CIRCA: Towards a modular and extensible framework for approximate circuit generation

Witschen, Linus; Rajput, Muhammad Awais; Mohammadi, Hassan Ghasemzadeh; Wiersema, Tobias; Platzner, Marco

doi:10.1016/j.microrel.2019.04.003

Cited by 15 publications

(4 citation statements)

References 2 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, the choice of a search method to explore the possible configurations of the approximate designs heavily impacts the quality of the results. In fact, the circuits found by one search method could sometimes be up to 28% smaller when compared to the circuits found by another search method [3]. However, this comes at a price of computational complexity.…”

Section: Search-based Design Space Exploration For Generation Of Appr...mentioning

confidence: 99%

“…A challenging task for automated approximate hardware generation is to handle the explosion of combinations that are possible during the process. Since exhaustive selection is not feasible, a more efficient approach is to employ heuristics to converge to a solution in a reasonable time [3].…”

Section: Related Workmentioning

confidence: 99%

“…and machine learning tasks [1], [2]. For a large number of approximation frameworks, component substitution has been extensively adopted as a design strategy for automated approximate circuits generation methodologies where a functionally accurate circuit is modified iteratively by applying a series of approximate transformations in order to get a number of approximate design(s) that adhere to the quality bounds [3], [4], [5], [6].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Improved Learning-Based Design Space Exploration for Approximate Instance Generation

et al. 2023

View full text Add to dashboard Cite

Design methodologies for approximation of medium to large-scale accelerators have largely relied on search-based design space exploration. Due to the enormously sized solution space, Artificial Intelligence (AI) based heuristic search has remained one of the most common techniques to explore cost and performance trade-offs. As a sub-class of AI techniques, Monte Carlo Tree Search (MCTS) has recently shown great potential as an intelligent stochastic search algorithm to solve computational problems with large branching factors. MCTS employs reinforced learning that combines past statistics and stochastic processes to traverse new paths in the search tree. Inspired by the success of MCTS in the games domain (AlphaGo), it has been recently applied to explore the solution space of approximate circuits. In this paper, a modified MCTS-based intelligent search technique is proposed that can handle huge search space for approximation of fairly large benchmark circuits. The proposed learning-based heuristic for MCTS directs the search towards deeper nodes in the search tree resulting in a rather asymmetric tree to efficiently utilize the search budget on the exploration of more promising nodes in the design space. The modified MCTS algorithm is based on reinforced learning where an agent uses past information to improve the overall gain. Experimental results confirm that the proposed heuristics can reach out to deeper nodes in the search tree which account for larger area savings in the context of approximate computing. The modified search algorithm enables 34.23% more area savings than the original search algorithm.

show abstract

Section: Search-based Design Space Exploration For Generation Of Appr...mentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Improved Learning-Based Design Space Exploration for Approximate Instance Generation

et al. 2023

View full text Add to dashboard Cite

show abstract

“…After the synthesis, the design is sent back to the user who runs the test vectors to verify the according to the given constraints, such as error metrics, but there is no verification performed by the user against potential hardware Trojans. To investigate the aforementioned threat model, the approximate circuit synthesis flows proposed in [108,109] could be manipulated with malicious approximate components (Trojan), which likely fulfills the constraints such as error and performance parameters, but could lead to catastrophic results later when the Trojan is activated. Moreover, the Trojan insertion has been so far targeted at the netlist level whereas the state-of-the-art approximate circuit synthesis frameworks, i.e., [108] target higher levels of abstraction such as SystemC.…”

Section: Chapter Conclusionmentioning

confidence: 99%

Hardware Trojans in Reconfigurable Computing

Ahmed

2021

2021 IFIP/IEEE 29th International Conference on Very Large Scale Integration (VLSI-SoC)

View full text Add to dashboard Cite

Inestimable appreciation and earnest gratitude for the help and support are extended to many remarkable people who have been around me throughout this journey and without whom the accomplishment of this thesis would not have been possible. I would like to meekly express my thanks to all my colleagues, friends, family, and specifically to those mentioned in the following lines.First and foremost, I would like to express my deepest thanks to my supervisor, Prof. Dr. Marco Platzner, for his invaluable guidance, persistent help, and immeasurable support throughout this research. He has always been kind and his generosity, dynamism, sincerity, scholarly advice, and professional attitude have deeply inspired, motivated and helped me to a great extent to complete this thesis. It was a great privilege and honor to work under his supervision where I have learnt a lot from his experience, skills, and leadership. Furthermore, I would like to thank him and Prof. Dr. Sybille Hellebrand for devoting their time and effort to review this thesis, together with all the committee members for evaluating my research.I am sincerely grateful to the people of the CEG for providing such a pleasant and professional working environment during my stay. A special thanks goes to my colleague cum mentor Tobias Wiersema who has been there all the time for scientific discussions in a very professional way and also his help regarding social matters in Germany is highly regarded.Besides, many thanks also goes to

show abstract