GB2184579A

GB2184579A - A multi-stage parallel binary adder

Info

Publication number: GB2184579A
Application number: GB08629927A
Authority: GB
Inventors: Richard David Simpson; Derek Roskell
Original assignee: Texas Instruments Ltd
Current assignee: Texas Instruments Ltd
Priority date: 1985-12-20
Filing date: 1986-12-15
Publication date: 1987-06-24
Also published as: US4858167A; GB8629927D0; JPS62241029A; GB8531380D0; JP2519227B2; GB2184579B

Description

1

SPECIFICATION

A multi-stage parallel binary adder i A GB 2 184 579 A 1 This invention relatesto a multi-stage parallel binary adder using semiconductor circuitry and having high speed carry propagation.

In the search for higher operating speeds for computing devices various proposals have been made to reduce thetime needed to add two numbers together by increasing the rate of propagation of carry signals along the stages of a parallel adder. Many of these proposals have required complex logical circuitrywhich is expensive to construct in terms of semiconductor components. It is clearly desirableto construct components for computing devices in the form of integrated circuits so thatthey can be produced at relatively low costand with high reliab[lity. In orderto include as many components as possible on an integrated circuit, it is advantageousto eliminate asfar as possible the need forcomplicated logical devices such as have been proposedfor certain carry chainswhilst retaining the advantages of higher speed of carry propagation, bearing in mind thatthe chief cause of delay in computing circuitry lies in the responsetime of the gates and switches used.

It is an object of the present invention to provide an improved multistage parallel binary addercircuit which is particularly suited to manufacture as all or partof an integrated circuit.

Accordingtothe present invention there is provided a multi-stage parallel binary adder circuit having a carry path and in which each stage has afirstoutput fora carry propagate signaland a second outputfora carrygenerate signal and logic means are providedfor each group of at leasttwo successive stages for combiningthe carry propagate signals andthe carry generate signaisfrom the stages ofthe particular groupto produce a group carrying propagate output and a group carry generate output, wherein for each group of stagesthe carry path has a groupcarry input conductor for receiving the carry outputfrom the immediately preceding.group of stages anda group carry output conductor for conveying thecarry output from the particular group to the immediately following group, a f irsttransistor connecting the input conductorto a first reference voltage in response to a pre-charge clocksignal prior to the generation of carry propagate signals and carry generate signals bythe stages of the circuit, a second transistor responsive to the group carry propagate signal from the particular group to connectthe group carry inputconductor to the group carry output conductorfor that group, and a third transistor responsive to the group carry generate signal from the particular group to. connectthe group carry output conductorto a second reference voltage.

Thetransistors may be MOS..transistors.

The logic means and the adder staú[es may also be constructed from dynamic MOS circuits and the entire 60. adder circuit may b--form,edas. all or part of. an Integrated circuit.

A group of stages may consist of 2 stages and the circuit may include for each group of stages a fourth MOS transistor connected from the group carry input % conductorand responsivetothe carry propagate signal from the first stage of the groupto apply a carry Inputsignal tothesecond stage of the groupthrough an OR gate,the second inputof which is connectedto receive a carry generate signal from thefirststage.

Thecarryand othersignalsmay be positive or negative going as requiredwith inverters included where necessary, andthe conductivity types of the MOS transistor channels should be selected to suitthe signal polarities.

In orderthatthe invention maybefully understood and readilycarried out into effectitwill nowbe describedwith referencetothe accompanying drawings, of which:- FIGURE 1 is a diagram of part of an embodiment of the invention; and Figure 2 shows a part of Figure 1 in more detail.

Figure 1 showsfive adding stages of a multi-stage parallel binary adder; the stages are represented by the blocks 1, 2,3,4 and 5 and have inputsforthe respective bits of numbers A and B togetherwith the carryfrom the preceding stage. Thus the block 2 represents stage 1 and has inputs All and B1 representing the digits of the numbers A and B Co representing the carry from stage 0. Each stage produces in addition to the sum output D a carry propagate signal P and a carry generate signal G. An incoming carry Cin is applied in inverted form to a conductor 6which forms the beginning of the carry propagate chain of the adder. The conductor 6 is connectedto conductor7 through a transistor 8 which in turn is connected to a conductor 9 through a transistor 10 which in turn is connected to a conductor 11 through a transistor 12. The conductors 6,7 and 9 are connected to a voltage supply Wethrough transistors 13,14 and 15 resepctively, which transistors are rendered conducting by a precharge signal 0P.

In the control of the propagation and generation of carries, the stages of the adderare grouped in pairs, stages 0 and 1 together, stages 2 and 3together, stage 4with stage 5, not shown, and so on. Sincethe circuits arethe sameforeach pairof stages, onlythe circuit associated with stages 0 and 1 will be described. The incoming carry signal Cin is applied via an inverter 16 to the carryinput of stage 0. The conductor 6 is also connected through a transistor 17controlled by PO, the carry propagate signal fromstage 0,to an input of a NOR gate 18 connected to the carry inputof stage 1. The signal PO is applied togetherwith a signal P1 from stage 1 tothe inputs of a NAND gate 19,the output of which controls the transistor 8through an inverter 20. Sincethe signals PO and P1 represenithe carry propagate signaisfrom the stages 0 and 1, itwill be a ppa rentthat when combined bythe NAND gate 19 they serve to convey an inverted carry input signal through the transistor 8to the conductor7. The carry generate signal G0from the stage Gafter inversion by an inverter21 is applied to a second input of the NOR gate 18which conveysthe carryto stage 1. The signal The drawing(s) originally filed was (were) informal and the print here reproduced is taken from a later filed format copy.

2 GB 2 184 579 A 2 GO is also applied to a NAND gate 22 together with the carry propagate signal P1 from stage 1 and the output of the gate 2 after inversion by an inverter 23 is applied to a NOR gate 24the output of which inverted by an inverter 25 controls a transistor26 connected from the 70 conductor7 to earth. The carry generate signal G1 from stage 1 is applied to a second input of the NOR gate 24.

The circuitry just described performs the logical operations to combine the ca rry propagate and carry generate signals from the stages 0 and 1 sothata signal representing the inverted carry from the stage 1 is correctly produced with the only delay imposed on the propagation of a carry input applied in inverted form to the conductor 6 is that due to the transistor 8.

The form of logic used in the carry shown is known as dynamic MOS logic and operates bythe precharging of the conductors 6,7,9, etc. to the voltage Vcc by the precharge signal OP. A carry is represented bythe discharge of the appropriate conductorto earth, the conductor 6 being discharged by means not shown although the input is represented by Cin. The transis tor 26 serves to discharge the conductor7 if the transistor8 does not conduct, this beingthe situation if the stages 0 and 1 produce a carry output although they may not necessarily propagate a carry from the conductor 6. If both stages 0 and 1 produce a carry propagate signal, that is to say both PO and P1 a re present, then the NAND gate 19 will respond and cause the inverter20 to apply a signal to the transistor 95 Sto cause itto conduct so that a ny inverted carry input signals wil 1 discharge the conductor 7. Stages 0 and 1 together wil 1 produce a carry output regardless of the input U either the stage 1 produces a carry generate signal G1 or stage 0 produces a carry generate signal 100 GO and stage 1 produces a carry propagate signal P1.

In the latter case, the gate 22 reacts and applies the signal tothe gate 24 and in the former case the signal G1 is applied directly to the gate 24. When either of these is present, the gate 24 reacts and causes the inverter 25 to switch on the transistor 26 and discharge theconductor7.

Incoming carries to stage 0 can be derived from the conductor 6 only and are fed to the stage through the inverter 16. Incoming carries to stage 1 canbederived from the conductor 6 if stage 0 is producing a carry propagat e signal PO when the transistor 17 conducts.

In addition, the stage 0 could produce a carry generate signal GO and this is applied through the inverter 21 to the NOR gate 18 which could receive an invertedcarry from the conductor 6 via the transistor 17.

Figure 2 shows in greater detail the construction of a pair of stages of the adder shown in Figure 1 using dynamic MOS logic. As faras possible the same reference notation is used in Figure 2 aswas used in Figure 1 and components corresponding to those just described with referenceto Figure 1 have the same references as in thatfigure. Itshould be noted that whereas in Figure 1 thestage numbers increase from leftto rightwith the carry being propagated In the 125 ame direction, in Figure 2 the stage numbers increase andthe carry propagation takes placefrom rightto.

left. Another point of dissimilarity between Figures 1 and 2 is that in Figure 2 each adding stage is in two parts, the stage n consisting of thetransistors 39 to 43 130 and the transmission gates 30 and 31. Atransmission gate consists of an N- channel MOSFET and a Pchannel MOSFET connected in parallel with appropriate signals applied to their gates so thatthey are turned off and on together; such agate hasthe advantage that it produces substantially no attenuation of the logic signal levels.

Considering the stage n of the adder, the transistors 39 to 43 together with the associated inverters function as a half adder in response to inverted input digits An and Bn to produce a carry propagate signal Pn and a carry generate signal Gn. The signal Pn is applied to the inputs of the transmission gates 30 and 31, through an inverter in the case of the gate 31 and these gates are control led by a carry signal Cri-1 derived by inverting the sig nal on the conductor 6. The output of the tranmission gates 30 and 31 forms the sum signa 1 Dn of the stage n. The stage n + 1 is of similar construction to the stage n and produces a carry propagate signal Pn+ 1 and a ca rry generate signal Gn+ 1. As in Figu re 1, the carry propagate signals Pn and Pri+1 control thetransistor8 by combination in a NAND gate 19the outputof which is inverted bythe inverter20. Carry generate signals Gn and Gn+l are combined by a logic arrangement which differs slightly from that shown in Figure 1 in thatthe NOR gate 24of Figure 1 is replaces by a NAND gate24A becausethe inverters 23 and 25 are omitted, andthe carry generate signal Gn+l is inverted before application to the gate 24A.The production of the carry inputto the stage n+l,the signal Cri, is achieved by transistors 34,35 and 36 of which thetransistor36 corresponds in function to the transistor 17 of Figure 1 and.its logical combination with the carry inputto the stage n, Cn-1, is provided by the transistor 35 which is connected in serieswith the transistor 36. The alternative source of the signal Cri which isthe carry generatesignal Gn from the stage n is provided bythe transistor 34which is connected in parallel with the series combination of transistors 35 and36. - The precharge clocksignal OP is used in Figure 2 in the samewayas in Figure 1 and alsoto enablethe functioning of the logical circuitryof the adding stages. An evaluate clock OE is employed in Figure 2 to enable the selective discharge of the conductors precharged by the precharge clock in response to the logical combination of signals applied to the transistorsofthe circuits involved.The precharge clock OP andthe evaluateclockOE are applied altenatelyto their respective connections.

From a consideration of the circuits described above with reference to Figures 1 and 2, it will be apparent thatthe carry chain includes only a single series connected transistorfor every two stages of the adder so thatthe propagation delay of the carry signal along the carry chain will be about half of what itwould be for a conventional carry chain having a transistorfor each stage of the adder. The invention therefore provides a significant increase in the speed of operation of a parallel adderfor only a small increase in the complexity of the carry propagation and generation circuitry.

The carry propagate andcarry generate signals from groups of morethan two stages could be combined logically, butthe logic circuitrywould be f- t, 3 GB 2 184 579 A 3 X v A A much more complicatedthan that described above, and the additional saving in propagation time along thecarrychain dueto the reduction in the number of series transistors would be offsetto some extent by the extra delay imposed on the generation of the control signals forthosetransistors; bythe more complicated logicclrcuitry.

Although the invention has been described with reference to embodiments using dynamic MOS logic some or all of the MOS transistors could be replaced byjunction FET's or bipolartrans[stors. Bipolar transistors have different characteristics from MOS transistors and their use in place of MOS transistors in the circuits described above may enhance the per-

Claims

formance of the circuits in certain applications. Moreoverthe chip area neededto accommodate a circuit can be reduced if certain of thetransistors are bipoiarones. CLAIMS

1. A multi-stage parallel binary adder circuit having a carrypath and in which each stage has afirst output fora carry propagate sig nai and a second outputfora carry generate signal and logic means are provided for each group of at leasttwo successive stages for combining the carry propagate signals and the carry generate signals from the stages of the particular group to produce a group carry propagate output and a group carry generate output, wherein for each group of stages the carry path has a group carry input conductorfor receiving the carry outputfrom the immediately preceding group of stages and a group carry output conductorfor conveying the carry output from the particular group to the immediately following group, a firsttransistor connecting the input conductorto a first reference voltage in response to a pre-charge clock signal priorto the generation of carry propagate signals and carry generate signals by the stages of the circuit, a second transistor responsive to the group carry propagate signal from the particular group to conneetthe group carry input conductorto the group carry output conductor for that group, and a third transistor responsive to the group carry generate signal from the particular group to conneetthe group carry output conductorto a second reference voltage.

2. An adder circuit according to claim 1 wherein the transistors are MOS transistors.

3. An adder circuit according to claim 1 or 2 wherein the logic means for each group of stages is constructed from dynamic MOS transistor logic circuits and further includes logic circuits connected to the group carry input conductor and thecarry propagate and carry generate outputs of all stages in the group exceptthe lastto produce carry input signals which are applied as carry signal inputs of all stages in the group exceptth.e first.

4. An adder circuit according to claim 3 wherein each group consists of two,stages and the logic means of a group hastwo MOS transistors connected in a series pathshunted by another MOStransistor, from a 6Q cQnductoretementeon,nectedto be precharged tothe first reference voltage through a precharge MOS transistor responsiveto a precharge clock signal,to an evaluating MOS transistor connected tothe second reference voltage and responsive to anevaluateclock signal, the two MOS transistors connected in a series path being respectively responsive to the carry signal on the group carry input conductor and the carry propagate signal from the first stage of the group, and the other MOS transistor being responsive to the carry generate signal from the first stage of the group, whereby a carry signal to be applied to the second stage of the group is produced on the conductor element.

5. An adder circuit according to claim 1 or 2 wherein each group of stages consists of two stages and for each group of stages there is provided a fourth transistor connected from the group carry input conductor and responsive to the carry propagate signal from thefirst stage of the group to apply a carry input signal to the second stage of the group through an OR gate, a second input of which is connected to receive the carry generate signal f rom the first stage of the group.

6. An adder circuit according to any preceding claim wherein each stage includes first and second conductor elements connected to be precharged to the first reference voltage through respective transistors in response to the pre-charge clocksignal, first and second evaluate transistors connected to a second reference voltage and responsive to an evaluate clocksignal, a first pair of transistors connected in series from the first conductor element to the first evaluate transistor, a second pair of transistors connected in series from the second conductor element to the second evaluate transistor, a further transistor connected from the first conductor elementto the intermediate connection of thesecond pair of transistors, and first and second inputterminals for signals representing digits to be added together, the first input terminal being connected to apply the digit signal in upright form to a transistor of each of the first and second pairs, and in inverted form to the further transistor, and the second input terminal being connected to apply the digit signal in upright form to a transistor of the second pair and in inverted form to a transistor of the first pair, whereby an inverted carry propagate signal is selectively produced on thefirst conductor element and an inverted carry generate signal is selectively produced on the second conductor element in responseto the digit signals after an evaluate clock signal has occurred.

7. A adder circuit according to claim 6 wherein each stage further includes two transmission gates to which the carry propagate signal is applied in upright and inverted form respectively and which are controlled by the carry signal from the preceding stage to produc6 a sum digit output from the stage.

8. An adder circuit substantially as described herein with referenceto either Figure 1 or Figure 2 of thedrawings.

Printed in the United Kingdom for Her Majety's Stationery Office by the Tweeddale Press Group, 8991685, 6187 18996. Published at the Patent Office, 25 Southampton Buildings, London WC2A 1AV, from which copies may be obtained.