Documente Academic
Documente Profesional
Documente Cultură
S0 = A0B0 (1)
Fig. 1. Multiplication of two decimal numbers
C1S1 = A0B1 + A1B0 (2)
C2S2 = C1 + A1B1 (3)
III. ARCHITECTURE AND IMPLEMENTATION OF VEDIC
MULTIPLIER B. 4 X 4 Vedic Multiplier Block
In the previous section Vedic multiplication technique for In this section 4 X 4 Vedic multiplier architecture is
decimal numbers is described. This section explains discussed. Consider two 4-bits numbers such as A and B
architecture of N by N bit multiplier. The Vedic multiplication where A = A3A2A1A0 and B = B3B2B1B0. The LSBs of two
technique can be used for multiplication of binary numbers. numbers (A0 X B0) are multiplied to generate LSB S0 of final
Implementation of 2 X 2 Vedic multiplier block is prime result. Same procedure is hunted here as 2 X 2 Vedic
important in the implementation of 4 X 4 and 8 X 8 Vedic multiplication procedure discussed earlier. Initially pre-carry
multiplier architecture. In this section implementation of 2 X is set to zero. In each and every step generated carry is
2, 4 X 4 and 8 X 8 Vedic multiplier architecture are explained. forwarded to next step and process goes on. At the end C6 and
S6 is obtained. Output lines C6S6S5S4S3S2S1S0 gives finally
A. 2 X 2 Vedic Multiplier Block generated result.
Here, actual Urdhva -Tiryagbhyam Sutra multiplication The 4 X 4 Vedic multiplication block diagram is shown in
technique is applied by to 2 binary numbers with 2 bits each. fig. 3. Here, 2 X 2 Vedic multipliers are used to implement 4
Considering two numbers are A and B where A = A1A0 and B X 4 Vedic multiplier to generate partial product. Three ripple
= B1B0. Least Significant Bit (LSB) of first number A that is carry adders of 4 bits each are used for addition of generated
A0 and LSB of second number B that B0 are multiplied with partial products. The carry output of first two ripple carry
each other (vertical). Generated product is saved as LSB of adders are ORed and output of this OR gate is given to next
final result. Thus, AND gate is used for multiplication of A0 ripple carry adder. Zero inputs are given to some of the ripple
and B0. Fig. 2, shows 2 X 2 multiplication process. Next step carry adders wherever required. The arrangement of ripple
is to multiply LSB of number A with MSB of number B that is carry adders is made in such way that computation time
(A0 X B1) and MSB of number A with LSB of number B that required for whole multiplication process is reduced and speed
is (A1 X B0). Thus, two AND gates are required for this of working is increased.
multiplication.
296
C. 8 X 8Vedic Multiplier Block
Here, implementation of 8 by 8 Vedic multiplier is
outlined. Consider two 8-bits binary numbers namely A and B
where A = A7A6A5A4A3A2A1A0 and B = B7B6B5B4B3B2B1B0.
8 X 8 Vedic multiplication process is mirror to 4 X 4 Vedic
multiplier expressed in the previous sub-section B. Pre-carry
at first step is set to zero. In every step generated carry is
shifted to next step for addition and process continues. At the
end C14 and S14 is obtained. Finally generated result is given in
terms of C14S14S13S12S11S10S9S8S7S6S5S4S3S2S1S0.
Implementation of 8 by 8 Vedic multiplier is clearly
understood from the block diagram as shown in fig. 4. Here,
four 4 X 4 Vedic multiplier blocks and three carry select
adders of 8 bits each are used. The arrangement of the carry
select adders is made in different way such that it requires less
computation time. Some of the carry select adders are given Fig. 5. BEC-Based CSLA
with zero inputs, wherever required. Output of middle
multipliers are added using first CSLA. Output of first CSLA According to the input carry Cin = ‘0’ or ‘1’, final output
and first Vedic multiplier are added using second CSLA. is selected. If Cin = ‘0’ then multiplexer will select output of
Carry outputs from first two CSLAs are ORed and given as RCA else it will select output of BEC. BEC takes the output of
input to the third CSLA to generate final result. RCA and gives output to multiplexer. As the name suggests,
BEC generates output by adding one to its input. Thus,
whenever Cin = ‘1’ multiplexer will select output of BEC.
Accordingly final sum and carry is generated. By using this
BEC-based CSLA, Vedic multiplier [1] is implemented.
B. MCSLA
B. K. Mohanty and S. K. Patel proposed modified CSLA
[3]. The block diagram of that carry select adder is shown in
fig. 6. This CSLA has one Half Sum Generation (HSG) unit,
one Final Sum Generation (FSG) unit, one Carry Generation
(CG) unit, and one Carry Selection (CS) unit. The CG unit
composed of two CGs (CG0 and CG1) corresponding to input-
carry ‘0’ and ‘1’ [3]. Two n bit operands (A and B) are given
to HSG unit that generates n bits S0 and n bits C0. These
generated S0 and C0 are given to CG0 as well as CG1 and n-bit
full carry words C01 and C11. CS unit selects one of the outputs
from full carry words C01 and C11. If Cin = ‘0’ then CS will
select C01, else it will select C11. By using this modified
CSLA, Vedic multiplier [8] is implemented.
Fig. 4. 8 X 8 Vedic multiplier architecture
297
V. PROPOSED DESIGN
BEC-based CSLA and MCSLA are briefly studied and
noticed the possibility of designing more improved CSLA.
Proposed CSLA consists of basically two units 1) sum and
carry generation unit 2) sum and carry selection unit.
Redundant operations present in previous CSLA designs are
identified and eliminated. This has lent a hand to reduce
number of gates required for the proposed design. So the aim
is to reduce number gates required for the design of CSLA.
Minimizing the number of gates of design involves two steps
• Generate all the prime implicants for the given logic
function f [9].
• Find the set of essential prime implicants. Fig.7. Single stage CSLA
Accordingly new area-efficient carry select adder is
designed. B. Multistage CSLA
By cascading single stage CSLAs, multistage CSLAs are
A. Single Stage CSLA formed. Here different bit widths of CSLA are designed such
as 8-16-32 bits. Proposed CSLA structure at gate level is
In proposed design some logic formulations are made.
shown in fig. 8 where fig. 8 (a) stands for gate level design of
Depending upon the sum and carry functions equal to ‘1’
carry generation unit. Gate level implementation of sum
prime implicants are noted down. Essential prime impicants
generation unit is shown in fig. 8 (b). Gate level design of sum
are combined so that final sum and final carry can be
and carry selection unit is shown in fig. 8 (c). For selection of
generated. Separate sum and carry selection units are used
sum and carry multiplexer is utilized. These are 4:2
after sum and carry generation units to eliminate the redundant
multiplexers obtained from two 2:1 multiplexers one for carry
operations. At first, single stage CSLA is implemented.
selection and other for sum selection.
Depending upon bit pattern of input carry ‘Cin’ final sum and
carry is selected. ‘If Cin = ’0’ then carry selection unit will Multistage CSLA of ‘n’ bit width is shown in fig. 9.
select output ‘p’ from carry generation unit else it will select Equations for sum generation carry generation as well as sum
‘q’. If Cin = ‘0’ then sum selection unit will select output ‘r’ and carry selection are given below.
from sum generation unit else it will select‘s’ as shown in the
fig. 7.
Fig.8. (a) Carry generation unit (b) Sum generation unit (c) Sum and carry selection unit
298
MCSLA based Vedic multiplier. Vedic multiplier is also
compared in terms of delay. Delay is maximum
combinational path delay obtained from the final report.
Proposed Vedic multiplier shows lesser delay than Vedic
multiplier using MCSLA but it has higher delay than BEC-
based Vedic multiplier as shown in fig. 11. Proposed Vedic
multiplier is also compared with Booth multiplier [4] in
terms of area and delay. Results obtained from these
implementations are shown in Table III. Vedic multiplier
shows better performance results than Booth multiplier in
terms of area as well as delay as shown in fig. 12 and fig. 13
respectively. Proposed Vedic multiplier is faster compared
to Booth multiplier.
299
VII. CONCLUSION
46 Vedic multiplier using carry select adder is implemented
45 here. At first conventional CSLAs like (BEC based CSLA
44 BEC Based Vedic
43 Multiplier and MCSLA) as well as proposed CSLA are implemented.
42 MCSLA Based According to the results obtained, it is concluded that
41 Vedic Multiplier proposed carry select adder requires less number of gates
40
Proposed Vedic
than both MCSLA and BEC based CSLA. On an average,
39
Multiplier proposed carry select adder requires 21 % less area than
Delay MCSLA and 44% less area than BEC based CSLA for
(ns) different bit widths. Hence proposed Vedic multiplier is
implemented using this improved CSLA. Proposed Vedic
multiplier requires less number of gates. On an average,
Fig. 11. Delay comparison of proposed and conventional Vedic multiplier proposed Vedic multiplier requires 6% less area than Vedic
multiplier using MCSLA and 16% less area than Vedic
multiplier using BEC based CSLA. Thus proposed Vedic
TABLE III. COMPARISON OF DELAY AND AREA BETWEEN BOOTH AND multiplier is more area-efficient than conventional Vedic
PROPOSED VEDIC MULTIPLIER
multipliers. Proposed Vedic multiplier is also compared
with conventional Vedic multipliers in terms of delay.
Design
Width Area Delay Proposed Vedic multiplier requires less delay than Vedic
(n) (gate count) (ns) multiplier using MCSLA whereas it requires more delay
than Vedic multiplier using BEC – based CSLA. Proposed
Vedic multiplier is also compared with the Booth multiplier.
Booth Proposed Vedic multiplier has approximately 43 % less area
8 bits 2301 52.677
Multiplier [4] than Booth multiplier. Proposed multiplier is faster than
Booth multiplier as it has approximately 15 % less delay
than Booth multiplier. Thus, Proposed Vedic multiplier is
Proposed Vedic superior to Booth multiplier in terms of delay as well as
8 bits 1293 44.358
Multiplier
area.
2500
References
2000 [1] P. Y. Bhavani, G. Chokkakula, S. P. Reddy and N. R. Samhitha,
“Design of low power and high speed modified carry select adder for
1500
16 bit Vedic multiplier,”IEEE, Feb 2014.
1000 Area [2] B. Ramkumar and H. M. Kittur, “Low-power and area-efficient carry-
select adder,” IEEE Trans. Very Large Scale Integr. (VLSI) Syst.,
500 vol. 20, no. 2, pp. 371–375, Feb. 2012.
[3] B. K. Mohanty and S. K. Patel, “Area–delay–power efficient carry-
0
select adder,” IEEE Trans. Circuits and Systems. vol. 61, no. 6, pp.
Booth Vedic 418-422, June 2014.
[4] R. P. Rajput and M. N. Shanmukha Swamy, “High speed modified
Booth encoder multiplier for signed and unsigned numbers”, UKSim
Fig. 12. Area comparison of Booth and proposed Vedic multiplier 14th International Conference on Modelling and Simulation, pp. 649-
654, March 2012.
[5] H. Thapliyal and H. R. Arbania. “A time-area-power efficient
multiplier and square architecture based on ancient Indian Vedic
54 mathematics”, proceedings of the 2004, International Conference on
VLSI (VLSI’04), Las Vegas, Nevada, pp. 434-439, June 2004.
52 [6] H. D. Tiwari, G. Gankhuyag, C. M. Kim, and Y. B. Cho, “Multiplier
50 design based on ancient Indian Vedic mathematics”, Proc. Int SoC
48 Design Conf., pp.65-68. 2008.
Delay [7] H. Thapliyal and M. B. Srinivas, “VLSI implementation of RSA
46
encryption system using ancient Indian Vedic mathematics”, Center
44 for VLSI and Embedded System Technologies, International Institute
42 of Information Technology Hyderabad-500019, India.
40 [8] G. Gokhale and P. D. Bahirgonde, “Design of Vedic Multiplier using
Area-Efficient Carry Select Adder”, 4th IEEE International
Booth Vedic
Conference on Advances in Computing, Communications and
Informatics (ICACCI-2015), Kochi, August 10-13, 2015, India.
[9] S. Brown and Z. Vranesic, “Fundamentals of digital logic design with
Fig. 13. Delay comparison of Booth and proposed Vedic multiplier VHDL. New Delhi, India: Tata McGraw-Hill 2007.
300