Performing Stochastic Computation Deterministically

Najafi, M. Hassan; Jenson, Devon; Lilja, David J.; Riedel, Marc

doi:10.1109/iscas45731.2020.9180521

Cited by 2 publications

(6 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It can be deduced that the smallest possible latency for a stochastic multiplier design is equal to

2^{N}

clock cycles, because any design faster than that would not even be able to receive complete inputs. Unlike deterministic techniques in [4], in order to achieve this minimum latency, repetition of input streams is not needed.…”

Section: Latency and Variable Precisionmentioning

confidence: 99%

“…The result at

2^{N}

th clock cycle is the exact full‐precision product. Latency is reduced by order of

2^{N}

compared to all the sequences and techniques suggested in [4].…”

Section: Latency and Variable Precisionmentioning

confidence: 99%

“…Comparison with state‐of‐the‐art in [4] is done by simulating the proposed design with all the possible input pairs for

N = 8

. The proposed design is implemented with three types of SNGs, namely counter, linear feedback shift register (LFSR), and Sobol sequence.…”

Section: Latency and Variable Precisionmentioning

confidence: 99%

“…Mean absolute error (MAE) is then calculated for different numbers of operation cycles. The results are recorded in Table 2, each of them compared with the smallest MAE obtained from [4] for conventional unary multiplier. MAE is defined as

MAE ()(%) = \frac{\sum_{i = 0}^{255} \sum_{j = 0}^{255} (||, P_{normalfull} - P_{normalpartial})}{2^{16}} \times 100

where

P_{normalfull}

is the full‐precision 16‐bit product of i and j acting as reference, whereas

P_{normalpartial}

is the obtained product with lower precision which is scaled to 16 bits.…”

Section: Latency and Variable Precisionmentioning

confidence: 99%

“…Najafi et al . [4] introduced various deterministic techniques including rotation, clock division and relatively‐prime methods to perform error‐free stochastic computing. Through ‘convolution’ of the input bit streams, every bit in a stream sees every bit in the other streams.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Fast and exact multiple‐input unary‐to‐binary multiplier with variable precision for stochastic computing

Chuah

Kumar

2020

Electron. lett.

View full text Add to dashboard Cite

Stochastic computing offers an area-efficient solution for power-greedy computations in error-resilient applications such as image processing. However, stochastic computing produces only approximated results and suffers from long latency especially for multiplication. The authors present a multiplier design technique for stochastic computing which guarantees exact output and has latency in the order of 2 N (N = input precision), utilising the counter for unary-to-binary conversion. Compared to state-of-the-art deterministic methods the proposed design is faster by a factor of 2 N. Besides, the area of proposed design scales up in proportion to N instead of N 2 , as opposed to conventional binary multipliers. The design technique can be further extended for multiple inputs and parallel computing, Extensive mathematical analysis and simulation results are presented for each variation throughout this Letter.

show abstract

“…It can be deduced that the smallest possible latency for a stochastic multiplier design is equal to

2^{N}

Section: Latency and Variable Precisionmentioning

confidence: 99%

“…The result at

2^{N}

th clock cycle is the exact full‐precision product. Latency is reduced by order of

2^{N}

compared to all the sequences and techniques suggested in [4].…”

Section: Latency and Variable Precisionmentioning

confidence: 99%

“…Comparison with state‐of‐the‐art in [4] is done by simulating the proposed design with all the possible input pairs for

N = 8

. The proposed design is implemented with three types of SNGs, namely counter, linear feedback shift register (LFSR), and Sobol sequence.…”

Section: Latency and Variable Precisionmentioning

confidence: 99%

MAE ()(%) = \frac{\sum_{i = 0}^{255} \sum_{j = 0}^{255} (||, P_{normalfull} - P_{normalpartial})}{2^{16}} \times 100

where

P_{normalfull}

is the full‐precision 16‐bit product of i and j acting as reference, whereas

P_{normalpartial}

is the obtained product with lower precision which is scaled to 16 bits.…”

Section: Latency and Variable Precisionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Fast and exact multiple‐input unary‐to‐binary multiplier with variable precision for stochastic computing

Chuah

Kumar

2020

Electron. lett.

View full text Add to dashboard Cite

show abstract

A Stochastic Logic-Based Fuzzy Logic Controller: First Experimental Results of a Novel Architecture

2021

View full text Add to dashboard Cite

In stochastic computing (SC) systems numbers are represented with mean values of random binary sequences. This paper introduces a novel fuzzy inference architecture, in which the computational mechanism is based on stochastic logic (SL). First, the basic concept of SL is described, then the architecture of the SL-based fuzzy logic controller (SFLC) is built up systematically using the derived stochastic elements. The second part of the paper demonstrates the application of the proposed techniques, where the SFLC-based control performance is evaluated on a real mechatronic system. The results show that the SL-based approach provides effective and robust control performance, simple architecture and high noise tolerance. The proposed method is also benchmarked against conventional FLCs indicating that the robustness of the stochastic architecture allowed to outperform the benchmark controllers in noisy environments.INDEX TERMS stochastic logic, fuzzy logic controller, fuzzy hardware, self-balancing robot.

show abstract

Performing Stochastic Computation Deterministically

Cited by 2 publications

References 0 publications

Fast and exact multiple‐input unary‐to‐binary multiplier with variable precision for stochastic computing

Fast and exact multiple‐input unary‐to‐binary multiplier with variable precision for stochastic computing

A Stochastic Logic-Based Fuzzy Logic Controller: First Experimental Results of a Novel Architecture

Contact Info

Product

Resources

About