0% found this document useful (0 votes)
96 views4 pages

Design of 128 Bit Low Power and Area Efficient Carry Select Adder

The document proposes a design modification to reduce the area and power consumption of carry select adders (CSLA). It replaces the regular ripple carry adders in the CSLA with binary to excess-1 converters (BEC) to achieve lower area and power. Simulation results show the proposed 128-bit CSLA has a 21.4% lower area-delay product and 6.3% lower power-delay product compared to the regular CSLA, with only a minor increase in delay. The modified CSLA structure is then implemented in the adder part of an FIR filter to reduce the area and power consumption compared to using the regular CSLA.

Uploaded by

Rama Sukumar
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
96 views4 pages

Design of 128 Bit Low Power and Area Efficient Carry Select Adder

The document proposes a design modification to reduce the area and power consumption of carry select adders (CSLA). It replaces the regular ripple carry adders in the CSLA with binary to excess-1 converters (BEC) to achieve lower area and power. Simulation results show the proposed 128-bit CSLA has a 21.4% lower area-delay product and 6.3% lower power-delay product compared to the regular CSLA, with only a minor increase in delay. The modified CSLA structure is then implemented in the adder part of an FIR filter to reduce the area and power consumption compared to using the regular CSLA.

Uploaded by

Rama Sukumar
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
You are on page 1/ 4

International Journal of Advanced Research in Engineering (IJARE) Vol 1, Issue 1,2012 Page

31 3!

Design Of 128 Bit Low Power and Area Efficient Carry Select Adder
1
1

A.Andamut u! S."it anyaa

M.E (VLSI Design), Department of ECE, Bannari Amman Institute of Tec no!og". # M.Sc (IT), Department of Information Tec 2 no!og", $ongu Engineering Co!!ege. 1
[email protected], [email protected]

Abstract Design of area% an& po'er%efficient ig %spee& &ata pat !ogic s"stems forms t e !argest areas of researc in VLSI s"stem &esign. In &igita! a&&ers, t e spee& of a&&ition is !imite& (" t e time re)uire& to transmit a carr" t roug t e a&&er. Carr" Se!ect A&&er (CSLA) is one of t e fastest a&&ers use& in man" &ata%processing processors to perform fast arit metic functions. *rom t e structure of t e CSLA, it is c!ear t at t ere is span for re&ucing t e area an& po'er consumption in t e CSLA. T is 'or+ uses a simp!e an& efficient gate%!e,e! mo&ification to &rastica!!" re&uce t e area an& po'er of t e CSLA. Base& on t is mo&ification -, 1., /#, .0 an& 1#-%(it s)uare%root CSLA (S12T CSLA) arc itectures a,e (een &e,e!ope& an& compare& 'it t e regu!ar S12T CSLA arc itecture. T e propose& &esign as re&uce& area an& po'er as compare& 'it t e regu!ar S12T CSLA 'it on!" a minor increase in t e &e!a". T is 'or+ estimates t e performance of t e propose& &esigns in terms of &e!a", area, po'er, an& t eir pro&ucts are imp!emente& in 3i!in4 *56A. T e resu!ts ana!"sis s o's t at t e propose& CSLA structure is (etter t an t e regu!ar S12T CSLA. T e propose& &esign is app!ie& to t e *I2 fi!ter structure in t e a&&er part. Index TermsApp!ication%specific integrate& circuit (ASIC), area%efficient, CSLA, !o' po'er. 1. #$%"OD&C%#O$ Area an& po'er re&uction in &ata pat !ogic s"stems are t e main area of researc in VLSI s"stem &esign. 7ig % spee& a&&ition an& mu!tip!ication as a!'a"s (een a fun&amenta! re)uirement of ig %performance processors an& s"stems. In &igita! a&&ers, t e spee& of a&&ition is !imite& (" t e time re)uire& to propagate a carr" t roug t e a&&er. T e sum for eac (it position in an e!ementar" a&&er is generate& se)uentia!!" on!" after t e pre,ious (it position as (een summe& an& a carr" propagate& into t e ne4t position. T e ma8or spee& !imitation in an" a&&er is in t e pro&uction of

carries an& man" aut ors a,e consi&ere& t e a&&ition pro(!em. T e CSLA is use& in man" computationa! s"stems to mo&erate t e pro(!em of carr" propagation &e!a" (" in&epen&ent!" generating mu!tip!e carries an& t en se!ect a carr" to generate t e sum. 7o'e,er, t e CSLA is not area efficient (ecause it uses mu!tip!e pairs of 2ipp!e Carr" A&&ers (2CA) to generate partia! sum an& carr" (" consi&ering carr" input an& t en t e fina! sum an& carr" are se!ecte& (" t e mu!tip!e4ers (mu4). To o,ercome a(o,e pro(!em, t e (asic i&ea of t e propose& 'or+ is (" using n%(it (inar" to e4cess%1 co&e con,erters (BEC) to impro,e t e spee& of a&&ition. T is !ogic can (e imp!emente& 'it an" t"pe of a&&er to furt er impro,e t e spee&. 9sing Binar" to E4cess%1 Con,erter (BEC) instea& of 2CA in t e regu!ar CSLA to ac ie,e !o'er area an& po'er consumption. T e main a&,antage of t is BEC !ogic comes from t e !esser num(er of !ogic gates t an t e *u!! A&&er (*A) structure. Digita! Signa! 5rocessing (DS5) &ea!s 'it t e manipu!ation of &igita! signa!s using comp!e4 signa! processing s"stems (ui!t from (asic (ui!&ing (!oc+s !i+e fi!ters. T e propose& 'or+ is imp!ementation in t e *I2 fi!ter in a&&er parts. T is re&uces t e area an& !o' po'er consumption in *I2 fi!ter. ##. BAS#C '&$C%#O$ A$D S%"&C%&"E O' BEC LO(#C T e (asic 'or+ is to use Binar" to E4cess%1 Con,erter (BEC) instea& of 2CA 'it Cin:1 in t e regu!ar CSLA to ac ie,e !o'er area an& po'er consumption. T e main a&,antage of t is BEC !ogic comes from t e !esser num(er of !ogic gates t an t e n%(it *u!! A&&er (*A) structure As state& a(o,e t e main i&ea of t is 'or+ is to use BEC instea& of t e 2CA 'it Cin:1 in or&er to re&uce t e area an& po'er consumption of t e regu!ar CSLA. To rep!ace t e n%(it 2CA, an n;1%(it BEC is re)uire&. A structure an& t e function

co"

Page 31

ta(!e of a 0%(it BEC are s o'n in *igure.# an& Ta(!e .#, respecti,e!".

###. BAS#C S%"&C%&"E O' "E(&LA" 1/0B#% S1"% CSLA A 1.%(it carr" se!ect as t'o t"pes of (!oc+ siDe name!" uniform (!oc+ siDe an& ,aria(!e (!oc+ siDe. A 1.%(it carr" se!ect a&&er 'it a uniform (!oc+ siDe as t e &e!a" of four fu!! a&&er &e!a"s an& t ree M93 &e!a"s. E i!e a 1.%(it carr" se!ect a&&er 'it ,aria(!e (!oc+ siDe as t e &e!a" of t'o fu!! a&&er &e!a"s, an& four mu4 &e!a"s. T erefore 'e use 1.%(it carr" se!ect a&&er 'it ,aria(!e (!oc+ siDe. 2ipp!e%carr" a&&ers are t e simp!est an& most compact fu!! a&&ers, (ut t eir performance is !imite& (" a carr" t at must ripp!e from t e !east%significant to t e most%significant (it. A carr"%se!ect a&&er ac ie,es spee&s 0=F to G=F faster (" performing a&&itions in para!!e! an& re&ucing t e ma4imum carr" pat .

*igure #< 0%(inar" to e4cess%1 con,erter Ta(!e.# *unction ta(!e of t e 0%(it

B)*+,-

.)*+,-

==== ===1 >

===1 ==1= > >

*igure /< 2egu!ar 1.%(it S12T CSLA A carr"%se!ect a&&er is &i,i&e& into sectors, eac of ' ic , e4cept for t e !east significant performs t'o a&&itions in para!!e!, one assuming a carr"%in of Dero, t e ot er a carr"% in of one 'it in t e sector, t ere are t'o 0%(it ripp!e% carr" a&&ers recei,ing t e same &ata inputs (ut &ifferent Cin. T e upper a&&er as a carr"%in of Dero, t e !o'er a&&er a carr"%in of one. T e actua! Cin from t e prece&ing sector se!ects one of t e t'o a&&ers. If t e carr"%in is Dero, t e sum an& carr"%out of t e upper a&&er are se!ecte&. If t e carr"%in is one, t e sum an& carr"%out of t e !o'er a&&er are se!ecte&. Logica!!", t e resu!t is not &ifferent if a sing!e ripp!e%carr" a&&er 'ere use&. *irst t e co&ing for fu!! a&&er an& &ifferent mu!tip!e4ers of .</, -<0, 1=<H, an& 1#<. 'as &one. T en #, /, 0, H%(it ripp!e carr" a&&er 'as &one (" ca!!ing t e fu!! a&&er. T e regu!ar .0%(it CSLA 'as create& (" ca!!ing t e ripp!e carr" a&&ers an& a!! mu!tip!e4ers (ase& on circuit. *ina!!", regu!ar 1#-%(it 'as imp!emente& in t e *I2 fi!ter &esign (sectionH).

> 111= 1111

> > 1111 ====

T e Boo!ean e4pressions of t e 0%(it BEC is !iste& as (note t e functiona! s"m(o!s ? @AT, B A@D, C3A2) 3= : ?B= 31 : B=CB1 3# : B#C (B= B B1) 3/ : B/C (B= B B1 B B#)

remaining groupLs t e arri,a! time of mu4 se!ection input is a!'a"s greater t an t e arri,a! time of &ata inputs from t e BECLs. T us, t e &e!a" of t e remaining groups &epen&s on t e It is simi!ar to regu!ar 1.%(it S12T CSLA. An!" arri,a! time of mu4 se!ection input an& t e mu4 &e!a". c ange is t at in (asic (!oc+s a,ing t'o ripp!e%carr" a&&ers, one ripp!e carr" a&&er fe& 'it a constant 1 carr"%in is rep!ace& (" BEC. T e area estimation of eac group is 2. #3PLE3E$%A%#O$ #$ '#" '#L%E" ca!cu!ate&. A fi!ter is use& to mo&if" an input signa! in or&er to faci!itate furt er processing. A &igita! fi!ter 'or+s on a &igita! Base& on t e consi&eration of &e!a" ,a!ues, t e T e most common &igita! fi!ter is t e Linear Time%In,ariant arri,a! time of se!ection input C1 Itime (T) :JK of .</ mu4 is (LTI) fi!ter. Designing an LTI in,o!,es arri,ing at t e fi!ter ear!ier t an t e s/ It :GK an& c/ It :JK an& !ater t an t e s# It coefficients ' ic , in turn, represents t e impu!se response of :0K. T us, t e sum/ an& fina! c/ (output from mu4) are t e propose& fi!ter &esign. T ese coefficients, in !inear &epen&ing on s/ an& mu4 an& partia! c/ (input to mu4) an& con,o!ution 'it t e input se)uence 'i!! resu!t in t e &esire& mu4, respecti,e!". T e sum# &epen&s on c1 an& mu4. *or t e remaining parts t e arri,a! time of mu4 se!ection input is a!'a"s greater t an t e arri,a! time of &ata inputs from t e BECLs. T us, t e &e!a" of t e remaining M93 &epen&s on t e arri,a! time of mu4 se!ection input an& t e mu4 &e!a". *irst t e co&ing for fu!! a&&er an& mu!tip!e4ers of .</, -<0, 1=<H, an& 1#<. 'as &one. T e BEC program 'as &esign (" using @AT, 3A2 an& A@D gates. T en #, /, 0, H%(it ripp!e carr" a&&er 'as &one (" ca!!ing t e fu!! a&&er. T e mo&ifie& 1.%(it CSLA 'as create& (" ca!!ing t e ripp!e carr" a&&ers, BEC an& a!! mu!tip!e4ers (ase& upon t e circuit. *ina!!", mo&ifie& .0%(it 'as imp!emente& in t e *I2 fi!ter &esign (section H).

#2 BAS#C S%"&C%&"E O' "E(&LA" 1/0B#% S1"% CSLA

output. In t e a(o,e figure.H, t e a&&er part is rep!ace& (" t e regu!ar an& mo&ifie& S12T CSLA. T en output is compare& (et'een t e regu!ar an& mo&ifie& S12T CSLA. 2#. #3PLE3E$%A%#O$ "ES&L%S T e &esign propose& in t is paper as (een &e,e!ope& using Veri!og%7DL an& s"nt esiDe& in 3i!in4 ISE G.1i. *or eac 'or& siDe of t e a&&er, t e same ,a!ue c ange& &ump (VCD) fi!e is generate& for a!! possi(!e input con&itions an& importe& t e same to 3i!in4 ISE G.1i 5o'er Ana!"sis to perform t e po'er simu!ations. T e simi!ar &esign f!o' is fo!!o'e& for (ot t e regu!ar an& mo&ifie& S12T CSLA. Ta(!e H e4 i(its t e simu!ation resu!ts of (ot t e CSLA fir fi!ter structures in terms of &e!a", area an& po'er. T e po'erM&e!a" pro&uct of t e propose& -%(it is ig er t an t at of t e regu!ar S12T CSLA (" H.#F an& t e area% &e!a" pro&uct is !o'er (" #.GF. 7o'e,er, t e po'er%&e!a" pro&uct of t e propose& 1.%(it S12T CSLA re&uces (" 1.J.F an& for t e /#%(it, .0%(it an& (" as muc as -.1-F, an& 1#.#-F respecti,e!".

*igure. .. Mo&ifie& 1.%(it S12T CSLA. T e arri,a! time of se!ection input of .</ mu4 is ear!ier. T us, t e sum/ an& fina! c/ (output from mu4) &epen&s on s/ an& mu4 an& partia! c/ (input to mu4) an& mu4, respecti,e!". T e sum# &epen&s on c1an& mu4. *or t e input (a se)uence of num(ers, resu!ting from samp!ing an& )uantiDing an ana!og signa!) an& pro&uces a &igita! output.

%ABLE 4 CO3PA"#SO$ O' %5E "E(&LA" A$D 3OD#'#ED S1"% CSLA #$ '#" '#L%E" S.$ O ADDE"S DELA 6 7ns8 *I2 fi!ter using regu!ar .0%(it 1 *I2 fi!ter using mo&ifie& .0%(it *I2 fi!ter using regu!ar 1#-%(it # *I2 fi!ter using mo&ifie& 1#-%(it 1=#.#H 1,.#,H.J JH0.01 G/.GH/ -0.H. 0.,JG= 1,G=,G0J /-/.JJ GHJ.H. J1.-J/ 0J,#J/ 0##./J A"EA 7gates8 PO9E" 7m98

2##. CO$CL&S#O$ A simp!e approac is propose& in t is paper to re&uce t e area an& po'er of S12T CSLA arc itecture. T e re&uce& num(er of gates of t is 'or+ offers t e great a&,antage in t e re&uction of area an& a!so t e tota! po'er. T e compare& resu!ts s o' t at t e mo&ifie& S12T CSLA as a s!ig t!" !arger &e!a" (on!"/.J.F), (ut t e area an& po'er of t e 1#-%(it mo&ifie& S12T CSLA are significant!" re&uce& (" 1J.0F an& 1H.0F respecti,e!". T e po'er%&e!a" pro&uct an& a!so t e area%&e!a" pro&uct of t e propose& &esign s o' a &ecrease for 1., /#,.0 an& 1#-%(it siDes ' ic in&icates t e success of t e met o& an& not a mere tra&eoff of &e!a" for po'er an& area. T e mo&ifie& CSLA arc itecture is t erefore, !o' area, !o' po'er, simp!e an& efficient for VLSI ar&'are imp!ementation. "E'E"E$CES I1K Be&ri8, A. N., (1G.#), OCarr"%se!ect Trans. Electron. Comput., pp./0=M/00 . a&&er,P IRE I0K 2am+umar,B. , $ittur, 7.M. an& $annan ,5. M. ,(#=1= ),OASIC imp!ementation of mo&ifie& faster carr" sa,e a&&er,P Eur. '. $c&. Res., vol. 42, no. 1,pp.H/MH-. IHK $im ,Q. an& $im ,L.%S.,((a%2001", O.0%(it carr"% se!ect a&&er 'it re&uce& area,PElectron Lett., vol. 3), no. 10, pp. *14 *1!. I.K E. A(u%S ama an& M. Ba"oumi, OA @e' ce!! for !o' po'er a&&ers,P in 5roc.Int. Mi&'est S"mp. Circuits an& S"stems, 1GGH, pp. 1=10M1=1J IJK '''.4i!in4.com.

I#K Ceiang ,T. Q. an& 7siao,M. N. ,(Oct1998 ),OCarr"% se!ect a&&er using sing!e ripp!e carr" a&&er,P Electron. Lett., vol. 34, no. 22, pp. 2101 2103 I/K 7e, Q. , C ang ,C. 7. an& 6u, N. , (200!", OAn A rea efficient .0%(it s)uare root carr"%se!ect a&&er for !o' po'er app!ication,P in #roc. IEEE Int. $%mp.C&rcu&ts $%st. vol. 4, pp. 4082 408!.

You might also like