Abstract
Machine learning accelerators (MLAs) are increasingly important in many applications such as image and video processing, speech recognition, and natural language processing. To achieve the needed performances and power efficiencies, MLAs are highly concurrent. The correctness of MLAs hinges on the concept of sequential consistency, i.e., the concurrent execution of a program by an MLA must be equivalent to a sequential execution of the program. In this paper, we certify the sequential consistency of modular MLAs using theorem proving. We first provide a formalization of the MLAs and define their sequential consistency. After that, we introduce our certification methodology based on inductive theorem proving. Finally, we demonstrate the feasibility of our approach through the analysis of the NVIDIA Deep Learning Accelerator and the Versatile Tensor Accelerator.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Bertot, Y., Castéran, P.: Interactive Theorem Proving and Program Development: Coq’Art: the Calculus of Inductive Constructions. Springer, Heidelberg (2013). https://doi.org/10.1007/978-3-662-07964-5
Cyrluk, D., Rajan, S., Shankar, N., Srivas, M.K.: Effective theorem proving for hardware verification. In: Kumar, R., Kropf, T. (eds.) TPCD 1994. LNCS, vol. 901, pp. 203–222. Springer, Heidelberg (1995). https://doi.org/10.1007/3-540-59047-1_50
Damm, W., Pnueli, A.: Verifying out-of-order executions. In: Advances in Hardware Design and Verification. IAICT, pp. 23–47. Springer, Boston, MA (1997). https://doi.org/10.1007/978-0-387-35190-2_3
Hoare, C.A.R.: Communicating sequential processes. Commun. ACM 21(8), 666–677 (1978)
Jouppi, N.P., et al.: In-datacenter performance analysis of a tensor processing unit. In: Proceedings of the 44th Annual International Symposium on Computer Architecture, pp. 1–12 (2017)
Kaufmann, M., Moore, J.S.: ACL2: an industrial strength version of Nqthm. In: Proceedings of 11th Annual Conference on Computer Assurance. COMPASS 1996, pp. 23–34. IEEE (1996)
Keller, R.M.: Formal verification of parallel programs. Commun. ACM 19(7), 371–384 (1976)
Kroening, D., Paul, W.J., Mueller, S.M.: Proving the correctness of pipelined micro-architectures. In: MBMV, pp. 89–98 (2000)
Leino, K.R.M.: Dafny: an automatic program verifier for functional correctness. In: Clarke, E.M., Voronkov, A. (eds.) LPAR 2010. LNCS (LNAI), vol. 6355, pp. 348–370. Springer, Heidelberg (2010). https://doi.org/10.1007/978-3-642-17511-4_20
Moreau, T., et al.: A hardware-software blueprint for flexible deep learning specialization. IEEE Micro 39(5), 8–16 (2019)
Nvidia: Nvidia deep learning accelerator (2018). http://nvdla.org/primer.html
Pnueli, A., Shankar, N., Singerman, E.: Fair synchronous transition systems and their liveness proofs. In: Ravn, A.P., Rischel, H. (eds.) FTRTFT 1998. LNCS, vol. 1486, pp. 198–209. Springer, Heidelberg (1998). https://doi.org/10.1007/BFb0055348
Rao, N., et al.: Intel Nervana: a next-generation neural network processor. In: Proceedings of the 2017 IEEE Hot Chips Symposium on High Performance Chips (HOTCHIPS), pp. 1–28. IEEE, Cupertino, CA, USA, August 2017
Sawada, J., Hunt, W.A. Jr.: Processor verification with precise exceptions and speculative execution. In: CAV, vol. 98, pp. 135–146 (1998)
Vijayaraghavan, M., Chlipala, A., Dave, N.: Modular deductive verification of multiprocessor hardware designs. In: Kroening, D., Păsăreanu, C. (eds.) Computer Aided Verification. CAV 2015. LNCS, San Francisco, CA, USA, July 18–24, 2015, Proceedings, Part II 27, vol. 9207, pp. 109–127. Springer, Cham (2015). https://doi.org/10.1007/978-3-319-21668-3_7
Acknowledgment
This research is partially supported by a gift from Intel Corporation.
Author information
Authors and Affiliations
Corresponding authors
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2023 The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.
About this paper
Cite this paper
Wu, H., Xie, F., Yang, Z. (2023). Certifying Sequential Consistency of Machine Learning Accelerators. In: Li, Y., Tahar, S. (eds) Formal Methods and Software Engineering. ICFEM 2023. Lecture Notes in Computer Science, vol 14308. Springer, Singapore. https://doi.org/10.1007/978-981-99-7584-6_10
Download citation
DOI: https://doi.org/10.1007/978-981-99-7584-6_10
Published:
Publisher Name: Springer, Singapore
Print ISBN: 978-981-99-7583-9
Online ISBN: 978-981-99-7584-6
eBook Packages: Computer ScienceComputer Science (R0)