Evaluating the Impact of Optimizations for Dynamic Binary Modification on 64-bit RISC-V

John Alistair Kressel; Guillermo Callaghan; Cosmin Gorgovan; Mikel Luján

doi:10.1109/ISPASS57527.2023.00017

Evaluating the Impact of Optimizations for Dynamic Binary Modification on 64-bit RISC-V

John Alistair Kressel, Guillermo Callaghan, Cosmin Gorgovan, Mikel Luján

Advanced Processor Technology

Research output: Chapter in Book/Conference proceeding › Conference contribution › peer-review

367 Downloads (Pure)

Abstract

Dynamic Binary Modification (DBM) is an important technique used in computer architecture simulators, virtualization, and program analysis, to name a few examples. The software ecosystem of RISC-V is maturing at pace, but is still missing a high-performance, optimized DBM. Addressing this requirement is key to improving the overall software ecosystem.
This paper presents a comprehensive performance evaluation study for a DBM (MAMBO) which has been ported and optimized for 64-bit RISC-V. The main optimizations for DBM on RISC architectures have been implemented and tuned for RISC-V to address specific architectural features. For example, jump trampolines have been specifically developed to address the short direct branch range specified by the RISC-V ISA.
The evaluation shows that for SPEC CPU2006 the geometric mean overhead is of 14.5%, with SPECint having the largest contribution with a geometric mean of 28.5%, while SPECfp has only an overhead of 5.6%. Concretely, this results in a reduction in runtime for h264ref from over 75 hours using the baseline DBM, to 2.2 hours with optimizations applied.

Original language	English
Title of host publication	2023 IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS)
Publisher	IEEE
Pages	81-91
Number of pages	11
ISBN (Electronic)	9798350397390
ISBN (Print)	9798350397390
DOIs	https://doi.org/10.1109/ISPASS57527.2023.00017
Publication status	Published - 24 Apr 2023

Publication series

Name	Proceedings - 2023 IEEE International Symposium on Performance Analysis of Systems and Software, ISPASS 2023

Keywords

Binary Code Profiling
Dynamic Binary Instrumentation
Dynamic Binary Modification
RISC-V

Access to Document

10.1109/ISPASS57527.2023.00017Licence: Other

MAMBO_RISCV_ISPASS2023_PREPRINTAccepted author manuscript, 0.99 MBLicence: CC BY

Cite this

Kressel, J. A., Callaghan, G., Gorgovan, C., & Luján, M. (2023). Evaluating the Impact of Optimizations for Dynamic Binary Modification on 64-bit RISC-V. In 2023 IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS) (pp. 81-91). (Proceedings - 2023 IEEE International Symposium on Performance Analysis of Systems and Software, ISPASS 2023). IEEE. https://doi.org/10.1109/ISPASS57527.2023.00017

Kressel, John Alistair ; Callaghan, Guillermo ; Gorgovan, Cosmin et al. / Evaluating the Impact of Optimizations for Dynamic Binary Modification on 64-bit RISC-V. 2023 IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS). IEEE, 2023. pp. 81-91 (Proceedings - 2023 IEEE International Symposium on Performance Analysis of Systems and Software, ISPASS 2023).

@inproceedings{3f71f4aa6628440d93ad672da2d452d5,

title = "Evaluating the Impact of Optimizations for Dynamic Binary Modification on 64-bit RISC-V",

abstract = "Dynamic Binary Modification (DBM) is an important technique used in computer architecture simulators, virtualization, and program analysis, to name a few examples. The software ecosystem of RISC-V is maturing at pace, but is still missing a high-performance, optimized DBM. Addressing this requirement is key to improving the overall software ecosystem.This paper presents a comprehensive performance evaluation study for a DBM (MAMBO) which has been ported and optimized for 64-bit RISC-V. The main optimizations for DBM on RISC architectures have been implemented and tuned for RISC-V to address specific architectural features. For example, jump trampolines have been specifically developed to address the short direct branch range specified by the RISC-V ISA.The evaluation shows that for SPEC CPU2006 the geometric mean overhead is of 14.5%, with SPECint having the largest contribution with a geometric mean of 28.5%, while SPECfp has only an overhead of 5.6%. Concretely, this results in a reduction in runtime for h264ref from over 75 hours using the baseline DBM, to 2.2 hours with optimizations applied.",

keywords = "Binary Code Profiling, Dynamic Binary Instrumentation, Dynamic Binary Modification, RISC-V",

author = "Kressel, {John Alistair} and Guillermo Callaghan and Cosmin Gorgovan and Mikel Luj{\'a}n",

note = "Funding Information: This research was partially supported by the UK Industrial Strategy Challenge Fund (ISCF) under the Digital Security by Design (DSbD) Programme delivered by UKRI as part of the MoatE (10017512) and Soteria (75243) projects, and by EPSRC Rain Hub (EP/R026084/1), and LAMBDA (EP/N035127/1) projects. Mikel Luj an is supported by a Royal Society Wolfson Fellowship and an Arm/RAEng Research Chair Award. SiFive donated one SiFive Unmatched board used in the experiments. Publisher Copyright: {\textcopyright} 2023 IEEE.",

year = "2023",

month = apr,

day = "24",

doi = "10.1109/ISPASS57527.2023.00017",

language = "English",

isbn = "9798350397390",

series = "Proceedings - 2023 IEEE International Symposium on Performance Analysis of Systems and Software, ISPASS 2023",

publisher = "IEEE",

pages = "81--91",

booktitle = "2023 IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS)",

address = "United States",

}

Kressel, JA, Callaghan, G, Gorgovan, C & Luján, M 2023, Evaluating the Impact of Optimizations for Dynamic Binary Modification on 64-bit RISC-V. in 2023 IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS). Proceedings - 2023 IEEE International Symposium on Performance Analysis of Systems and Software, ISPASS 2023, IEEE, pp. 81-91. https://doi.org/10.1109/ISPASS57527.2023.00017

Evaluating the Impact of Optimizations for Dynamic Binary Modification on 64-bit RISC-V. / Kressel, John Alistair; Callaghan, Guillermo; Gorgovan, Cosmin et al.
2023 IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS). IEEE, 2023. p. 81-91 (Proceedings - 2023 IEEE International Symposium on Performance Analysis of Systems and Software, ISPASS 2023).

Research output: Chapter in Book/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Evaluating the Impact of Optimizations for Dynamic Binary Modification on 64-bit RISC-V

AU - Kressel, John Alistair

AU - Callaghan, Guillermo

AU - Gorgovan, Cosmin

AU - Luján, Mikel

N1 - Funding Information: This research was partially supported by the UK Industrial Strategy Challenge Fund (ISCF) under the Digital Security by Design (DSbD) Programme delivered by UKRI as part of the MoatE (10017512) and Soteria (75243) projects, and by EPSRC Rain Hub (EP/R026084/1), and LAMBDA (EP/N035127/1) projects. Mikel Luj an is supported by a Royal Society Wolfson Fellowship and an Arm/RAEng Research Chair Award. SiFive donated one SiFive Unmatched board used in the experiments. Publisher Copyright: © 2023 IEEE.

PY - 2023/4/24

Y1 - 2023/4/24

N2 - Dynamic Binary Modification (DBM) is an important technique used in computer architecture simulators, virtualization, and program analysis, to name a few examples. The software ecosystem of RISC-V is maturing at pace, but is still missing a high-performance, optimized DBM. Addressing this requirement is key to improving the overall software ecosystem.This paper presents a comprehensive performance evaluation study for a DBM (MAMBO) which has been ported and optimized for 64-bit RISC-V. The main optimizations for DBM on RISC architectures have been implemented and tuned for RISC-V to address specific architectural features. For example, jump trampolines have been specifically developed to address the short direct branch range specified by the RISC-V ISA.The evaluation shows that for SPEC CPU2006 the geometric mean overhead is of 14.5%, with SPECint having the largest contribution with a geometric mean of 28.5%, while SPECfp has only an overhead of 5.6%. Concretely, this results in a reduction in runtime for h264ref from over 75 hours using the baseline DBM, to 2.2 hours with optimizations applied.

AB - Dynamic Binary Modification (DBM) is an important technique used in computer architecture simulators, virtualization, and program analysis, to name a few examples. The software ecosystem of RISC-V is maturing at pace, but is still missing a high-performance, optimized DBM. Addressing this requirement is key to improving the overall software ecosystem.This paper presents a comprehensive performance evaluation study for a DBM (MAMBO) which has been ported and optimized for 64-bit RISC-V. The main optimizations for DBM on RISC architectures have been implemented and tuned for RISC-V to address specific architectural features. For example, jump trampolines have been specifically developed to address the short direct branch range specified by the RISC-V ISA.The evaluation shows that for SPEC CPU2006 the geometric mean overhead is of 14.5%, with SPECint having the largest contribution with a geometric mean of 28.5%, while SPECfp has only an overhead of 5.6%. Concretely, this results in a reduction in runtime for h264ref from over 75 hours using the baseline DBM, to 2.2 hours with optimizations applied.

KW - Binary Code Profiling

KW - Dynamic Binary Instrumentation

KW - Dynamic Binary Modification

KW - RISC-V

UR - http://www.scopus.com/inward/record.url?scp=85164539620&partnerID=8YFLogxK

UR - https://www.mendeley.com/catalogue/71ca2df4-a724-3e7e-8950-400329d94dbd/

U2 - 10.1109/ISPASS57527.2023.00017

DO - 10.1109/ISPASS57527.2023.00017

M3 - Conference contribution

SN - 9798350397390

T3 - Proceedings - 2023 IEEE International Symposium on Performance Analysis of Systems and Software, ISPASS 2023

SP - 81

EP - 91

BT - 2023 IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS)

PB - IEEE

ER -

Kressel JA, Callaghan G, Gorgovan C, Luján M. Evaluating the Impact of Optimizations for Dynamic Binary Modification on 64-bit RISC-V. In 2023 IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS). IEEE. 2023. p. 81-91. (Proceedings - 2023 IEEE International Symposium on Performance Analysis of Systems and Software, ISPASS 2023). doi: 10.1109/ISPASS57527.2023.00017

Evaluating the Impact of Optimizations for Dynamic Binary Modification on 64-bit RISC-V

Abstract

Publication series

Keywords

Access to Document

Other files and links

Fingerprint

Robotics and Artificial Intelligence for Nuclear (RAIN)

Cite this

Evaluating the Impact of Optimizations for Dynamic Binary Modification on 64-bit RISC-V

Abstract

Publication series

Keywords

Access to Document

Other files and links

Fingerprint

Projects

Robotics and Artificial Intelligence for Nuclear (RAIN)

Cite this