.scopus.com/inward/record.url?scp=84880285819&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84880285819&partnerID=8YFLogxK

U2 - 10.1109/HPCA.2013.6522329

DO - 10.1109/HPCA.2013.6522329

M3 - Conference contribution

AN - SCOPUS:84880285819

SN - 9781467355858

SP - 318

EP - 329

BT - Proceedings - International Symposium on High-Performance Computer Architecture

ER -