Pith. sign in

REVIEW 1 major objections 2 minor 50 references

A twit-based multiplier enables efficient generic modular multiplication for RNS moduli of the form 2^n ± δ by deferring carry propagation to the final stage.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · grok-4.3

2026-06-27 14:29 UTC pith:RFTDFHOT

load-bearing objection The paper gives a working twit-based RNS multiplier with synthesis numbers, but the claim that carry propagation stays deferred for every delta is not isolated in the results. the 1 major comments →

arxiv 2606.09965 v1 pith:RFTDFHOT submitted 2026-06-08 cs.AR

A Generic Modulo-(2^npmδ) RNS Multiplier Based on Twit Representation

Saeid Gorgin , Amirhossein Sadr , Behzad Salami , Dara Rahmati This is my paper
classification cs.AR
keywords residue number systemmodular multipliertwit representationRNS arithmetichardware implementation45nm synthesismodulo 2^n deltacryptography
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper develops a modular multiplier for residue number systems that works with the twit representation for moduli 2^n plus or minus delta. The design splits operands, generates modular partial products, accumulates them with carry-save adders, folds overflows, and finishes with a compatible addition. By keeping carries short until the end, it cuts the critical path compared to multiply-then-reduce methods. Synthesis results in 45 nm technology confirm average improvements of about 20 percent in speed, 13 percent in area, and 28 percent in power over baselines. The approach supports various channel widths and provides wide dynamic range for applications in cryptography and signal processing.

Core claim

The proposed architecture computes the product of two residues through operand splitting, modular partial-product generation, carry-save accumulation, overflow folding, and a twit-compatible final modular addition. This organization avoids the long critical paths of conventional designs and is compatible with the twit representation for the full range of admissible deltas.

What carries the argument

Twit representation of residues combined with overflow folding after carry-save accumulation in the multiplication pipeline.

Load-bearing premise

The twit representation remains compatible with the new partial-product generation, carry-save accumulation, and overflow-folding steps across the entire admissible delta range without introducing hidden long carry chains or requiring post-synthesis fixes.

What would settle it

Synthesis results for a specific delta in the admissible range showing no reduction or an increase in delay, area, or power would falsify the efficiency claim.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • The multiplier integrates directly with prior twit-based adders and subtractors for complete RNS arithmetic.
  • It supports 5-bit channels that together give sufficient dynamic range due to the wide delta range.
  • Performance gains hold for 8-bit and 11-bit channel widths as well.
  • End-to-end latency decreases in RNS-based workloads involving many multiplications and additions.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Similar techniques might apply to other arithmetic operations in RNS beyond add, subtract, and multiply.
  • Adoption could lower power in machine-learning accelerators that use RNS for efficiency.
  • Further work could explore automated generation of such multipliers for arbitrary delta values.
  • Comparison against other generic RNS multipliers not based on twit would quantify the specific benefit.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

1 major / 2 minor

Summary. The paper proposes a generic twit-based modular multiplier for RNS channels with moduli of form 2^n ± δ (0 ≤ δ ≤ 2^{n-1}-1). The architecture performs operand splitting, modular partial-product generation, carry-save accumulation, overflow folding, and a final twit-compatible modular addition, with the goal of deferring all carry propagation to the last stage. Synthesis in FreePDK 45 nm for 5-/8-/11-bit channels reports average reductions of 20.5% delay, 13.2% area, and 28.0% power versus baselines, plus system-level latency gains over mixed multiply-add workloads.

Significance. If the constant-depth claim holds across the full δ range, the work supplies a missing multiplier primitive that pairs with prior twit-based addition/subtraction, enabling flexible RNS datapaths without modulus-specific redesign. The concrete 45 nm synthesis numbers and workload study constitute reproducible, falsifiable evidence of the gains; this is a practical strength for RNS applications in cryptography and accelerators.

major comments (1)
  1. [§3] §3 (Architecture, overflow-folding paragraph): the assertion that folding maps the CSA sum back into twit form without introducing δ-dependent carry chains is load-bearing for the constant-critical-path claim, yet the description provides neither an explicit bound on carry-propagation length inside the folding logic nor a separate synthesis isolation of folding delay versus δ; the 5-/8-/11-bit results therefore rest on an unverified assumption.
minor comments (2)
  1. [Abstract] Abstract and §5: the baseline architectures, exact δ values chosen for each channel width, and any error-bar information on the reported percentage reductions are not stated; adding these details would allow readers to reproduce the comparison.
  2. [§5] §5 synthesis tables: column headings for the compared designs and the precise modulus set (including chosen δ) are needed for clarity.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. The major comment highlights a point where additional clarification and evidence can strengthen the constant-critical-path claim. We address it point-by-point below and commit to revisions that directly respond to the concern.

read point-by-point responses

  1. Referee: [§3] §3 (Architecture, overflow-folding paragraph): the assertion that folding maps the CSA sum back into twit form without introducing δ-dependent carry chains is load-bearing for the constant-critical-path claim, yet the description provides neither an explicit bound on carry-propagation length inside the folding logic nor a separate synthesis isolation of folding delay versus δ; the 5-/8-/11-bit results therefore rest on an unverified assumption.

    Authors: We acknowledge that the current manuscript does not supply an explicit analytical bound on carry length within the overflow-folding logic or an isolated synthesis sweep versus δ. The overall 45 nm results for the complete multiplier already incorporate the folding stage for representative δ values within each channel width, and the reported delay reductions are consistent with the claim. However, to make the independence from δ fully verifiable, we will revise §3 to include (1) a short proof that the folding operation produces at most a two-bit carry chain independent of δ (arising from the twit digit bounds and the specific overflow-correction constants), and (2) a supplementary table isolating post-synthesis delay of the folding block alone across the full admissible δ range for the 8-bit channel. These additions will be placed in the revised version. revision: yes

Circularity Check

0 steps flagged

Minor self-citation on twit representation; central claims rest on synthesis results

full rationale

The paper presents an architectural construction (operand splitting, modular PP generation, CSA accumulation, overflow folding, final twit addition) whose performance numbers are obtained from FreePDK 45 nm synthesis rather than from any equation or derivation that reduces to a fitted constant or prior result by construction. Reliance on the twit representation is explicitly cited from prior work and does not create a load-bearing circular chain inside the present derivation or claims. No self-definitional, fitted-prediction, or uniqueness-imported steps are present.

Axiom & Free-Parameter Ledger

2 free parameters · 1 axioms · 0 invented entities

The design rests on the twit representation and its addition properties from prior work; channel widths and delta ranges are chosen for evaluation rather than derived.

free parameters (2)
  • residue channel width
    5-bit, 8-bit, and 11-bit widths selected for concrete evaluation and synthesis
  • delta range
    Admissible 0 ≤ δ ≤ 2^{n-1}-1 taken as given from prior twit work
axioms (1)
  • domain assumption Twit representation permits efficient modular addition and subtraction for all admissible δ
    Invoked as the foundation that the new multiplier must remain compatible with

pith-pipeline@v0.9.1-grok · 5873 in / 1443 out tokens · 21488 ms · 2026-06-27T14:29:58.827094+00:00 · methodology

0 comments
read the original abstract

Modular multiplication is a fundamental arithmetic primitive in Residue Number Systems (RNS) and is often the dominant source of delay, area, and energy consumption in RNS datapaths used in cryptography, signal processing, and machine-learning accelerators. Recent work introduced a twit-based residue representation for moduli of the form $2^n \pm \delta$, with $0 \le \delta \le 2^{n-1}-1$, and showed that it enables efficient generic modular addition and subtraction across the full admissible $\delta$ range. However, an efficient modular multiplier compatible with the same representation has remained unavailable. This paper presents a generic twit-based modulo-$(2^n \pm \delta)$ multiplier for RNS channels. The proposed architecture computes the product through operand splitting, modular partial-product generation, carry-save accumulation, overflow folding, and a twit-compatible final modular addition. By deferring carry propagation to the final stage, the resulting organization avoids the long critical paths characteristic of conventional multiply-then-reduce designs. To demonstrate the effectiveness of the proposed approach, we study a modulus set with 5-bit residue channels and show that, owing to the broad admissible range of $\delta$, it can provide a sufficiently wide dynamic range. Moreover, additional 8-bit and 11-bit configurations are used to evaluate the proposed approach at larger channel widths. We implement and synthesize the proposed multiplier in a FreePDK 45\,nm flow, and the results show average reductions of 20.5\% in delay, 13.2\% in area, and 28.0\% in power relative to baseline designs. A system-level study further indicates that these circuit-level improvements translate into lower end-to-end latency over a broad range of modular multiplication and addition workloads.

Figures

Figures reproduced from arXiv: 2606.09965 by Amirhossein Sadr, Behzad Salami, Dara Rahmati, Saeid Gorgin.

Figure 1
Figure 1. Figure 1: Abstract views of generic modular multipliers: (a) [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: The overall structure of the proposed multiplier [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Overall architecture of the proposed modulo- [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: As highlighted in Table I, unlike prior designs that rely [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 4
Figure 4. Figure 4: Analytical delay (∆G) and hardware-cost (#G) trends of the compared generic modulo-(2 n ±δ) multipliers for (3 ≤ n ≤ 16). C. Circuit-Level Synthesis Results While the analytical evaluation is useful for exposing structural trends, it does not capture technology-dependent effects such as logic mapping, fan-out, interconnect overhead, and gate sizing. To obtain a more realistic comparison, all architectures … view at source ↗
Figure 5
Figure 5. Figure 5: Synthesis comparison for the representative [PITH_FULL_IMAGE:figures/full_fig_p009_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Area and power of the moduli supported by all compared architectures under various timing constraints. [PITH_FULL_IMAGE:figures/full_fig_p010_6.png] view at source ↗
Figure 8
Figure 8. Figure 8: Application-level delay evaluation as a function of the [PITH_FULL_IMAGE:figures/full_fig_p011_8.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

50 extracted references

  1. [1]

    Up to 8k-bit Modular Montgomery Multiplication in Residue Number Systems With Fast 16-bit Residue Channels,

    Z. Ahmadpour and G. Jaberipur, “Up to 8k-bit Modular Montgomery Multiplication in Residue Number Systems With Fast 16-bit Residue Channels,”IEEE Transactions on Computers, vol. 71, no. 6, pp. 1399– 1410, 2021

    2021

  2. [2]

    RNS-based FPGA accelerators for high-quality 3D medical image wavelet processing using scaled filter coefficients,

    N. N. Nagornov, P. A. Lyakhov, M. V . Valueva, and M. V . Bergerman, “RNS-based FPGA accelerators for high-quality 3D medical image wavelet processing using scaled filter coefficients,”IEEE Access, vol. 10, pp. 19 215–19 231, 2022

    2022

  3. [3]

    Application of the residue number system to reduce hardware costs of the convolutional neural network implementation,

    M. V . Valueva, N. Nagornov, P. A. Lyakhov, G. V . Valuev, and N. I. Chervyakov, “Application of the residue number system to reduce hardware costs of the convolutional neural network implementation,” Mathematics and computers in simulation, vol. 177, pp. 232–243, 2020. 12 TABLE III: Synthesis results of the compared generic modulo-(2 n ±δ)multipliers for...

    2020

  4. [4]

    1.19 1.072010.48 0.75 1272 0.75 1513.55 0.81

  5. [5]

    1.32 1.19 2964.10 1.11 2221 1.31 2938.61 1.57 Proposed1.11 1.002679.70 1.00 1689 1.00 1871.75 1.00 +3 259

  6. [6]

    1.70 1.40 9140.09 3.11 3374 1.61 5731.41 2.26

  7. [7]

    1.36 1.12 3435.28 1.17 2685 1.28 3646.77 1.44 Proposed1.21 1.00 2938.76 1.00 2094 1.00 2534.16 1.00 −9 247

    2094

  8. [8]

    1.31 1.16 2592.41 1.011668 0.982188.75 1.13

  9. [9]

    1.26 1.12 2774.03 1.08 1994 1.17 2515.63 1.30 Proposed1.13 1.00 2558.15 1.001709 1.001931.68 1.00 +9 265

    1994

  10. [10]

    1.59 1.22 8061.17 2.04 3198 1.15 5072.03 1.41

  11. [11]

    1.49 1.143882.52 0.983052 1.10 4549.31 1.26 Proposed1.30 1.003953.38 1.002769 1.00 3608.84 1.00 −127 129 [14] 1.20 1.192351.66 0.901574 1.00 1886.60 1.20 Proposed1.01 1.002622.45 1.001569 1.00 1577.79 1.00 +127 383 [14] 1.70 1.12 6750.88 1.52 2911 1.12 4951.03 1.26 Proposed1.52 1.00 4452.78 1.00 2594 1.00 3930.43 1.00 11 −3 2045

    2045

  12. [12]

    1.60 1.203329.68 0.77 2230 0.73 3567.33 0.87

  13. [13]

    1.70 1.27 5136.02 1.18 4036 1.32 6851.11 1.68 Proposed1.34 1.004339.15 1.00 3050 1.00 4078.15 1.00 +3 2051

    2051

  14. [14]

    2.21 1.57 54554.70 10.04 9283 2.18 20541.41 3.43

  15. [15]

    1.76 1.25 5989.68 1.10 5092 1.20 8969.05 1.50 Proposed1.41 1.00 5431.68 1.00 4250 1.00 5984.42 1.00 −9 2039

    2039

  16. [16]

    1.70 1.21 4291.75 1.022836 0.954821.20 1.15

  17. [17]

    1.72 1.22 4676.57 1.12 3733 1.25 6418.52 1.53 Proposed1.41 1.00 4192.26 1.002985 1.004199.60 1.00 +9 2057

    2057

  18. [18]

    2.35 1.56 64344.38 12.98 9338 2.47 21957.43 3.86

  19. [19]

    1.88 1.25 6302.23 1.27 5305 1.41 9965.44 1.75 Proposed1.51 1.00 4955.34 1.00 3775 1.00 5689.68 1.00 −1023 1025 [14] 2.08 1.19 13422.91 2.01 5360 1.33 11157.90 1.58 Proposed1.75 1.00 6687.65 1.00 4042 1.00 7081.58 1.00 +1023 3071 [14] 2.31 1.23 50857.13 7.14 7920 1.84 18273.02 2.27 Proposed1.87 1.00 7126.43 1.00 4300 1.00 8044.44 1.00

  20. [20]

    Res-dnn: A residue number system-based dnn accelerator unit,

    N. Samimi, M. Kamal, A. Afzali-Kusha, and M. Pedram, “Res-dnn: A residue number system-based dnn accelerator unit,”IEEE Transactions on Circuits and Systems I: regular papers, vol. 67, no. 2, pp. 658–671, 2019

    2019

  21. [21]

    Computer arithmetic: Algorithms and hardware designs,

    B. Parhami, “Computer arithmetic: Algorithms and hardware designs,” Oxford University Press, 2010

    2010

  22. [22]

    Efficient VLSI implementation of modulo(2 n ±1) addition and multiplication,

    R. Zimmermann, “Efficient VLSI implementation of modulo(2 n ±1) addition and multiplication,” inProceedings 14th IEEE Symposium on Computer Arithmetic (Cat. No.99CB36336), 1999, pp. 158–167

    1999

  23. [23]

    Efficient diminished-1 modulo 2 n +1 multipliers,

    C. Efstathiou, H. T. Vergos, G. Dimitrakopoulos, and D. Nikolos, “Efficient diminished-1 modulo 2 n +1 multipliers,”IEEE Transactions on Computers, vol. 54, no. 4, pp. 491–496, 2005

    2005

  24. [24]

    Design of efficient modulo 2 n +1 multipliers,

    H. T. Vergos and C. Efstathiou, “Design of efficient modulo 2 n +1 multipliers,”IET Computers & Digital Techniques, vol. 1, no. 1, pp. 49–57, 2007

    2007

  25. [25]

    On the Design of Modulo 2n +1 Multipliers,

    C. Efstathiou, K. Pekmestzi, and N. Axelos, “On the Design of Modulo 2n +1 Multipliers,” in14th Euromicro conference on digital system design. IEEE, 2011, pp. 453–459

    2011

  26. [26]

    Efficient modulo 2n +1 multiply and multiply-add units based on modified Booth encoding,

    C. Efstathiou, N. Moshopoulos, N. Axelos, and K. Pekmestzi, “Efficient modulo 2n +1 multiply and multiply-add units based on modified Booth encoding,”Integration, vol. 47, no. 1, pp. 140–147, 2014

    2014

  27. [27]

    Arithmetic units for RNS moduli{2 n −3}and{2 n +3}operations,

    P. M. Matutino, R. Chaves, and L. Sousa, “Arithmetic units for RNS moduli{2 n −3}and{2 n +3}operations,” in13th Euromicro Conference on Digital System Design. IEEE, 2010, pp. 243–246

    2010

  28. [28]

    Improved modulo-(2 n ±3)multipliers,

    H. Ahmadifar and G. Jaberipur, “Improved modulo-(2 n ±3)multipliers,” in17th CSI International Symposium on Computer Architecture & Digital Systems. IEEE, 2013, pp. 31–35

    2013

  29. [29]

    Modulo-(2 q − 3)Multiplication with Fully Modular Partial Product Generation and Reduction,

    G. Jaberipur, S. Gorgin, N. Ahamadian, and J.-A. Lee, “Modulo-(2 q − 3)Multiplication with Fully Modular Partial Product Generation and Reduction,” in30th Symposium on Computer Arithmetic. IEEE, 2023, pp. 68–75

    2023

  30. [30]

    New efficient structure for a modular multiplier for RNS,

    A. A. Hiasat, “New efficient structure for a modular multiplier for RNS,” IEEE Transactions on Computers, vol. 49, no. 2, pp. 170–174, 2000

    2000

  31. [31]

    RNS Arithmetic Units for Modulo 2 n ±k,

    P. M. Matutino, H. Pettenghi, R. Chaves, and L. Sousa, “RNS Arithmetic Units for Modulo 2 n ±k,” in15th Euromicro Conference on Digital System Design. IEEE, 2012, pp. 795–802

    2012

  32. [32]

    A generic modulo-(2 n ±δ) addition algorithm via two-valued digit encoding,

    S. Gorgin, A. Sadr, D. Rahmati, and J. Kim, “A generic modulo-(2 n ±δ) addition algorithm via two-valued digit encoding,” in32nd Symposium on Computer Arithmetic. IEEE, 2025, pp. 85–92

    2025

  33. [33]

    P. A. Mohan,Residue Number Systems: Theory and Applications, 1st ed. Birkh¨auser Basel, 2016

    2016

  34. [34]

    Modular multiplication and base extensions in residue number systems,

    J.-C. Bajard, L.-S. Didier, and P. Kornerup, “Modular multiplication and base extensions in residue number systems,” in15th Symposium on Computer Arithmetic. IEEE, 2001, pp. 59–65

    2001

  35. [35]

    Area-Power Efficient Modulo 2 n −1 and Modulo 2n +1 Multipliers for{2 n −1,2 n,2 n +1}Based RNS,

    R. Muralidharan and C.-H. Chang, “Area-Power Efficient Modulo 2 n −1 and Modulo 2n +1 Multipliers for{2 n −1,2 n,2 n +1}Based RNS,”IEEE Transactions on Circuits and Systems I: Regular Papers, vol. 59, no. 10, pp. 2263–2274, 2012

    2012

  36. [36]

    Novel approaches to the design of VLSI RNS multipliers,

    D. Radhakrishnan and Y . Yuan, “Novel approaches to the design of VLSI RNS multipliers,”IEEE Transactions on Circuits and Systems II Analog and Digital Signal Processing, vol. 39, no. 1, pp. 52–57, 1992

    1992

  37. [37]

    A universal architecture for designing efficient modulo 2n +1 multipliers,

    L. Sousa and R. Chaves, “A universal architecture for designing efficient modulo 2n +1 multipliers,”IEEE Transactions on Circuits and Systems I: Regular Papers, vol. 52, no. 6, pp. 1166–1178, 2005

    2005

  38. [38]

    Koren,Computer Arithmetic Algorithms, 2nd ed

    I. Koren,Computer Arithmetic Algorithms, 2nd ed. Natick, MA, USA: A K Peters, 2002

    2002

  39. [39]

    Multiplication of Multidigit Numbers on Automata,

    A. Karatsuba and Y . Ofman, “Multiplication of Multidigit Numbers on Automata,”Soviet Physics Doklady, vol. 7, p. 595, 12 1962

    1962

  40. [40]

    Weighted two-valued digit- set encodings: unifying efficient hardware representation schemes for redundant number systems,

    G. Jaberipur, B. Parhami, and M. Ghodsi, “Weighted two-valued digit- set encodings: unifying efficient hardware representation schemes for redundant number systems,”IEEE Transactions on Circuits and Systems I: Regular Papers, vol. 52, no. 7, pp. 1348–1357, 2005

    2005

  41. [41]

    Fast parallel-prefix modulo 2n +1 adders,

    C. Efstathiou, H. T. Vergos, and D. Nikolos, “Fast parallel-prefix modulo 2n +1 adders,”IEEE Transactions on Computers, vol. 53, no. 9, pp. 1211–1216, 2004

    2004

  42. [42]

    New memoryless, mod(2 n ±1)residue multiplier,

    A. Hiasat, “New memoryless, mod(2 n ±1)residue multiplier,”Elec- tronics Letters, vol. 28, pp. 314–315, 1992

    1992

  43. [43]

    Combinational logic approach for designing RNS multipliers,

    A. A. Hiasat and H. Abdel-Aty-Zohdy, “Combinational logic approach for designing RNS multipliers,” in39th Midwest Symposium on Circuits and Systems, vol. 1. IEEE, 1996, pp. 541–543

    1996

  44. [44]

    Novel Modulo 2 n +1 Multipliers,

    H. Vergos and C. Efstathiou, “Novel Modulo 2 n +1 Multipliers,” in 9th Euromicro Conference on Digital System Design. IEEE, 2006, pp. 491–496

    2006

  45. [45]

    Faster modulo 2 n +1 multipliers without booth recoding,

    R. Chaves and L. Sousa, “Faster modulo 2 n +1 multipliers without booth recoding,” inConf. Design of Circuits and Integrated Systems, 2005

    2005

  46. [46]

    (4+2logn)∆GParallel Prefix Modulo-(2 n −3)Adder via Double Representation of Residues in [0, 13 2],

    G. Jaberipur and S. H. F. Langroudi, “(4+2logn)∆GParallel Prefix Modulo-(2 n −3)Adder via Double Representation of Residues in [0, 13 2],”IEEE Transactions on Circuits and Systems II: Express Briefs, vol. 62, no. 6, pp. 583–587, 2015

    2015

  47. [47]

    High-Performance Multiplication Modulo 2 n–3,

    P.-M. Seidel, “High-Performance Multiplication Modulo 2 n–3,” in52nd Asilomar Conference on Signals, Systems, and Computers, 2018, pp. 130–134

    2018

  48. [48]

    A low-complexity com- binatorial RNS multiplier,

    V . Paliouras, K. Karagianni, and T. Stouraitis, “A low-complexity com- binatorial RNS multiplier,”IEEE Transactions on Circuits and Systems II: Analog and Digital Signal Processing, vol. 48, no. 7, pp. 675–683, 2001

    2001

  49. [49]

    RNS arithmetic multiplier for medium and large moduli,

    A. A. Hiasat, “RNS arithmetic multiplier for medium and large moduli,” IEEE Transactions on Circuits and Systems II: Analog and Digital Signal Processing, vol. 47, no. 9, pp. 937–940, 2002

    2002

  50. [50]

    Double-least-significant-bits 2’s-complement number rep- resentation scheme with bitwise complementation and symmetric range,

    B. Parhami, “Double-least-significant-bits 2’s-complement number rep- resentation scheme with bitwise complementation and symmetric range,” IET Circuits, Devices & Systems, vol. 2, pp. 179–186, 2008. Saeid Gorgin(Senior Member, IEEE) received the B.S. degree in Computer Engineering from Azad University, South Tehran Branch, Tehran, Iran, in 2001, the M.S....

    2008