Publication detail

Approximating Complex Arithmetic Circuits with Guaranteed Worst-Case Relative Error

MATYÁŠ, J. PANKUCH, A. VOJNAR, T. ČEŠKA, M. ČEŠKA, M.

Original Title

Approximating Complex Arithmetic Circuits with Guaranteed Worst-Case Relative Error

Type

conference paper

Language

English

Original Abstract

We present a novel method allowing one to approximate complex arithmetic circuits with formal guarantees on the worst-case relative error, abbreviated as WCRE. WCRE represents an important error metric relevant in many applications including, e.g., approximation of neural network HW architectures. The method integrates SAT-based error evaluation of approximate circuits into a verifiability-driven search algorithm based on Cartesian genetic programming. We implement the method in our framework ADAC that provides various techniques for automated design of arithmetic circuits. Our experimental evaluation shows that, in many cases, the method offers a superior scalability and allows us to construct, within a few hours, high-quality approximations (providing trade-offs between the WCRE and size) for circuits with up to 32-bit operands. As such, it significantly improves the capabilities of ADAC.

Keywords

automated circuit design, approximate circuits, SAT

Authors

MATYÁŠ, J.; PANKUCH, A.; VOJNAR, T.; ČEŠKA, M.; ČEŠKA, M.

Released

15. 4. 2020

Publisher

Springer Verlag

Location

Cham

ISBN

978-3-030-45092-2

Book

International Conference on Computer Aided Systems Theory (EUROCAST'19)

Edition

Lecture Notes in Computer Science

Pages from

482

Pages to

490

Pages count

9

BibTex

@inproceedings{BUT168134,
  author="Jiří {Matyáš} and Adam {Pankuch} and Tomáš {Vojnar} and Milan {Češka} and Milan {Češka}",
  title="Approximating Complex Arithmetic Circuits with Guaranteed Worst-Case Relative Error",
  booktitle="International Conference on Computer Aided Systems Theory (EUROCAST'19)",
  year="2020",
  series="Lecture Notes in Computer Science",
  volume="12013",
  pages="482--490",
  publisher="Springer Verlag",
  address="Cham",
  doi="10.1007/978-3-030-45093-9\{_}58",
  isbn="978-3-030-45092-2"
}