# US 6,732,259 B1 · Processor having a conditional branch extension of an instruction set architecture

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 6,732,259 B1
- Document ID: US6732259B1
- Status: Granted
- Kind: B1
- Application number: 09/364,789
- Filed: 1999-07-30
- Granted: 2004-05-04
- Earliest claimed: 1999-07-30
- Assignees: MIPS Technologies, Inc.
- Inventors: Radhika Thekkath; G. Michael Uhler; Ying-wai Ho; Chandlee B. Harrell
- Web page: https://patentgrep.com/patent/US6732259B1

## CPC

- G06F 9/30014: {with variable precision}
- G06F 9/30094: {Condition code generation, e.g. Carry, Zero flag}
- G06F 9/3885: {using a plurality of independent parallel functional units}
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

A processor having a conditional branch extension of an instruction set architecture which incorporates a set of high performance floating point operations. The instruction set architecture incorporates a variety of data formats including single precision and double precision data formats, as well as the paired-single data format that allows two simultaneous operations on a pair of operands. The extension includes instructions directed to branching if, for example, either one of two condition codes is false or true, if any of three condition codes are false or true, or if any one of four condition codes are false or true.

## Claims

### Claim 1

In a general purpose processor, a method for performing computer graphics calculations, said method comprising:
representing a vertex in a computer graphics image with coordinates;
transforming said coordinates;
performing a comparison of a plurality of transformed coordinates with a value representing a plurality of edges of a specified view volume;
setting a plurality of condition code bits to one or more specific states to indicate results of said comparison; and
performing a conditional branch instruction based on one of a plurality of conditions,
wherein said conditions are based on said plurality of condition code bits,
wherein at least a portion of said method is performed using one or more instructions that are part of a group of instructions collectively tailored for geometry processing.

### Claim 2 (depends on 1)

The method for performing computer graphics calculations as recited in claim 1 further comprising:
a) computing a branch target address;
b) jumping to said branch target address; and
c) executing an instruction at said branch target address;
wherein said computing, jumping and executing are dependent upon condition code state.

### Claim 3 (depends on 2)

The method for performing computer graphics calculations as recited in claim 2 further comprising:
comparing one of said plurality of condition code bits to an indicator bit in said conditional branch instruction to determine if said one condition code bit is set to a pre-determined state; and
performing a)-c) if said one condition code bit is set to said pre-determined state.

### Claim 4 (depends on 2)

The method for performing computer graphics calculations as recited in claim 2 further comprising:
setting a first condition code bit to a predetermined state if a graphics primitive crosses a first edge of said specified view volume;
setting a second condition code bit to said predetermined state if said graphics primitive crosses a second edge of said specified view volume;
performing a)-c) if either of said first condition code bit or said second condition code bit is set to said pre-determined state.

### Claim 5 (depends on 4)

The method for performing computer graphics calculations as recited in claim 4 further comprising:
comparing said first condition code bit and said second condition code bit to an indicator bit in said conditional branch instruction to determine if said first condition code bit or said second condition code bit is set to said pre-determined state.

### Claim 6 (depends on 2)

The method for performing computer graphics calculations as recited in claim 2 further comprising:
setting a first condition code bit;
setting a second condition code bit;
setting a third condition code bit;
setting a fourth condition code bit;
performing a)-c) if any of said first condition code bit, said second condition code bit, said third condition code bit, or said fourth condition code bit is set to said pre-determined state.

### Claim 7 (depends on 6)

The method for performing computer graphics calculations as recited in claim 6 further comprising:
comparing said first condition code bit, said second condition code bit, said third condition code bit, and said fourth condition code bit to an indicator bit in said conditional branch instruction to determine if said first condition code bit, said second condition code bit, said third condition code bit or said fourth condition code bit is set to said pre-determined state.

### Claim 8 (depends on 2)

The method for performing computer graphics calculations as recited in claim 2 further comprising:
adding an offset to an address of an instruction following said conditional branch instruction in a branch delay slot to form a target address;
executing an instruction in said delay slot; and
branching to said target address after execution of said instruction in said delay slot.

### Claim 9 (depends on 2)

The method for performing computer graphics calculations as recited in claim 2, wherein steps a)-c) are performed only if either one of two specified condition code bits is set in a false state.

### Claim 10 (depends on 2)

The method for performing computer graphics calculations as recited in claim 2, wherein steps a)-c) are performed only if either one of two specified condition code bits is set in a true state.

### Claim 11 (depends on 2)

The method for performing computer graphics calculations as recited in claim 2, wherein steps a)-c) are performed only if any one of three specified condition code bits is set in a false state.

### Claim 12 (depends on 2)

The method for performing computer graphics calculations as recited in claim 2, wherein steps a)-c) are performed only if any one of three specified condition code bits is set in a true state.

### Claim 13 (depends on 2)

The method for performing computer graphics calculations as recited in claim 2, wherein steps a)-c) are performed only if any one of four specified condition code bits is set in a false state.

### Claim 14 (depends on 2)

The method for performing computer graphics calculations as recited in claim 2, wherein steps a)-c) are performed only if any one of four specified condition code bits is set in a true state.

### Claim 15

A general purpose processor for computer graphics calculations, said processor comprising:
a bus;
an instruction dispatch unit coupled to said bus, said instruction dispatch unit for dispatching instructions to a floating point unit, wherein one or more of said instructions are part of a group of instructions collectively tailored for geometry processing; and
said floating point unit coupled to said bus, said floating point unit for executing said instructions to implement a method for performing computer graphics calculations, said method comprising:
representing a vertex in a computer graphics image with coordinates;
transforming said coordinates;
performing a comparison of a plurality of transformed coordinates with a value representing a plurality of edges of a specified view volume;
setting a plurality of condition code bits to one or more specific states to indicate results of said comparison; and
performing a conditional branch instruction based on one of a plurality of conditions,
wherein said conditions are based on said plurality of condition code bits.

### Claim 16 (depends on 15)

The processor of claim 15 wherein said method for performing computer graphics calculations further comprises:
jumping to a branch target address if any one of said plurality of condition code bits is set to a pre-determined state.

### Claim 17 (depends on 16)

The processor of claim 16 wherein said method for performing computer graphics calculations further comprises:
comparing said plurality of condition code bits to an indicator bit in said conditional branch instruction to determine if any one of said plurality of condition code bits is set to said pre-determined state.

### Claim 18 (depends on 15)

The processor of claim 15 wherein said method for performing computer graphics calculations further comprises:
adding an offset to an address of an instruction following said conditional branch instruction in a branch delay slot to form a target address;
executing an instruction in said delay slot; and
branching to said target address after execution of said instruction in said delay slot.

### Claim 19 (depends on 15)

The processor of claim 15 wherein the conditional branch instruction is a BC1ANY2F instruction.

### Claim 20 (depends on 15)

The processor of claim 15 wherein the conditional branch instruction is a BC1ANY2T instruction.

### Claim 21 (depends on 15)

The processor of claim 15 wherein the conditional branch instruction is a BC1ANY4F instruction.

### Claim 22 (depends on 15)

The processor of claim 15 wherein the conditional branch instruction is a BC1ANY4T instruction.

### Claim 23

In a general purpose processor, a method comprising:
providing a plurality of bits set to one or more states, said bits being held in a storage device within said general processor;
processing said plurality of bits or any combination of said plurality of bits to generate a combined bit; and
performing a conditional branch instruction based on said combined bit,
wherein at least a portion of said method is performed using one or more instructions that are part of a group of instructions collectively tailored for geometry processing.

### Claim 24 (depends on 23)

The method of claim 23 wherein said processing comprises logically combining said plurality of bits to determine whether a branching condition is satisfied.

### Claim 25 (depends on 24)

The method of claim 24 wherein said logically combining is effected through a logical OR process.

### Claim 26 (depends on 24)

The method of claim 24 wherein said logically combining is effected through a logical AND process.

### Claim 27 (depends on 25)

The method of claim 25 wherein said logically combining is effected through circuitry representative of a boolean equation.

### Claim 28 (depends on 27)

The method of claim 27 wherein said circuitry is provided in a programmable logic array.

### Claim 29

In a system having a memory coupled to a general purpose processor, a method comprising:
storing a first instruction and second instruction in said memory, wherein said first instruction is formatted to operate on a plurality of operands;
processing said first instruction in said general purpose processor, wherein said first instruction operates on said plurality of operands to perform a plurality of magnitude compare operations in parallel;
setting a plurality of bits to one or more specific states in response to said magnitude compare operations; and
processing said second instruction in said processor, wherein said second instruction responds to said plurality of bits or a subset of said plurality of bits to selectively initiate a branch operation,
wherein said first and second instructions are part of a group of instructions collectively tailored for geometry processing.

### Claim 30 (depends on 29)

The method of claim 29 wherein said processing comprises:
combining said plurality of bits to generate a combined bit;
comparing said combined bit to an indicator bit in said second instruction to determine if said combined bit is set to a predetermined state; and
jumping to a branch target address if said combined bit is set to said predetermined state.

### Claim 31 (depends on 30)

The method of claim 30 wherein said combined bit is set to said predetermined state when any one of said plurality of bits is set to said predetermined state.

### Claim 32 (depends on 30)

The method of claim 30 wherein said combined bit is set to said predetermined state when two or more of said plurality of bits is set to said predetermined state.

### Claim 33

In a system having a memory coupled to a general purpose processor, a method comprising:
storing a first instruction, second instruction and third instruction in said memory;
processing said first instruction in said general purpose processor, wherein said processor operates on a first operand to perform a first magnitude compare operation and generate a first result bit;
processing said second instruction in said general purpose processor, wherein said processor operates on a second operand to perform a second magnitude compare operation and generate a second result bit; and
processing said third instruction in said general purpose processor, wherein said processor jumps to a branch target address if either of said first or second result bit is set to a predetermined state,
wherein said first, second and third instructions are part of a group of instructions collectively tailored for geometry processing.

### Claim 34

A computer program product comprising a computer-readable medium having a plurality of instructions stored thereon, the plurality of instructions for enabling a general purpose processor to perform certain operations, wherein the plurality of instructions include:
a first instruction to enable the general purpose processor to perform a plurality of magnitude compare operations in parallel and set a plurality of result bits to one or more specific states; and
a second instruction to enable the general processor to jump to a branch target address in response to said plurality of result bits,
wherein said first and second instructions are part of a group of instructions collectively tailored for geometry processing.

### Claim 35 (depends on 1)

The method of claim 1 wherein said performing a comparison of a plurality of transformed coordinates is carried out in parallel.

### Claim 36 (depends on 1)

The method of claim 1 wherein said coordinates and said plurality of transformed coordinates are in a paired-single data format.

### Claim 37 (depends on 1)

The method of claim 1 wherein said condition branch instruction is a BC1ANY2F instruction.

### Claim 38 (depends on 1)

The method of claim 1 wherein said condition branch instruction is a BC1ANY2T instruction.

### Claim 39 (depends on 1)

The method of claim 1 wherein said condition branch instruction is a BC1ANY4F instruction.

### Claim 40 (depends on 1)

The method of claim 1 wherein said condition branch instruction is a BC1ANY4T instruction.

### Claim 41 (depends on 2)

The method of claim 2 further comprising:
logically combining said plurality of condition code bits to generate a combined bit;
comparing said combined bit to an indicator bit in said conditional branch instructions to determine if said combined bit is set to said predetermined state; and
performing a)-c) if said combined bit is set to said predetermined state.

### Claim 42 (depends on 4)

The method of claim 4 further comprising:
logically combining said first condition code bit and said second condition code bit to generate a combined bit; and
comparing said combined bit to an indicator bit in said conditional branch instruction to determine if said first condition code bit or said second condition code bit is set to said predetermined state.

### Claim 43 (depends on 6)

The method of claim 6 further comprising:
logically combining said first, second, third and fourth condition code bits to generate a combined bit; and
comparing said combined bit to an indicator bit in said conditional branch instruction to determine if said first, second, third or fourth condition code bit is set to said predetermined state.

### Claim 44 (depends on 15)

The processor of claim 15 wherein said performing a comparison of a plurality of transformed coordinates is carried out in parallel.

### Claim 45 (depends on 15)

The processor of claim 15 wherein said coordinates and said plurality of transformed coordinates are in a paired-single data format.

### Claim 46 (depends on 16)

The processor of claim 16 wherein said method for performing computer graphics calculations further comprises:
logically combining said plurality of condition code bits to generate a combined bit; and
comparing said combined bit to an indicator bit in said conditional branch instruction to determine if any one of said plurality of condition code bits is set to a predetermined state.

### Claim 47 (depends on 23)

The method of claim 23 wherein said performing comprises:
jumping to a branch target address if said combined bit is set to a predetermined state.

### Claim 48 (depends on 47)

The method of claim 47 wherein said combined bit is set to said predetermined state when any one of said plurality of bits is set to said predetermined state.

### Claim 49 (depends on 48)

The method of claim 48 wherein said combined bit is set to said predetermined state when two or more of said plurality of bits is set to said predetermined state.

### Claim 50 (depends on 23)

The method of claim 23 wherein said performing a condition branch instruction is carried out in a single clock cycle.

### Claim 51

An extension to an instruction set architecture, comprising:
a plurality of extension instructions for reducing the number of instructions otherwise required to perform a given operation, said plurality of extension instructions enabling a general purpose processor to accelerate the speed at which said general purpose processor performs, said plurality of extension instructions capable of using a plurality of data formats, said plurality of data formats including single precision, double precision, and paired-single data formats, wherein said paired-single data format provides two simultaneous operations on a pair of operands, said plurality of extension instructions including:
a first instruction to enable said general purpose processor to perform a plurality of magnitude compare operations in parallel and set a plurality of result bits to one or more specific states; and
a second instruction to enable said general purpose processor to jump to a branch target address based on a subset of one of consecutive result bits and non-consecutive result bits used to determine whether said second instruction is performed.

### Claim 52 (depends on 1)

The method of claim 1, wherein said plurality of conditions includes:
whether any one of said plurality of condition code bits is true, whether any one of said plurality of condition code bits is false, whether a combination of said plurality of condition code bits is true, whether a combination of said plurality of condition code bits is false, whether a subset of said plurality of condition code bits is true, whether a subset of said plurality of condition code bits is false, whether a combination of a subset of said plurality of condition code bits is true, and whether a combination of a subset of said plurality of condition code bits is false.

### Claim 53 (depends on 15)

The method of claim 15, wherein said plurality of conditions includes:
whether any one of said plurality of condition code bits is true, whether any one of said plurality of condition code bits is false, whether a combination of said plurality of condition code bits is true, whether a combination of said plurality of condition code bits is false, whether a subset of said plurality of condition code bits is true, whether a subset of said plurality of condition code bits is false, whether a combination of a subset of said plurality of condition code bits is true, and whether a combination of a subset of said plurality of condition code bits is false.

### Claim 54 (depends on 1)

The method of claim 1, wherein said one or more instructions are in addition to an instruction set architecture.

### Claim 55 (depends on 54)

The method of claim 54, wherein said instruction set architecture is a general purpose instruction set architecture.

### Claim 56 (depends on 1)

The method of claim 1, wherein said one or more instructions are from an application specific extension to an instruction set architecture.

### Claim 57 (depends on 1)

The method of claim 1, wherein said one or more instructions accelerate the speed at which said general purpose processor performs said computer graphics calculations.

### Claim 58 (depends on 15)

The method of claim 15, wherein said one or more of said instructions are in addition to an instruction set architecture.

### Claim 59 (depends on 58)

The method of claim 58, wherein said instruction set architecture is a general purpose instruction set architecture.

### Claim 60 (depends on 15)

The method of claim 15, wherein said one or more of said instructions are from an application specific extension to an instruction set architecture.

### Claim 61 (depends on 15)

The method of claim 15, wherein said one or more of said instructions accelerate the speed at which said general purpose processor performs said computer graphics calculations.

### Claim 62 (depends on 23)

The method of claim 23, wherein said one or more instructions are in addition to an instruction set architecture.

### Claim 63 (depends on 62)

The method of claim 62, wherein said instruction set architecture is a general purpose instruction set architecture.

### Claim 64 (depends on 23)

The method of claim 23, wherein said one or more instructions are from an application specific extension to an instruction set architecture.

### Claim 65 (depends on 23)

The method of claim 23, wherein said one or more instructions accelerate the speed at which said general purpose processor performs said method.

### Claim 66 (depends on 29)

The method of claim 29, wherein said first and second instructions are in addition to an instruction set architecture.

### Claim 67 (depends on 66)

The method of claim 66, wherein said instruction set architecture is a general purpose instruction set architecture.

### Claim 68 (depends on 29)

The method of claim 29, wherein said first and second instructions are from an application specific extension to an instruction set architecture.

### Claim 69 (depends on 29)

The method of claim 29, wherein said first and second instructions accelerate the speed at which said general purpose processor performs said method.

### Claim 70 (depends on 33)

The method of claim 33, wherein said first, second and third instructions are in addition to an instruction set architecture.

### Claim 71 (depends on 70)

The method of claim 70, wherein said instruction set architecture is a general purpose instruction set architecture.

### Claim 72 (depends on 33)

The method of claim 33, wherein said first, second and third instructions are from an application specific extension to said instruction set architecture.

### Claim 73 (depends on 33)

The method of claim 33, wherein said first, second and third instructions reduce the number of instructions otherwise required to perform a given operation, thereby accelerating the speed at which said general purpose processor performs said method.

### Claim 74 (depends on 34)

The computer program product of claim 34, wherein said first and second instructions are in addition to an instruction set architecture.

### Claim 75 (depends on 74)

The computer program product of claim 74, wherein said instruction set architecture is a general purpose instruction set architecture.

### Claim 76 (depends on 34)

The computer program product of claim 34, wherein said first and second instructions are from an application specific extension to said instruction set architecture.

### Claim 77 (depends on 34)

The computer program product of claim 34, wherein said first and second instructions accelerate the speed at which said general purpose processor performs.

### Claim 78 (depends on 34)

The computer program product of claim 34, wherein said first and second instructions are in one of a plurality of formats, including single precision, double precision, and paired-single data format, wherein said instructions for said paired-single data format provide two simultaneous operations on a pair of operands.
