請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/49281
完整後設資料紀錄
DC 欄位 | 值 | 語言 |
---|---|---|
dc.contributor.advisor | 洪士灝(Shih-Hao Hung) | |
dc.contributor.author | Chun-Yen Ho | en |
dc.contributor.author | 何駿延 | zh_TW |
dc.date.accessioned | 2021-06-15T11:21:59Z | - |
dc.date.available | 2026-12-29 | |
dc.date.copyright | 2017-03-09 | |
dc.date.issued | 2016 | |
dc.date.submitted | 2016-08-17 | |
dc.identifier.citation | Bibliography
[1] Intel® architecture instruction set extensions programming reference. http://www.cs.utexas.edu/~marijn/class/2016-spring/cs350c/documents/Intel-x86-Docs/64-ia-32-architectures-instruction-set-extensions-reference-manual.pdf, 2013. [2] Intel® vtune™ amplifier 2016. https://software.intel.com/en-us/intel-vtune-amplifier-xe, 2013. [3] Intel® xeon phi™ coprocessor 7120p. http://ark.intel.com/zh-tw/products/75799/Intel-Xeon-Phi-Coprocessor-7120P-16GB-1_238-GHz-61-core, 2013. [4] Intel® xeon phi™ products. http://www.intel.com.tw/content/www/tw/zh/processors/xeon/xeon-phi-detail.html, 2013. [5] Intel® xeon® processor e5-2630 v2. http://ark.intel.com/zh-tw/products/75790/Intel-Xeon-Processor-E5-2630-v2-15M-Cache-2_60-GHz, 2013. [6] J. Cassidy, L. Lilge, and V. Betz. Fast, power-efficient biophotonic simulationsfor cancer treatment using fpgas. In Field-Programmable Custom Computing Machines(FCCM), 2014 IEEE 22nd Annual International Symposium on, pages 133–140. IEEE, 2014. [7] S.-H. Hung, M.-Y. Tsai, B.-Y. Huang, and C.-H. Tu. A platform-oblivious approach for heterogeneous computing: A case study with monte carlo-based simulation for medical applications. In Proceedings of the 2016 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays, pages 42–47. ACM, 2016. [8] B. Shareef. Monte carlo simulations on xeon phi: Offload and native mode. [9] L. Wang, S. L. Jacques, and L. Zheng. Mcml—monte carlo modeling of light transportin multi-layered tissues. Computer methods and programs in biomedicine, 47(2):131–146, 1995. | |
dc.identifier.uri | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/49281 | - |
dc.description.abstract | 近年來,在醫療方面特別是癌症治療光子被視為不可或缺的角色,為了增進醫療的品質,我們需要透過電腦精確的運算或模擬,一般來說以蒙地卡羅方法為基礎的模擬被大眾視為比較精準的方法,卻需要龐大的運算資源。身為最普及運算資源的中央運算元(CPU)在這類的應用中扮演重要的角色,但卻很少人有效地使用向量來加速在中央運算元的運算,且在Xeon Phi協同處理器中效能的好壞取決於向量化的程度。在這篇的論文中,我使用OpenMP平行並向量化了MCML程式,而後也進一步討論在這些平行與向量化過後的MCML程式跑在Xeon CPU及Xeon Phi協同處理器上的效能。 | zh_TW |
dc.description.abstract | Recently, photon is indispensable in many medical applications, such as cancer treatment. After accurate computing or simulating, it improve the quality of medical treatments. Generally, Monte Carlo-based simulations are considered to deliver accurate results, but require intensive computational resources. CPU (Central Processing Unit), the most universal resource, plays an important role in computing the application. However, less researches effectively use vector to accelerate the computations on CPU. Furthermore, on Xeon Phi coprocessor, the performance depends on degree of vectorization. In this paper, we parallelized and vectorized the Monte Carlo modeling of light transport in multi-layered tissues (MCML) program with OpenMP.We then discussed the performance of the parallelized and vectorized MCML kernel program runs on Xeon CPU and Xeon Phi coprocessor. | en |
dc.description.provenance | Made available in DSpace on 2021-06-15T11:21:59Z (GMT). No. of bitstreams: 1 ntu-105-R03922077-1.pdf: 2108446 bytes, checksum: 31ddc0d9ece4de6d05669015563e3d27 (MD5) Previous issue date: 2016 | en |
dc.description.tableofcontents | 口試委員會審定書 iii
誌謝 v 摘要 vii Abstract ix 1 Introduction 1 2 Background 3 2.1 MCML . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 2.2 Xeon Phi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 2.3 AVX-512 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 3 MCML ACCELERATION 7 3.1 Parallel Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 3.1.1 Branch A: Trace . . . . . . . . . . . . . . . . . . . . . . . . . . 8 3.1.2 A.1 No Memset . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 3.1.3 Branch B: No Trace . . . . . . . . . . . . . . . . . . . . . . . . 9 3.1.4 B.1 Dynamic Scheduling . . . . . . . . . . . . . . . . . . . . . . 10 3.2 Vector Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 3.2.1 B.3: Auto Vectorization . . . . . . . . . . . . . . . . . . . . . . 11 3.2.2 B.4: Random Function Vectorization . . . . . . . . . . . . . . . 12 3.2.3 B.5: Use Intel Math Kernel Library . . . . . . . . . . . . . . . . 14 3.2.4 B.6: Vectorization on HopDropSpin Functions . . . . . . . . . . 14 3.2.5 B.7: Vectorization on Step Functions . . . . . . . . . . . . . . . 14 3.2.6 B.8: Intrinsics on MCML . . . . . . . . . . . . . . . . . . . . . 16 3.2.7 Further Reducing Branch Divergence? . . . . . . . . . . . . . . . 16 4 Experimental Results 19 4.1 Hardware Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . 19 4.2 Original MCML . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 4.3 Parallel Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 5 Conclusion 25 5.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 5.2 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 Bibliography 27 | |
dc.language.iso | en | |
dc.title | 加速異質架構上蒙地卡羅於光子治療的模擬 | zh_TW |
dc.title | Accelerating Monte Carlo Simulation for Photon Therapy with Heterogeneous Computing | en |
dc.type | Thesis | |
dc.date.schoolyear | 104-2 | |
dc.description.degree | 碩士 | |
dc.contributor.oralexamcommittee | 廖世偉(Shih-wei Liao),施吉昇(Chi-Sheng Shih),李曜琮,涂嘉恆(Chia-Heng Tu) | |
dc.subject.keyword | 至,MCML,蒙地卡羅,AVX-512, | zh_TW |
dc.subject.keyword | Xeon Phi,MCML,Monte Carlo,AVX-512, | en |
dc.relation.page | 28 | |
dc.identifier.doi | 10.6342/NTU201603202 | |
dc.rights.note | 有償授權 | |
dc.date.accepted | 2016-08-19 | |
dc.contributor.author-college | 電機資訊學院 | zh_TW |
dc.contributor.author-dept | 資訊工程學研究所 | zh_TW |
顯示於系所單位: | 資訊工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-105-1.pdf 目前未授權公開取用 | 2.06 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。