An FPGA acceleration for the kd-tree search in photon mapping

Takuya Kuhara, Takaaki Miyajima, Masato Yoshimi, Hideharu Amano

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

5 Citations (Scopus)

Abstract

Photon mapping is a kind of rendering techniques which enables depicting complicated light concentrations for 3D graphics. Searching kd-tree of photons with k-near neighbor search (k-NN) requires a large amount of computations. As k-NN search includes high degree of parallelism, the operation can be accelerated by GPU and recent multi-core microprocessors. However, memory access bottleneck will limit their computation speed. Here, as an alternative approach, an FPGA implementation of k-NN search operation in kd-tree is proposed. In the proposed design, we maximized the effective throughput of the block RAM by connecting multiple Query Modules to both ports of RAM. Furthermore, an implementation of the discovery process of the max distance which is not depending on the number of Estimate-Photons is proposed. Through the implementation on Spartan6, Virtex6 and Virtex7, it appears that 26 fundamental modules can be mounted on Virtex7. As a result, the proposed module achieved the throughput of approximately 282 times as that of software execution at maximum.

Original languageEnglish
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Pages25-36
Number of pages12
Volume7806 LNCS
DOIs
Publication statusPublished - 2013
Event9th International Symposium on Applied Reconfigurable Computing, ARC 2013 - Los Angeles, CA, United States
Duration: 2013 Mar 252013 Mar 27

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume7806 LNCS
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other9th International Symposium on Applied Reconfigurable Computing, ARC 2013
CountryUnited States
CityLos Angeles, CA
Period13/3/2513/3/27

Fingerprint

Kd-tree
Nearest Neighbor Search
Field Programmable Gate Array
Field programmable gate arrays (FPGA)
Photon
Photons
Random access storage
Module
Throughput
3D Graphics
FPGA Implementation
Microprocessor
Rendering
Parallelism
Microprocessor chips
Query
Data storage equipment
Software
Alternatives
Estimate

Keywords

  • Acceleration
  • FPGA
  • k-NN
  • kd-tree
  • Photon Mapping

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Kuhara, T., Miyajima, T., Yoshimi, M., & Amano, H. (2013). An FPGA acceleration for the kd-tree search in photon mapping. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 7806 LNCS, pp. 25-36). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7806 LNCS). https://doi.org/10.1007/978-3-642-36812-7_3

An FPGA acceleration for the kd-tree search in photon mapping. / Kuhara, Takuya; Miyajima, Takaaki; Yoshimi, Masato; Amano, Hideharu.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 7806 LNCS 2013. p. 25-36 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7806 LNCS).

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

Kuhara, T, Miyajima, T, Yoshimi, M & Amano, H 2013, An FPGA acceleration for the kd-tree search in photon mapping. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). vol. 7806 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 7806 LNCS, pp. 25-36, 9th International Symposium on Applied Reconfigurable Computing, ARC 2013, Los Angeles, CA, United States, 13/3/25. https://doi.org/10.1007/978-3-642-36812-7_3
Kuhara T, Miyajima T, Yoshimi M, Amano H. An FPGA acceleration for the kd-tree search in photon mapping. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 7806 LNCS. 2013. p. 25-36. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-642-36812-7_3
Kuhara, Takuya ; Miyajima, Takaaki ; Yoshimi, Masato ; Amano, Hideharu. / An FPGA acceleration for the kd-tree search in photon mapping. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 7806 LNCS 2013. pp. 25-36 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{564432b2d6bf4fcdb69705a28a38827d,
title = "An FPGA acceleration for the kd-tree search in photon mapping",
abstract = "Photon mapping is a kind of rendering techniques which enables depicting complicated light concentrations for 3D graphics. Searching kd-tree of photons with k-near neighbor search (k-NN) requires a large amount of computations. As k-NN search includes high degree of parallelism, the operation can be accelerated by GPU and recent multi-core microprocessors. However, memory access bottleneck will limit their computation speed. Here, as an alternative approach, an FPGA implementation of k-NN search operation in kd-tree is proposed. In the proposed design, we maximized the effective throughput of the block RAM by connecting multiple Query Modules to both ports of RAM. Furthermore, an implementation of the discovery process of the max distance which is not depending on the number of Estimate-Photons is proposed. Through the implementation on Spartan6, Virtex6 and Virtex7, it appears that 26 fundamental modules can be mounted on Virtex7. As a result, the proposed module achieved the throughput of approximately 282 times as that of software execution at maximum.",
keywords = "Acceleration, FPGA, k-NN, kd-tree, Photon Mapping",
author = "Takuya Kuhara and Takaaki Miyajima and Masato Yoshimi and Hideharu Amano",
year = "2013",
doi = "10.1007/978-3-642-36812-7_3",
language = "English",
isbn = "9783642368110",
volume = "7806 LNCS",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "25--36",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

TY - GEN

T1 - An FPGA acceleration for the kd-tree search in photon mapping

AU - Kuhara, Takuya

AU - Miyajima, Takaaki

AU - Yoshimi, Masato

AU - Amano, Hideharu

PY - 2013

Y1 - 2013

N2 - Photon mapping is a kind of rendering techniques which enables depicting complicated light concentrations for 3D graphics. Searching kd-tree of photons with k-near neighbor search (k-NN) requires a large amount of computations. As k-NN search includes high degree of parallelism, the operation can be accelerated by GPU and recent multi-core microprocessors. However, memory access bottleneck will limit their computation speed. Here, as an alternative approach, an FPGA implementation of k-NN search operation in kd-tree is proposed. In the proposed design, we maximized the effective throughput of the block RAM by connecting multiple Query Modules to both ports of RAM. Furthermore, an implementation of the discovery process of the max distance which is not depending on the number of Estimate-Photons is proposed. Through the implementation on Spartan6, Virtex6 and Virtex7, it appears that 26 fundamental modules can be mounted on Virtex7. As a result, the proposed module achieved the throughput of approximately 282 times as that of software execution at maximum.

AB - Photon mapping is a kind of rendering techniques which enables depicting complicated light concentrations for 3D graphics. Searching kd-tree of photons with k-near neighbor search (k-NN) requires a large amount of computations. As k-NN search includes high degree of parallelism, the operation can be accelerated by GPU and recent multi-core microprocessors. However, memory access bottleneck will limit their computation speed. Here, as an alternative approach, an FPGA implementation of k-NN search operation in kd-tree is proposed. In the proposed design, we maximized the effective throughput of the block RAM by connecting multiple Query Modules to both ports of RAM. Furthermore, an implementation of the discovery process of the max distance which is not depending on the number of Estimate-Photons is proposed. Through the implementation on Spartan6, Virtex6 and Virtex7, it appears that 26 fundamental modules can be mounted on Virtex7. As a result, the proposed module achieved the throughput of approximately 282 times as that of software execution at maximum.

KW - Acceleration

KW - FPGA

KW - k-NN

KW - kd-tree

KW - Photon Mapping

UR - http://www.scopus.com/inward/record.url?scp=84875486846&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84875486846&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-36812-7_3

DO - 10.1007/978-3-642-36812-7_3

M3 - Conference contribution

SN - 9783642368110

VL - 7806 LNCS

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 25

EP - 36

BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

ER -