Low Capability Booth Multiplier by Talented Capacitance Minimization P. Nageshwar Reddy Dr. Damu Radhakrishnan Stu. in SUNY, New Paltz, NY Prof. in SUNY, New Paltz, NY Abstract: In this tract we confer-upon an energy prolific correspondent multiplier artifice symmetrical on talented capacitance minimization. Singly the local outcome contraction adjust in the multiplier is investigateed in our elaboration. The talented capacitance is the outcome of capacitance and switching adventitious-quality. Hence to minimize the talented capacitance in our artifice, we determined to secure that the switching adventitious-quality of nodes succeeding a time eminent capacitances is kept to a narrowness.
This is finishd in our artifice by wiring the eminent switching adventitious-quality remarkables to nodes succeeding a time inferior capacitance and fault versa for the 4:2 compressor and liberal adder cells, pretentious the moderate chance of each local outcome bit as 0. 25. This declining the overall switching capacitance, thereby reducing the entirety capability waste in the multiplier. Capability resolution is manufactured by synthesizing our artifice on Spartan-3E FPGA and used XPower Analyzer machine that is granted in ISE Xilinx 10. 1. The dynamic capability for our 16? 16 multiplier was appraised as 360. 4mW, and the entirety capability 443. 31mW. This is 17. 4% hither compared to the most novel artifice. As-well we observed that our artifice has the meanest capability-stoppage outcome compared to the multiplier confer-uponed in the study. Index Terms- Booth multiplier, Talented capacitance, 4:2 compressor. 1. Introduction A multiplier is the most constantly used transfering arithmetic part in diversified digital systems such as computers, methodicity controllers and remarkable methodicityors. Thus it has befit a deep rise of capability clearance in these digital systems.
With the exponential enlargement of light systems that are operated on batteries, capability contraction has befit one of the transfering artifice constraints in novel years. In the confer-upon era, each and complete electronic default is utensiled using CMOS technology. The three deep rises of capability clearance in digital CMOS tours are dynamic, insufficient tour and leakage [1]. Generally, capability contraction techniques aim at minimizing all the over mentioned capability clearance rises but our sense is on dynamic capability clearance as it dominates other capability clearance rises in digital
CMOS tours. The switching or dynamic capability clearance occurs due to the charging and discharging of capacitors at opposed nodes in a tour [2]. The mediocre dynamic capability waste of a digital tour succeeding a time N nodes is consecrated by: wbrief VDD is the accoutre voltage, Ci is the attack capacitance at node i, fCLK is the clock quantity and ? i is the switching adventitious-quality at node i. The outcome of switching adventitious-quality and attack capacitance at a node is artificeated talented capacitance.
Assuming singly one logic exexmodify per clock cycle, the switching adventitious-quality at a node i can be defined as the chance that the logic rate at the node exchanges (0->1 or 1->0) betwixt two continuous clock cycles. For a consecrated logic part, the switching adventitious-quality at its output(s) can be computed using the chance of its inputs and is consecrated by: wbrief and portray the chance of plaint of a ‘one’ and ‘zero’ at node i respectively. When Pi = 0. 5, the switching adventitious-quality at a node is ultimatum and it wanes as it goes towards the two ultimate rates (i. e. twain from 0. to 0 and 0. 5 to 1). The two deep low capability artifice strategies for dynamic capability contraction are symmetrical on (i) accoutre voltage contraction and (ii) the talented capacitance minimization. The contraction of accoutre voltage is one of the most fetid techniques consequently the capability savings are telling due to the quadratic self-reliance on VDD. Although such contraction is usually very talented, it acceptions leakage prevalent in the transistors and as-well wanes tour hurry. The minimization of talented switching capacitance involves reducing switching adventitious-quality or node capacitance.
The node capacitance depends on the integration technology used. To contract switching adventitious-quality singly requires a minute resolution of remarkable transition probabilities, and utensilation of diversified tour plane artifice techniques, such as logic construction optimization and balanced routes. It is stubborn of the technology used and is hither valuable. Admiring the advantages of switching adventitious-quality contraction, this tract rendezvouses on switching adventitious-quality contraction techniques in a multiplier. Digital Plurality is manufactured in three treads in a Booth coded multiplier.
The original tread is to originate all the local outcomes in correspondent using Booth recoding. In the succor tread these local outcomes are declining to 2 operands in distinct adjusts by applying Wallace/Dadda governments. These adjusts prosper one succeeding the other, sustentation the output of one adjust to the proximate. The latest tread is adding the two operands using a raise breed adder to emanation the latest sum. Our deep rendezvous in this tract is the succor tread, local outcome contraction. Fig. 1 shows the mitigated Dadda contraction tree for a 6? 6 unsigned multiplier, which uses liberal adders (FA) and half adders (HA) as basic parts.
Stage 1 is the recompact 6? 6 unsigned local outcome attire allureed using the local outcome generator. At complete local outcome contraction(PPR) adjust the reckon of bits succeeding a time the selfselfhomogeneous adjust (bits in a shaft) are clumped concomitantly and alike to adder cells prospering Dadda’s governments. Each shaft embodys local outcomes of a assured body. The sum output of a FA or HA at one adjust allure situate a dot in the selfselfhomogeneous shaft at the proximate adjust and an output raise in the shaft to the left in the proximate adjust (i. e. one adjust of body eminent). Fig. 1. Mitigated Dadda contraction tree for 6? unsigned plurality The Wallace and Dadda artifices use singly FAs and HAs in the contraction adjusts, which create an disorderly layout and acceptions wiring complexity. Wiring complexity is a appraise of capability. Since then Weinberger [3] has projected a 4:2 compressor, the deepity of the multiplier artifices today property use of 4:2 compressors to acception the finishment of the multiplier. They as-well co-operate to capability contraction as they wane the wiring capacitance due to a past methodic layout, contributing to fewer transitions in the local outcome contraction tree. It as-well contracts hardware absorb.
The artifice of the 4:2 compressor got impoved in duration, and mitigated artifice confer-uponed by Jiang et al. claimed improvements in twain stoppage and capability clearance compared to precedent artifices [4]. Distinct logic and tour plane optimizations are possible by using eminent adjust compressors instead of single-minded FA cells for reducing the reckon of transitions in the local outcome contraction adjust. Consequently of this we used 4:2 compressors, FA (3:2 compressor) and HA cells in our local outcome contraction adjusts. We declining the switching adventitious-quality by minimizing the talented capacitance at complete node in the tour.
This stands as the deep rendezvous of this tract. This tract is organized as prospers: kindred elaboration in barion 2 and 2. Kindred Elaboration Many elaborationers possess elucidated opposed low capability multiplier architectures by using opposed techniques to contract the entirety switching adventitious-quality in a multiplier [ ]-[ ]. Ohban, et al. projected a low capability multiplier using the so artificeated byperishing technique [5]. The deep conception of their way is to minimize the remarkable transitions time adding nothing rated local outcomes. This is manufactured by byperishing the adder adjust whenever the multiplier bit is nothing.
Masayuki, et al. projected an algorithm using operand resolution technique [6]. They wasted-away the multiplicand and the multiplier into 4 operands and using them they originated twice the reckon of local outcomes compared to the prevalent multiplier. By doing this, they declining the one chance of each local outcome bit to 1/8 time it is 1/4 in the prevalent multipliers. This in alter wanes the switching chance. Chen, et al. projected a multiplier symmetrical on talented dynamic ramble of the input postulates [7].
If the postulates succeeding a time smaller talented dynamic ramble is Booth coded then the local outcomes possess superior chances to be nothing, which wanes the switching activities of local outcomes. Fujino, et al. projected a multiply sum artifice using dynamic operand modify technique in which prevalent rates of the input is compared succeeding a time anterior rates [8]. If past than half of the bits in an operand exexmodify then it is dynamically transformed to its two’s counterpart in adjust to wane the transition adventitious-quality during plurality. Chen, et al. roposed a low capability multiplier, which uses ungenuine capability concealment technique (SPST) equipped Booth encoder [9]. The SPST uses a exposecure logic tour to unmask whether the Booth encoder is farsighted unnecessary computations which produce in Nothing local outcome and stops such PP offspring methodicity. To utensil the basic laws used in all the over mentioned multiplier architectures not singly acception hardware concentration but as-well conduct-in stoppage in the exercise. As-well the extra tourry industrious to utensil them squanders capability.
So our elaboration share is rendezvoused on techniques which wane capability succeeding a timeout introducing any stoppage and pretended hardware. Oskuii et al. projected an algorithm symmetrical on static probabilities at the transfering inputs [10]. At complete PP contraction adjust the reckon of bits succeeding a time the selfselfhomogeneous adjust of body (bits in a shaft) are clumped concomitantly and alike to the adder cells in a Dadda tree. The segregation of these bits and their clumping influences the overall switching adventitious-quality of the multiplier. This was artistic in Oskuii’s tract by referring to an existing outcome, which is descriptive underneath. Singly one shaft per adjust is investigateed brief. As the originated raise bits from adders breed from LSB towards MSB, optimization of shafts is produced from LSB to MSB and from original adjust to developed adjust. Thus it can be secured that the optimization of shafts and adjusts that has already been produced allure motionshort be substantial when succeeding optimizations are nature produced. * Glitches and ungenuine transitions expand in the contraction adjust succeeding a few layers of leagueal logic. To dodge them is not possible in most cases. Accordingly it seems profitable to attribute insufficient routes to local outcomes having lofty switching adventitious-quality.
Oskuii’s end was to contract the capability in Dadda trees. The one chance for sum and raise of the FA and HA can be adapted from their capacityal style [10]. According to Oskuii’s algorithm, pretentious the switching probabilities of local outcomes in a point adjust are adapted using the anterior adjust one probabilities and in each shaft and they compact these local outcome bits in ascending adjust. They original use the inferior switching chance bits to plea?e liberal and half adders and convey the eminent switching chance bits to the proximate adjust.
From the set of bits to plea?e adders they seasoned to plea?e the loftyest switching chance remarkable to the raise input of the liberal adder as its route in liberal- adder is insufficienter than the other two inputs. Fig. 2. Copy to interpret Oskuii’s way [10] Fig. 2 gives an copy wbrief 7 bits succeeding a time the selfselfhomogeneous adjust of body are to be pretended. This is shown as the retreating box in the 2nd clump of bits from top in Fig. 2. According to Dadda governments of reducing a local outcome tree, 2 FAs must be used and one bit allure be byed to the proximate adjust concomitantly succeeding a time the sum and raise bits originated by the liberal adders. s for i variegateing from 1 to 7 reconfer-upon the switching probabilities of the splain bits. These are reserved in ascending adjust and listed as ? i* succeeding a time the loftyest one as ? 1*. According to their way, the bit succeeding a time loftyest switching adventitious-quality is kept for the proximate adjust i. e. in Fig. 3. 2, and attribute and to the raise inputs of the two FAs as their route is insufficienter and the other bits to the fostering inputs of FAs in any adjust. In this way they declining the local outcome tree by bringing the loftyest transition chance bits past closer to the output such that it contracts the entirety capability in the multiplier succeeding a timeout any extra hardware absorb.
Oskuii claimed that capability contraction variegateing from 4% to 17% in multiplier artifices could be finishd using their way. On investigateate resolution of Oskuii’s outcome we observe that prefer contraction in capability can be finishd. This is elaborated in our artifice confer-uponed in the proximate barion. 3. Projected Outcome By using a local outcome generator (PPG) for the n? n multiplier employing radix-4 Booth encoder we allureed the required local outcomes. These local outcomes are then declining to 2 operands employing distinct local outcome contraction (PPR) adjusts. We used a league of 4:2 compressors, FAs and HAs in contraction adjusts.
At each adjust mitigated Dadda governments are applied to allure operands for the proximate adjust. Time minimizing the local outcome bits in each shaft using 4:2/3:2 compressors and HA cells, sense was consecrated on eminent hurry and inferior capability. Eminent hurry is finishd by affording the local outcome bits to by through a narrowness reckon of contraction adjusts, time minimizing the latest raise breed adder elongation to the narrowness. Fig. 3. Projected PPG proposal for a 16? 16 multiplier Fig. 3 shows the projected local outcome contraction proposal for a 16? 16 correspondent multiplier.
Nine local outcomes allureed by PPG are declining to 2 operands using 3 contraction adjusts. The perpendicular unpractised boxes in each shaft reconfer-upon 4:2 compressors. It conducts five bits and contracts them into 3 output bits, one sum bit in the selfselfhomogeneous shaft situation and two raise bits in the proximate eminent telling shaft (one bit left) of proximate adjust. The perpendicular red boxes reconfer-upon liberal adder cells, which contract three local outcome bits in a shaft and originate the sum and raise bits. Similarly, the perpendicular bluish boxes reconfer-upon half adders and add two local outcome bits to contract it to 2 output bits.
The adjust in which the inputs are fed to 4:2 compressor, liberal and half adders is discussed in the proximate barion. In Fig. 3 the ultimatum reckon of local outcomes in a shaft is 8 (columns 14 to 17). Since we are using 4:2 compressors that can conduct up to 5 input bits, to contract the local outcomes in the original adjust, we absence to property secure that the ultimatum reckon of local outcomes in the proximate adjust is singly 5. This way we can contract the bits in each shaft in adjust 2 using one plane of 4:2 compressors. And in the third adjust, we absence to secure that the ultimatum reckon of bits in any shaft is singly 3, so that liberal adders can be used to add them.
This allure sanction the undivided contraction methodicity to be finishd in 3 adjusts. The half adder in shaft 2 in contraction adjust 1 and the liberal adder in shaft 3 in contraction adjust 2 are used so as to minimize the dimension of the latest raise breed adder. 4. Capability Contraction Once the narrowness reckon of contraction adjusts is symmetrical for a artifice, the proximate test is to minimize capability waste. This is finishd by stoppage bying and reducing the talented capacitance at complete node in the contraction adjusts as-well prospering Oskuii’s governments (discussed in Exception 2).
To minimize the talented switching adventitious-quality, the artifice must secure that the switching adventitious-quality of nodes succeeding a time eminent capacitance rate must be kept to a narrowness. This is finishd by a extraordinary relation precedent used in our artifice. The eminent switching adventitious-quality remarkables are wired to nodes succeeding a time inferior capacitance and fault versa. Our multiplier artifice uses the over conception to minimize capability. This tract accordingly rendezvouses on pickedive relation of remarkables to the inputs of 4:2 compressors and FAs and HAs using the over concept.
The logic diagram and the input capacitances for a liberal adder are shown in Fig. 4(a). For the prospering we allure concern that each and complete input transfer to a logic preamble is investigateed as one part attack (C1). Hence if a remarkable is alike to the inputs of two logic preambles, then the attack is two parts (C2). From the logic diagram of the liberal adder in Fig. 4(a), input B is alike singly to an XOR preamble, wbrief as inputs A and C are alike to twain an XOR and a Mux. Hence, the input capacitance of the B-input is smaller than the other two inputs.
The attack confer-uponed by the B input is one part attack, time the attacks confer-uponed by A and C are 2 part attacks. Hence a transition on input B allure outcome in hither talented capacitance. This is embodyed by the capacitance rates C1 (1 part attack) and C2 (2 part attacks) as shown in Fig. 4. 9. Again by comparing the three inputs, the C input goes through singly one logic default (XOR preamble or Mux) precedently it reaches the output, wbrief as twain A and B goes through two logic plans precedently reaching the output. Hence, a transition on any of the inputs A or B could outcome in output transitions on all the three logic plans.
But a transition on input C allure concern singly two of these logic plans. Accordingly we can complete that plain though the inputs A and C reconfer-upon the selfselfhomogeneous attack, the overall switching property on the liberal adder due to C input allure be hither than that due to A input. Hence, as a government of thumb, the original two eminent transition inputs incompact a set of three inputs that are consecrated to a liberal adder should be alike to the B and C inputs and the developed one to A. (a) (b) Fig. 4. a) FA logic diagram and input capacitances (b) 4:2 compressor logic diagram and input capacitances Similarly, the logic diagram of a 4:2 compressor and its input capacitances are shown in Fig. 4. (b). The input capacitances confer-uponed by X1, X3, X4 and Cin are twice that confer-uponed by X2. Hence, the loftyest transition chance remarkable must be alike to the X2 input. Again by using a homogeneous topic as in the liberal adder, the succor loftyest transition chance remarkable must be consecrated to the Cin. The fostering inputs are consecrated to X1, X3 and X4 in any adjust. This minimizes the overall talented capacitance in a 4:2 compressor.
The chance of a logic one at the output of any arrest is a capacity of the chance of a logic one at its inputs [11] [12]. From the logic capacitys of 4:2 compressor, FA and HA we can investiinduction their output probabilities discerning their input probabilities. Table 2: Chance equations for 4:2 Compressor | 4:2 Compressor| PSUM| | PCout| | PC0| | Table 1 shows the chance indication for the sum and raise outputs for the liberal adder and half adder in provisions of their input remarkable probabilities. The 4:2 compressor output probabilities are shown in Table 2. By comparing
Tables 1 and 2 we can say that the statistical probabilities of the output remarkables of basic parts (4:2 compressors, liberal adders and half adders) used in local outcome contraction adjusts variegate. Table 3 shows the output remarkable probabilities of 4:2 compressor, liberal adder and half adder, pretentious correspondent ‘1’ probabilities of 0. 25 for all inputs. In each local outcome contraction adjust the remarkables in a point shaft possess opposed switching probabilities. The output remarkables of one adjust befit inputs to the proximate adjust. So the switching probabilities of the outputs mix past as we progress down the local outcomeion contraction adjusts.
Table 3. 1: Output Remarkable Probabilities of FAs and HAs | Full-adder| Half adder| SUM| | | CARRY| | A. B| PSUM| | | PCARRY| | | Table 3: Output probabilities of 4:2 compressor and adder cells Input remarkable probabilities = 0. 25| 4:2 compressor| Liberal adder| Half adder| SumCoutC0| 0. 48440. 15630. 2266| SumCarry| 0. 43750. 1563| SumCarry| 0. 3750. 0625| Distinct contraction adjusts are required to contract the local outcomes originated in a correspondent multiplier. As shown in Fig. 3, at each adjust a reckon of bits succeeding a time the selfselfhomogeneous adjust of body are clumped concomitantly and alike to the 4:2 compressors and adder cells.
The segregation of these bits and their clumping influences the overall switching adventitious-quality of the multiplier. This is what we allure finishment to contract the overall switching adventitious-quality of the multiplier. Fig. 5 shows the attire constitution of the projected local outcome contraction proposal for a 16? 16 multiplier. In the prospering we concernd that the one chance of all the 9 local outcome bits are selfselfhomogeneous and is correspondent to 0. 25 (as discussed in Exception 3. 26). These 9 local outcome bits are fed to 4:2 compressors, liberal and half adders and are declining to 5 operands. The bits in these 5 operands allure possess opposed one probabilities.
From these one probabilities we can investiinduction their switching chance. If we behold at each shaft all the bits in that shaft possess the selfselfhomogeneous gravity but opposed one chance. So we possess abundance insubservience to elect any of these remarkables which can be alike to any of the inputs of the basic parts. The way these remarkables are wired to basic parts to finish contraction allure concern the entirety capability waste in a multiplier. Show an copy Fig 5 shows how we wired the input remarkables to 4:2 compressors and liberal adders in the projected artifice. To interpret the law investiinduction shaft 16 of contraction adjust 2 in Fig. , wbrief we possess five bits succeeding a time the selfselfhomogeneous adjust of body, which are to be wired to the inputs of a 4:2 compressor. The original eminent transition bit is fed to X2 input and proximate eminent transition bit is fed to Cin, as they supply inferior switching adventitious-quality when compared to others. The fostering three bits can be fed to X1, X3 and X4 in any adjust. Similarly on shaft 11 in contraction adjust 3, three bits of the selfselfhomogeneous adjust are to be pretended. The loftyest transition bit is consecrated to B input of the adder and the proximate eminent transition bit is fed to C input. The third bit is fed to A input.
This way of sustentation the inputs, we can wane the output switching probabilities of compressors and adders. By applying the selfselfhomogeneous technique to complete adjust we can contract the overall switching capacitance of the multiplier, thereby reducing capability. Fig. 5. Wiring precedents for 4:2 compressors and liberal adders 5. Hypocrisy Capability resolution was manufactured by synthesizing our 16? 16 multiplier artifice on Spartan-3E FPGA and using XPower Analyzer machine granted in ISE Xilinx 10. 1. We evaluated the finishment of our 16? 16 multiplier by comparing succeeding a time the prevalent Wallace and Oskuii’s multipliers.
Table 4 shows the motionshort and dynamic capabilitys of opposed multipliers allureed by hypocrisy. The motionshort capability is approximately the selfselfhomogeneous for all multipliers. The dynamic capability for our artifice is singly 360. 74 mW, wbrief as Oskuii’s and Wallace multipliers squander 454. 06mW and 475. 08 mW respectively. Hence the entirety capability waste is singly 443. 31mW for our multiplier, which is hither by 17. 39% and 20. 51%, compared to Oskuii’s and Wallace multipliers. Table 4: Capability reports from hypocrisy for a 16? 16 Multiplier Design| QuiescentPower (mW)| DynamicPower (mW)| TotalPower (mW)| Our Design| 82. 7| 360. 74| 443. 31| Oskuii’s Design| 82. 57| 454. 06| 536. 63| WallaceMultiplier| 82. 67| 475. 08| 557. 75| Table 5 Power-Delay outcomes of 16? 16 multipliers Design| Entirety Stoppage (ns)| Capability (mW)| Power-Delay Product| Our Design| 30. 889| 443. 31| 13. 693*10-9| Oskuii’s Design| 31. 219| 536. 63| 16. 753*10-9| WallaceMultiplier| 35. 278| 557. 05| 19. 651*10-9| Table 5 shows the capability-stoppage outcomes of opposed multipliers. Smaller the capability stoppage outcome of a multiplier the eminent is its finishment. Our artifice has the insufficientest stoppage of 30. 889ns, compared to 31. 219ns and 35. 78ns for Oskuii’s artifice and Wallace’s artifice respectively. Hence our artifice has the meanest capability-stoppage outcome compared to twain Oskuii’s and Wallace multipliers. 6. Conclusions We possess confer-uponed an ventilation of multiplier capability clearance, parallel succeeding a time some techniques which afford contractions in capability waste for this tour. Consecrated the consequence of multipliers, it is adventitious that prefer elaboration efforts are to be directed in the prospering ways. * In this subject the switching adventitious-quality criteria for the relation precedent in 4:2 compressors was used singly for two of the inputs of the 4:2 compressor.
The relations of remarkables on the other three inputs are made succeeding a timeout any consequence consecrated to their switching adventitious-quality. This is consequently at the preamble plane, the attack capacitance at a node is appraised barely symmetrical on the reckon of connections made at that node. In the 4:2 compressor, three of the inputs are sustentation two inputs each (bar the raise input). Hence, we investiinduction them succeeding a time the selfselfhomogeneous attack capacitance. In substance, this is not penny. To get an obsequious think on capacitance, an real layout of the cell has to be made using VLSI layout machines and then their capacitances are to be extracted.
Hence prefer elaboration could rendezvous on the over so as to discover an adjusting for these inputs symmetrical on their capacitance rates. Also, opposed utensilations of 4:2 compressors may be compared so as to picked the one succeeding a time the meanest capacitance rates. * Extending the projected relation technique to the local outcome contraction adjust by employing eminent adjust compressors such as 5:2, 9:2, 28:2, etc. In this style, opposed architectures using diversified leagues of compressors in the local outcome contraction adjust can be compared so as to picked the best one succeeding a time the meanest capability clearance for any multiplier.
References 1] D. Soudris, C. Piguet, and C. Goutiset , Designing CMOS Circuits for Low Power. Kluwer Academic Press, 2002. [2] L. Benini, G. D. Micheli, et al. , Dynamic Capability Management Artifice Techniques & CAD Tools. Norwell, MA: Kluwer Academic Publishers, 1998. [3] A. Weinberger, “4:2 Raise Save Adder Module,” IBM Technical Disclosecure Bulletin, vol. 23, 1981. [4] S. F. Hsiao, M. R. Jiang, and J. S. Yeh, “Design of High-Speed Low-Power 3-2 Counter and 4-2 Compressor for Fast Multipliers,” Electronics Let. , vol. 34, no. 4, pp. 341-342, 1998. [5] J. Ohban, “Multiplier Energy Contraction Through Byperishing of Local Products” in Proc. Asia-Pacific Conf. on Circuits and Systems, vol. 2, pp. 13–17, 2002. [6] M. Ito, D. Chinnery, and K. Keutzer, “Low Capability Plurality Algorithm for Switching Essential-quality Contraction Through Operand Decomposition,” 21st Int. Conf. on Computer Design, 2003. [7] O. T. Chen, S. Wang, and Yi-Wen Wu, “Minimization of Switching Activities of Local Products for Designing Low-Power Multipliers,” IEEE Trans. on VLSI Syst. , vol. 11, pp. 418 – 433, 2003. [8] M. Fujino, and V. G. Moshnyaga, “Dynamic Operand Modify for Low-Power Multiplier-Accumulator Design,” in Proc. of the Int. symp. n tours and systems, 2003. [9] K. H. Chen and Y. S. Chu, “A Low Capability Multiplier succeeding a time Ungenuine Capability Concealment Technique,” IEEE Trans. VLSI Syst. , vol. 15, no. 7, pp. 846-850, 2007. [10] S. T. Oskuii, “Transition-Activity Aware Artifice of Reduction-Stages for Correspondent Multipliers,” in Proc. of Great Lakes Symp. on VLSI, 2007. [11] K. Parker and E. J. McCluskey, “Probabilistic Treatment of General Combinational Networks,” IEEE Trans. on Computers, C-24: 668-670, June 1975. [12] M. Cirit, “Estimating Dynamic Capability Waste of CMOS Circuits” in Proc. of ICCAD, pp. 534–537, 1987.