Backward-compatible computer architecture with extended word size and address space

Claim 2

In a computer system including a processing unit and a memory subsystem, a processing unit comprising: means, responsive to instructions fetched from said memory subsystem, for specifying corresponding operations wherein an m-bit entity on which operations can be performed has a most significant bit designated bit(m-1) and a least significant bit designated bit(0), an N-bit entity on which operations can be performed has a most significant bit designated bit(N-1) and a least significant bit designated bit(0), and N is greater than m; the instructions including a first subset of instructions, called m-bit instructions, and a second subset of instructions, called N-bit instructions; the m-bit instructions including a number of m-bit load instructions, a number of logical operation instructions, a number of m-bit shift instructions, and a number of m-bit addition instructions; the N-bit instructions including a number of N-bit load instructions, a number of N-bit shift instructions, and a number of N-bit addition instructions; a load circuit for performing specified load operations to retrieve entities up to N bits long from said memory subsystem, including means, responsive to at least one m-bit load instruction but non-responsive to any N-bit load instructions, for converting a retrieved m-bit entity to an N-bit entity where bit(N-1) through bit(m) of the N-bit entity are all equal to the value of bit (m-1) of the m-bit entity and bit (m-1) through bit(0) of the N-bit entity are respectively equal to the values of bit(m-1) through bit(0) of the m-bit entity; a register file having a set of N-bit registers; a logic circuit for performing specified logical operations between two N-bit operands and providing an N-bit result; a shift circuit for performing specified shift operations on an N-bit operand and providing an N-bit result, including means, responsive to at least one m-bit shift instruction but non-responsive to any N-bit shift instructions, for changing the N-bit result of a shift operation by setting bit(N-1) through bit (m) of the N-bit result equal to the value of bit(m-1) of the N-bit result; an adder for performing specified addition operations between a pair of N-bit operands and providing an N-bit result; means, coupled to said adder and responsive to at least one m-bit addition instruction but non-responsive to any N-bit addition instructions, for requiring that the N-bit result of an operation in said adder have bit(N-1) through bit(m) equal to the value of bit(m-1) of the N-bit result; means defining N-bit data paths from output ports of each of said load circuit, said logic circuit, said shift circuit, and said adder to said register file; and means defining N-bit data paths from said register file to input ports of each of said logic circuit, said shift circuit, and said adder.

Claim 3

The processing unit of claim 1 wherein N=64 and m=32.

Claim 4

The processing unit of claim 1 wherein said load circuit further comprises means for sign-extending or zero-extending to m bits a retrieved entity of fewer than m bits to further define a retrieved entity of m bits.

Claim 5

The processing unit of claim 1 wherein: said adder has carryouts from bit positions bit(m-1) and bit(m-2), and provides a signal indicating a condition, referred to as m-bit overflow, where said carryouts from bit positions bit(m-1) and bit(m-2) differ; and said means for requiring operates to change the N-bit result of an operation by setting bit(N-1) through bit(m) of the N-bit result equal to the value of bit(m-1) of the N-bit result only if an m-bit overflow has occurred.

Claim 6

The processing unit of claim 1 wherein: said adder has carryouts from bit positions bit(m-1) and bit(m-2), and provides a signal indicating a condition, referred to as m-bit overflow, where said carryouts from bit positions bit(m-1) and bit(m-2) differ; said m-bit addition instructions include ADD and SUB instructions that trap on an m-bit overflow and produce no result, and ADDU and SUBU instructions that do not trap on an m-bit overflow and do produce a result; and said means for requiring operates to change the N-bit result of an operation in said adder by setting bit(N-1) through bit(m) of the N-bit result equal to the value of bit(m-1) of the N-bit result if an m-bit overflow has occurred in the course of executing said ADDU and SUBU instructions and does not operate in connection with executing said ADD and SUB instructions.

Claim 7

The processing unit of claim 1, wherein: said adder has carryouts from bit positions bit(N-1) and bit(N-2), and provides a signal indicating a condition, referred to as N-bit overflow, where the carryouts from bit positions bit(N-1) and bit(N-2) differ; and said N-bit addition instructions include DADD and DSUB instructions that trap on an N-bit overflow and produce no result, and DADDU and DSUBU instructions that do not trap on an N-bit overflow and do produce a result.

Claim 8

The processing unit of claim 1 wherein said shift circuit comprises: means, responsive to a first opcode, designated an N-bit logical right shift, for (1) shifting the N-bit operand a designated number of bit positions S to the right and, if the designated number is non-zero, (2) filling bit positions bit(N-1) through bit(N-S) of the N-bit result with 0's; means, responsive to a second opcode, designated an m-bit logical right shift, for (1) shifting the N-bit operand a designated number of bit positions S to the right and, if the designated number is non-zero, (2) filling bit positions bit(N-1) through bit(m-S) of the N-bit result with 0's; means, responsive to a third opcode, designated an N-bit left shift, for (1) shifting the N-bit operand a designated number of bit positions S to the left and, if the designated number is non-zero, (2) filling bit positions bit(S-1) through bit(0) of the N-bit result with 0's; and means, responsive to a fourth opcode, designated an m-bit left shift, for (1) shifting the N-bit operand a designated number of bit positions S to the left and, if the designated number is non-zero, (2) filling bit positions bit(S-1) through bit(0) with 0's to provide an N-bit initial result, and (3) setting bit(N-1) through bit(m) of the N-bit result to the value of bit(m-1) of the N-bit initial result.

Claim 9

In the execution unit of a data processor, an adder for combining two N-bit operands to produce an N-bit result, the result and each operand having a most significant bit designated bit(N-1) and a least significant bit designated bit(0), comprising: means, responsive to a first opcode, designated an N-bit add, for adding a pair of input operands and providing an N-bit result; and means, responsive to a second opcode, designated an m-bit add, for adding a pair of input operands and providing an N-bit initial result, and, in at least those cases where bit(m) of the initial result is unequal to bit(m-1) of the initial result, for changing the N-bit initial result by setting bit(N-1) through bit(m) of the N-bit initial result to the value of bit(m-1) of the initial result and producing the initial result, thus changed, as the result.

Claim 10

In the execution unit of a data processor, an adder for combining two N-bit operands to produce an N-bit result, the result and each operand having a most significant bit designated bit(N-1) and a least significant bit designated bit(0), comprising: means, responsive to a first opcode, designated an N-bit add, for adding a pair of input operands and providing an N-bit result; and means, responsive to a second opcode, designated an m-bit add, for adding a pair of input operands and providing an N-bit initial result, and, in at least those cases where bit(m) of the initial result is unequal to bit(m-1) of the initial result, for changing the N-bit initial result by setting bit(N-1) through bit(m) of the N-bit initial result to the value of bit(m-1) of the initial result and producing the initial result, thus changed, as the result.

Claim 11

In the execution unit of a data processor, a shift unit for shifting an N-bit input operand to produce an N-bit result, the input operand and the result each having a most significant bit designated bit(N-1) and a least significant bit designated bit(0), comprising: means, responsive to a first opcode, designated an N-bit logical right shift, for (1) shifting the input operand a designated number of bit positions S to the right and, if the designated number is non-zero, (2) filling bit positions bit(N-1) through bit(N-S) of the N-bit result with 0's; and means, responsive to a second opcode, designated an m-bit logical right shift, for (1) shifting the input operand a designated number of bit positions S to the right and, if the designated number is non-zero, (2) filling bit positions bit(N-1) through bit(m-S) of the N-bit result with 0's.

Claim 12

In the execution unit of a data processor, a shift unit for shifting an N-bit input operand to produce an N-bit result, the input operand and the result each having a most significant bit designated bit(N-1) and a least significant bit designated bit(0), comprising: means, responsive to a first opcode, designated an N-bit logical right shift, for (1) shifting the input operand a designated number of bit positions S to the right and, if the designated number is non-zero, (2) filling bit positions bit(N-1) through bit(N-S) of the N-bit result with 0's; and means, responsive to a second opcode, designated an m-bit logical right shift, for (1) shifting the input operand a designated number of bit positions S to the right and, if the designated number is non-zero, (2) filling bit positions bit(N-1) through bit(m-S) of the N-bit result with 0's.

Claim 13

In the execution unit of a data processor, a shift unit for shifting an N-bit input operand to produce an N-bit result, the input operand and the result each having a most significant bit being bit(N-1) and a least significant bit designated bit(0), comprising: means, responsive to a first opcode, designated an N-bit left shift, for (1) shifting the input operand a designated number of bit positions S to the left and, if the designated number is non-zero, (2) filling bit positions bit(S-1) through bit(0) of the N-bit result with 0's; and means, responsive to a second opcode, designated an m-bit left shift, for (1) shifting the input operand a designated number of bit positions S to the left and, if the designated number is non-zero, (2) filling bit positions bit(S-1) through bit(0) with 0's to provide an initial result, and (3) setting bit(N-1) through bit(m) of the N-bit result to the value of bit (m-1) of the initial result.

Claim 14

In the execution unit of a data processor, a shift unit for shifting an N-bit input operand to produce an N-bit result, the input operand and the result each having a most significant bit being bit(N-1) and a least significant bit designated bit(0), comprising: means, responsive to a first opcode, designated an N-bit left shift, for (1) shifting the input operand a designated number of bit positions S to the left and, if the designated number is non-zero, (2) filling bit positions bit(S-1) through bit(0) of the N-bit result with 0's; and means, responsive to a second opcode, designated an m-bit left shift, for (1) shifting the input operand a designated number of bit positions S to the left and, if the designated number is non-zero, (2) filling bit positions bit(S-1) through bit(0) with 0's to provide an initial result, and (3) setting bit(N-1) through bit(m) of the N-bit result to the value of bit (m-1) of the initial result.

Claim 15

A shift unit for performing specified shift operations on an N-bit operand and providing an N-bit result in a processing unit of a computer system wherein the processing unit executes instructions including a first subset of instructions, called m-bit shift instructions, and a second subset of instructions, called N-bit shift instructions, the input operand and the result each has a most significant bit designated bit(N-1) and a least significant bit designated bit(0), and N is greater than m, the shift unit comprising: an N-bit shifter having an N-bit input for receiving an N-bit entity to be shifted, an N-bit output for providing an N-bit entity resulting from a shift operation, a control input for specifying a direction and a designated number of bit positions S for a shift operation, a first shift input terminal for receiving a value to replace bit(N-1) for right shifts, and a second input terminal for receiving a value to replace bit(0) for left shifts; a first data selector having a first (N-m)-bit input coupled to receive bit(N-1) through bit(m) of said N-bit operand, a second input receiving (N-m) 0's, and an output coupled to bit positions bit(N-1) through bit(m) of said input of said shifter; said first data selector passing bit(N-1) through bit(m) of said N-bit operand for said N-bit shift instructions, and passing said (N-m) 0's for an instruction, designated an m-bit logical right shift, if the designated number of bit positions is non-zero; bit positions bit(m-1) through bit(0) of said input of said shifter being coupled to receive bit(m-1) through bit(0) of said N-bit operand; a second data selector having a first input coupled to receive bit(N-1) of said N-bit input operand, a second input receiving a 0, and an output coupled to said first shift input of said shifter; said second data selector passing bit(N-1) of said N-bit input operand for instructions, designated m-bit and N-bit arithmetic right shifts, and passing said 0 for instructions, designated m-bit and N-bit logical right shifts; said second shift input being coupled to receive a 0; a third data selector having a first input coupled to bit positions bit(N-1) through bit(m) of said output of said shifter, a second input coupled to receive (N-m) copies of bit(m-1) of said output of said shifter, and an output for providing bit(N-1) through bit(m) of said N-bit result; said third data selector passing bit positions bit(N-1) through bit(N-m) of said output of said shifter for said N-bit shift instructions, and passing said (N-m) copies of bit(m-1) of said output of said shifter for instructions, designated m-bit left shifts; bit positions bit(m-1) through bit(0) of said output of said shifter being coupled to provide bit(m-1) through bit(0) of said N-bit result.

Claim 16

A shift unit for performing specified shift operations on an N-bit operand and providing an N-bit result in a processing unit of a computer system wherein the processing unit executes instructions including a first subset of instructions, called m-bit shift instructions, and a second subset of instructions, called N-bit shift instructions, the input operand and the result each has a most significant bit designated bit(N-1) and a least significant bit designated bit(0), and N is greater than m, the shift unit comprising: an N-bit shifter having an N-bit input for receiving an N-bit entity to be shifted, an N-bit output for providing an N-bit entity resulting from a shift operation, a control input for specifying a direction and a designated number of bit positions S for a shift operation, a first shift input terminal for receiving a value to replace bit(N-1) for right shifts, and a second input terminal for receiving a value to replace bit(0) for left shifts; a first data selector having a first (N-m)-bit input coupled to receive bit(N-1) through bit(m) of said N-bit operand, a second input receiving (N-m) 0's, and an output coupled to bit positions bit(N-1) through bit(m) of said input of said shifter; said first data selector passing bit(N-1) through bit(m) of said N-bit operand for said N-bit shift instructions, and passing said (N-m) 0's for an instruction, designated an m-bit logical right shift, if the designated number of bit positions is non-zero; bit positions bit(m-1) through bit(0) of said input of said shifter being coupled to receive bit(m-1) through bit(0) of said N-bit operand; a second data selector having a first input coupled to receive bit(N-1) of said N-bit input operand, a second input receiving a 0, and an output coupled to said first shift input of said shifter; said second data selector passing bit(N-1) of said N-bit input operand for instructions, designated m-bit and N-bit arithmetic right shifts, and passing said 0 for instructions, designated m-bit and N-bit logical right shifts; said second shift input being coupled to receive a 0; a third data selector having a first input coupled to bit positions bit(N-1) through bit(m) of said output of said shifter, a second input coupled to receive (N-m) copies of bit(m-1) of said output of said shifter, and an output for providing bit(N-1) through bit(m) of said N-bit result; said third data selector passing bit positions bit(N-1) through bit(N-m) of said output of said shifter for said N-bit shift instructions, and passing said (N-m) copies of bit(m-1) of said output of said shifter for instructions, designated m-bit left shifts; bit positions bit(m-1) through bit(0) of said output of said shifter being coupled to provide bit(m-1) through bit(0) of said N-bit result.

Claim 17

A shift unit for performing specified shift operations on an N-bit operand and providing an N-bit result in a processing unit of a computer system wherein the processing unit executes instructions including a first subset of instructions, called m-bit shift instructions, and a second subset of instructions, called N-bit shift instructions, the input operand and the result each has a most significant bit designated bit(N-1) and a least significant bit designated bit(0), and N is greater than m, the shift unit comprising: an N-bit shifter having an N-bit input for receiving an N-bit entity to be shifted, an N-bit output for providing an N-bit entity resulting from a shift operation, a control input for specifying a direction and a designated number of bit positions for a shift operation, a first shift input terminal for receiving a value to replace bit(N-1) for right shifts, and a second input terminal for receiving a value to replace bit(0) for left shifts means for communicating to bit positions bit(N-1) through bit(m) of said input of said shifter (a) bit(N-1) through bit(m) of said N-bit operand for said N-bit shift instructions, and (b) (N-m) 0's for an instruction, designated an m-bit logical right shift, if the designated number of bit positions is non-zero; bit positions bit(m-1) through bit(0) of said input of said shifter being coupled to receive bit(m-1) through bit(0) of said N-bit operand; means for communicating to said first shift input of said shifter (a) bit(N-1) of said N-bit input operand for instructions, designated m-bit and N-bit arithmetic right shifts, and (b) a 0 for instructions, designated m-bit and N-bit logical right shifts; said second shift input being coupled to receive a 0; and means for providing as bit(N-1) through bit(m) of said N-bit result (a) bit positions bit(N-1) through bit(m) of said output of said shifter for said N-bit shift instructions, and (b) (N-m) copies of bit(m-1) of said output of said shifter for instructions, designated m-bit left shifts; bit positions bit(m-1) through bit(0) of said output of said shifter being coupled to provide bit(m-1) through bit(0) of said N-bit result.

Claim 18

A shift unit for performing specified shift operations on an N-bit operand and providing an N-bit result in a processing unit of a computer system wherein the processing unit executes instructions including a first subset of instructions, called m-bit shift instructions, and a second subset of instructions, called N-bit shift instructions, the input operand and the result each has a most significant bit designated bit(N-1) and a least significant bit designated bit(0), and N is greater than m, the shift unit comprising: an N-bit shifter having an N-bit input for receiving an N-bit entity to be shifted, an N-bit output for providing an N-bit entity resulting from a shift operation, a control input for specifying a direction and a designated number of bit positions for a shift operation, a first shift input terminal for receiving a value to replace bit(N-1) for right shifts, and a second input terminal for receiving a value to replace bit(0) for left shifts means for communicating to bit positions bit(N-1) through bit(m) of said input of said shifter (a) bit(N-1) through bit(m) of said N-bit operand for said N-bit shift instructions, and (b) (N-m) 0's for an instruction, designated an m-bit logical right shift, if the designated number of bit positions is non-zero; bit positions bit(m-1) through bit(0) of said input of said shifter being coupled to receive bit(m-1) through bit(0) of said N-bit operand; means for communicating to said first shift input of said shifter (a) bit(N-1) of said N-bit input operand for instructions, designated m-bit and N-bit arithmetic right shifts, and (b) a 0 for instructions, designated m-bit and N-bit logical right shifts; said second shift input being coupled to receive a 0; and means for providing as bit(N-1) through bit(m) of said N-bit result (a) bit positions bit(N-1) through bit(m) of said output of said shifter for said N-bit shift instructions, and (b) (N-m) copies of bit(m-1) of said output of said shifter for instructions, designated m-bit left shifts; bit positions bit(m-1) through bit(0) of said output of said shifter being coupled to provide bit(m-1) through bit(0) of said N-bit result.

Claim 19

In a computer system including a processing unit and a memory subsystem, a processing unit comprising: a decoder, responsive to instructions fetched from the memory subsystem, for specifying corresponding operations wherein an m-bit entity on which operations can be performed has a most significant bit designated bit(m-1) and a least significant bit designated bit(0), an N-bit entity on which operations can be performed has a most significant bit designated bit(N-1) and a least significant bit designated bit(0), and N is greater than m; the instructions including a first subset of instructions, called m-bit instructions, and a second subset of instructions, called N-bit instructions, the m-bit instructions including a number of m-bit load instructions, a number of logical operation instructions, a number of m-bit shift instructions, and a number of m-bit addition instructions; the N-bit instructions including a number of N-bit load instructions, a number of N-bit shift instructions, and a number of N-bit addition instructions; a load circuit for performing specified load operations to retrieve entities up to N bits long from said memory subsystem, the load circuit having an associated load sign-extension circuit, controllably operable in response to at least one m-bit load instruction, but not in response to any N-bit load instructions, to convert a retrieved m-bit entity to an N-bit entity where bit(N-1) through bit(N-m) of the N-bit entity are all equal to the value of bit(m-1) of the m-bit entity and bit(m-1) through bit(0) of the N-bit entity are respectively equal to the values of bit(m-1) through bit(0) of the m-bit entity; a register file having a set of N-bit registers; a logic circuit for performing specified logical operations between two N-bit operands and providing an N-bit result; a shift circuit for performing specified shift operations on an N-bit operand and providing an N-bit result; a shift sign-extension circuit, controllably operable in response to at least one m-bit shift instruction, but not in response to any N-bit shift instructions, to change the N-bit result of a shift operation by setting bit(N-1) through bit(N-m) of the N-bit result equal to the value of bit(m-1) of the N-bit result; an adder for performing specified addition operations between a pair of N-bit operands and providing an N-bit result; an add sign-extension circuit, controllably operable when at least one m-bit addition instruction is executed but not when any N-bit addition instruction is executed, to require the N-bit result of an operation in said adder have bit(N-1) through bit(m) of the N-bit result equal to the value of bit(m-1) of the result; N-bit data paths from output ports of each of said load circuit, said logic circuit, said shift circuit, and said adder to said register file; and N-bit data paths from said register file to input ports of each of said logic circuit, said shift circuit, and said adder.

Claim 20

In a computer system including a processing unit and a memory subsystem, a processing unit comprising: a decoder, responsive to instructions fetched from the memory subsystem, for specifying corresponding operations wherein an m-bit entity on which operations can be performed has a most significant bit designated bit(m-1) and a least significant bit designated bit(0), an N-bit entity on which operations can be performed has a most significant bit designated bit(N-1) and a least significant bit designated bit(0), and N is greater than m; the instructions including a first subset of instructions, called m-bit instructions, and a second subset of instructions, called N-bit instructions, the m-bit instructions including a number of m-bit load instructions, a number of logical operation instructions, a number of m-bit shift instructions, and a number of m-bit addition instructions; the N-bit instructions including a number of N-bit load instructions, a number of N-bit shift instructions, and a number of N-bit addition instructions; a load circuit for performing specified load operations to retrieve entities up to N bits long from said memory subsystem, the load circuit having an associated load sign-extension circuit, controllably operable in response to at least one m-bit load instruction, but not in response to any N-bit load instructions, to convert a retrieved m-bit entity to an N-bit entity where bit(N-1) through bit(N-m) of the N-bit entity are all equal to the value of bit(m-1) of the m-bit entity and bit(m-1) through bit(0) of the N-bit entity are respectively equal to the values of bit(m-1) through bit(0) of the m-bit entity; a register file having a set of N-bit registers; a logic circuit for performing specified logical operations between two N-bit operands and providing an N-bit result; a shift circuit for performing specified shift operations on an N-bit operand and providing an N-bit result; a shift sign-extension circuit, controllably operable in response to at least one m-bit shift instruction, but not in response to any N-bit shift instructions, to change the N-bit result of a shift operation by setting bit(N-1) through bit(N-m) of the N-bit result equal to the value of bit(m-1) of the N-bit result; an adder for performing specified addition operations between a pair of N-bit operands and providing an N-bit result; an add sign-extension circuit, controllably operable when at least one m-bit addition instruction is executed but not when any N-bit addition instruction is executed, to require the N-bit result of an operation in said adder have bit(N-1) through bit(m) of the N-bit result equal to the value of bit(m-1) of the result; N-bit data paths from output ports of each of said load circuit, said logic circuit, said shift circuit, and said adder to said register file; and N-bit data paths from said register file to input ports of each of said logic circuit, said shift circuit, and said adder.

Claim 21

The processing unit of claim 13 wherein N=64 and m=32.

Claim 22

The processing unit of claim 13 wherein: said adder has carryouts from bit positions bit(m-1) and bit(m-2), and provides a signal indicating a condition, referred to as m-bit overflow, in said adder where said carryouts from bit positions bit(m-1) and bit(m-2) differ; and said add sign-extension circuit operates to change the N-bit result of an operation in said adder by setting bit(N-1) through bit(m) of the N-bit result equal to the value of bit(m-1) of the N-bit result only if an m-bit overflow has occurred.

Claim 23

The processing unit of claim 1 wherein: said adder has carryouts from bit positions bit(m-1) and bit(m-2), and provides a signal indicating a condition, referred to as m-bit overflow, in said adder where said carryouts from bit positions bit(m-1) and bit(m-2) differ; said m-bit addition instructions include ADD and SUB instructions that trap on an m-bit overflow and produce no result, and ADDU and SUBU instructions that do not trap on an m-bit overflow and do produce a result; and said add sign-extension circuit operates to change the N-bit result of an operation in said adder by setting bit(N-1) through bit(m) of the N-bit result equal to the value of bit(m-1) of the N-bit result if an m-bit overflow has occurred in the course of executing said ADDU and SUBU instructions and does not operate in connection with executing said ADD and SUB instructions.

Claim 24

The processing unit of claim 1 wherein: said adder has carryouts from bit positions bit (N-1) and bit (N-2), and provides a signal indicating a condition, referred to as N-bit overflow, where the carryouts from bit positions bit (N-1) and bit (N-2) differ; and said N-bit addition instructions include DADD and DSUB instructions that trap on an N-bit overflow and produce no result, and DADDU and DSUBU instructions that do not trap on an N-bit overflow and do produce a result.