Method and apparatus for modified carry-save determination of arithmetic/logic zero results

Claim 2

In a computing system including an arithmetic operation unit responsive to a carry-in signal for performing an arithmetic operation on a pair of operands, A and B, each of said operands comprising an ordered sequence of N bits, A(i) and B(i), respectively, where i=0 . . . N-1, and, for each operand, the significance of operand bit i being greater than the significance of any operand bit i+M, where M=1 . . . N-1-i, and for producing a digital output signal representing a result of said arithmetic operation, an improvement to said computing system for predicting when said result has a zero magnitude, said improvement comprising: a carry-save adder circuit responsive to said operands for producing an ordered sequence of intermediate result signals by carry-free combination of operand A with operand B; and condition code prediction means connected to said carry-save adder circuit for generating a condition code signal based upon an algebraic combination of said carry-in signal with said ordered sequence of intermediate result signals and for providing an indication whether the magnitude of said result equals zero, said indication being provided substantially simultaneously with said result.

Claim 3

The improvement of claim 1, wherein said carry-save adder circuit produces intermediate result signals Pxi and Pyi, such that: Pxi=(A(i) V B(i))', and Pyi=(A(i)|B(i)), where ' is the logical complement, V is the exclusive-OR operation, and | is the OR operation.

Claim 4

The improvement of claim 2, wherein said carry-in signal is a digital signal having a value of "1" or "0", and said condition code prediction means receives said intermediate result signals and said carry-in signal to produce said condition code signal ("RESULT=0") according to: "RESULT=0"=(Px(N-1)·D), for carry-in=0, and "RESULT=0"=(Px(N-1)'·D), for carry-in=1, where D=[(Px(0) V Py(1))·(Px(1) V Py(2))· . . . ·(Px(N-2) V Py(N-1))], where "·" is the AND operation.

Claim 5

The improvement of claim 1, wherein said condition code prediction means includes means responsive to said condition code signal for producing a second condition code signal indicating a sign of said result.

Claim 6

The improvement of claim 1 wherein said carry-save adder circuit produces intermediate result signals Pxi and Pyi, such that: Pxi=(A(i) V B(i)')', and Pyi=(A(i)|B(i)'), where ' is the logical complement V is the exclusive-OR operation, and | is the OR operation.

Claim 7

The improvement of claim 5, wherein said carry-in signal is a digital signal having a value of "1" or "0", and said condition code prediction means receives said intermediate result signals and said carry-in signal to produce said condition code signal ("RESULT=0") according to: "RESULT=0"=(Px(N-1)·D) for carry-in=0, and "RESULT=0""=(Px(N-1)·D), for carry-in=1, where D=[(Px(0) V Py(1))·(Px(1) V Py(2))· . . . ·(Px(N-2) V Py(N-1))], where "·" is the AND operation

Claim 8

In a system including an arithmetic unit for selectively performing a plurality of arithmetic operations on a pair of operands A and B to produce a result R in response to a binary carry-in signal C in , said carry-in signal having a first state, "0", indicating a first class of arithmetic operations in which no carry-in to the arithmetic unit occurs and a second state, "1", indicating a second class of arithmetic operations in which a carry-in to the arithmetic unit occurs, each of said operands including an ordered sequence of N bits, A(i) and B(i), respectively, where i=0. . . N-1, a method for generating a predictive condition code signal "RESULT=0" reflecting whether a result produced by said arithmetic unit will be zero, said method being executed in an apparatus including a modified adder connected to a condition code prediction circuit, and including the steps of: algebraically combining said operands in said modified adder to produce a first sequence of intermediate term signals Px=Px(0) . . . Px(i) . . . Px(N-1) and Py=Py(0) . . . Py(i) . . . Py(N-1), wherein: when C in =0, Px(i)=(A(i) V B(i))' Py(i)=(A(i)|B(i); and when C in =1 Px(i)=(A(i) V B(i)')' Py(i)=(A(i)|B(i)') when C in =0, algebraically combining said first sequence of intermediate term signals in said condition code prediction circuit with a "1" to produce a second sequence of intermediate term signals according to: Px+Py+1=(Px(0+Py(1))||(Px(1)+Py(2))|| . . . ||(Px(N-2)+Py(N-1))||(Px(N-1)+1); and generating said predictive condition code signal to reflect production of a zero result when each term of said second sequence has a magnitude of zero; and when C in =1, algebraically combining said first sequence of intermediate term signals in said condition code prediction circuit with two "1"s to produce a third sequence of intermediate term signals according to: Px+Py+1+1=(Px(0)+Py(1))|| . . . ||(Px(N-2)+Py(N-1))||(Px(N-1)+1+1); and generating said predictive condition code signal to reflect production of a non-zero result when each term of said third sequence has a magnitude of zero, where 1. | represents logical OR, 2. · represents logical AND, 3. V represents logical exclusive-OR, 4. + represents addition, 5. - represents subtraction, 6. A' is the one's complement of A, and 7. || represents concatenation.

Claim 9

In a system including an arithmetic unit for selectively performing a plurality of arithmetic operations on a pair of operands A and B to produce a result R in response to a binary carry-in signal C in , said carry-in signal having a first state, "0", indicating a first class of arithmetic operations in which no carry-in to the arithmetic unit occurs and a second state, "1", indicating a second class of arithmetic operations in which a carry-in to the arithmetic unit occurs, each of said operands including an ordered sequence of N bits, A(i) and B(i), respectively, where i=0. . . N-1, a method for generating a predictive condition code signal "RESULT=0" reflecting whether a result produced by said arithmetic unit will be zero, said method being executed in an apparatus including a modified adder connected to a condition code prediction circuit, and including the steps of: algebraically combining said operands in said modified adder to produce a first sequence of intermediate term signals Px=Px(0) . . . Px(i) . . . Px(N-1) and Py=Py(0) . . . Py(i) . . . Py(N-1), wherein: when C in =0, Px(i)=(A(i) V B(i))' Py(i)=(A(i)|B(i); and when C in =1 Px(i)=(A(i) V B(i)')' Py(i)=(A(i)|B(i)') when C in =0, algebraically combining said first sequence of intermediate term signals in said condition code prediction circuit with a "1" to produce a second sequence of intermediate term signals according to: Px+Py+1=(Px(0+Py(1))||(Px(1)+Py(2))|| . . . ||(Px(N-2)+Py(N-1))||(Px(N-1)+1); and generating said predictive condition code signal to reflect production of a zero result when each term of said second sequence has a magnitude of zero; and when C in =1, algebraically combining said first sequence of intermediate term signals in said condition code prediction circuit with two "1"s to produce a third sequence of intermediate term signals according to: Px+Py+1+1=(Px(0)+Py(1))|| . . . ||(Px(N-2)+Py(N-1))||(Px(N-1)+1+1); and generating said predictive condition code signal to reflect production of a non-zero result when each term of said third sequence has a magnitude of zero, where 1. | represents logical OR, 2. · represents logical AND, 3. V represents logical exclusive-OR, 4. + represents addition, 5. - represents subtraction, 6. A' is the one's complement of A, and 7. || represents concatenation.

Claim 10

The method of claim 7, wherein said steps of algebraically combining include the exclusive disjunction of Px and Py by said condition code prediction circuit according to: D=[(Px(0) V Py(1))·(Px(1) V Py(2))· . . . ·(Px(N-2) V Py(N-1))]; and wherein, when C in =0, said predictive condition code signal is generated according to: "RESULT=0"=Px(N-1)·D; and when C in =1, said predictive condition code signal is generated by said condition code prediction circuit according to: "RESULT=0"=Px(N-1)'·D.

Claim 11

The method of claim 8, wherein said condition code prediction circuit includes means for generating indication signals J and K, such that J is true when C in =0, and K is true when C in =1, and selective means for combining D, Px(N-1), Px(N-1)' to produce said signal "RESULT=0", according to: "RESULT=0"=D·[(J·Px(N-1))|(K·Px(N-1)')]