Dual mode SIMD/MIMD processor providing reuse of MIMD instruction memories as data memories when operating in SIMD mode

• US 5,239,654 A
• Filed: 11/17/1989
• Issued: 08/24/1993
• Est. Priority Date: 11/17/1989
• Status: Expired due to Term

• Current Assignee: Texas Instruments, Inc.
• Original Assignee: Texas Instruments, Inc.
• Inventors: Ing-Simmons, Nicholas K., Guttag, Karl M., Balmer, Keith, Gove, Robert J.
• Primary Examiner(s): Anderson, Lawrence E.
• Assistant Examiner(s): ELLIS, RICHARD L
• Application Number:
  US07/437,851
• Time in Patent Office: 1,376 Days
• Field of Search: 395/325, 395/800
• US Class Current: 712/20
• CPC Class Codes: G06F 13/376 using a contention resolving method, e.g. collision detection, collision avoidance
• CPC Class Codes: G06F 15/17375 One dimensional, e.g. linear array, ring
• CPC Class Codes: G06F 15/8007 single instruction multiple data [SIMD] multiprocessors
• CPC Class Codes: G06F 9/30014 with variable precision
• CPC Class Codes: G06F 9/30025 Format conversion instructions, e.g. Floating-Point to Integer, decimal conversion
• CPC Class Codes: G06F 9/30032 Movement instructions, e.g. MOVE, SHIFT, ROTATE, SHUFFLE
• CPC Class Codes: G06F 9/30036 Instructions to perform operations on packed data, e.g. vector, tile or matrix operations
• CPC Class Codes: G06F 9/30094 Condition code generation, e.g. Carry, Zero flag
• CPC Class Codes: G06F 9/3842 Speculative instruction execution

Abstract

A multi-processor system and method arranged, in one embodiment, as an image and graphics processor. The multiprocessor system includes several individual processors all having communication links to several memories. Additional instruction memories are dedicated individually as cache memories to particular processors so that the processors can function in the multiple instruction, multiple data (MIMD) mode. When the processors function in the single instruction, multiple data mode (SIMD) the dedicated memories are reassigned for access by all of the processors for data. A crossbar switch serves to establish the processor memory links. The entire image processor, including the individual processors, the crossbar switch and the memories, is contained on a single silicon chip.

Claims

1. A multi-processor system having a single instruction multiple data (SIMD) mode and a multiple instruction multiple data (MIMD) mode, said multi-processor system comprising:

a plurality of processors, each processor generating an address request for a memory access;

a plurality of memories each having a unique addressable space, said plurality of memories including an instruction memory corresponding to each of said processors for supplying corresponding instructions and a plurality of data memories;

a switch matrix connected to each of said plurality of processors and each of said plurality of memories, said switch matrix receiving address requests from said processors for memory access and

when operating in the multiple instruction multiple data (MIMD) mode said switch matrix;

(1) granting access to a processor generating an address request if said address request is within said unique address space of said corresponding instruction memory for supplying corresponding instructions, (2) granting access to a processor generating an address request if said address request is within said unique address space of any of said plurality of memories except said instruction memories, and (3) prohibiting access to a processor generating an address request if said address request is within said unique address space of an instruction memory corresponding to another processor, and

when operating in a the single instruction multiple data (SIMD) mode said switch matrix;

(1) granting access to a first processor generating an address request if said address request is within said unique address space of said corresponding instruction memory for supplying corresponding instructions to all of said processors, (2) granting access to a processor generating an address request if said address request is within said unique address space of any of said plurality of memories except said instruction memory corresponding to said first processor, and (3) prohibiting access to a processor generating an address request if said address request is within said unique address space of said instruction memory corresponding to said first processor.

2. The multi-processor system claimed in claim 1, wherein:

said plurality of processors, said plurality of memories and said switch matrix are formed on a single integrated circuit.

3. A multi-processor system having a single instruction multiple data (SIMD) mode and a multiple instruction multiple data (MIMD) mode, said multi-processor system comprising:

a plurality of processors, each processors having a data port and an instruction memory port and operating from instructions provided to said instruction port;

a plurality of memories including at least one data memory corresponding to each of said processors and an instruction memory corresponding to each of said processors;

a switch matrix connected to each of said plurality of processors and each of said plurality of memories, said switch matrix

when operating in the multiple instruction multiple data (MIMD) mode said switch matrix;

(1) connecting between said instruction port of each processor and said corresponding instruction memory for supplying corresponding instructions to each of said processors, (2) connecting between said data port of each of said processors and each of said plurality of memories except said instruction memories, and (3) prohibiting connection between said data port of each of said processors and said instruction memories, and

when operating in the single instruction multiple data (SIMD) mode said switch matrix;

(1) permitting connecting between said instruction port of a first processor of said plurality of processors and said corresponding instruction memory for supplying a single instruction to all of said processors, (2) connecting between said data port of each of said plurality of processors and each of said plurality of memories including said instruction memories corresponding to processors other than said first processor, and (3) prohibiting connection between said data port of each of said processors and said instruction memory corresponding to said first processor.

4. The multi-processor system claimed in claim 3, wherein:

said switch matrix includes

a set of first links, each first link connected to a corresponding memory,

a set of second links, each second line connected to said data port of a corresponding processor,

a third link having a plurality of adjacent sections equal in number to the number of said processors, each section of said third link connected to said instruction port of a corresponding processor,

a plurality of buffers disposed between adjacent sections of said third link connecting said adjacent sections of said third link when in the single instruction multiple data (SIMD) mode, and isolating said adjacent sections of said third link when in the multiple instructions multiple (MIMD) mode, and

a plurality of crosspoints disposed at intersections between said first links and said second links and at intersections between said first links and said sections of said third link, said plurality of crosspoints including (1) a first crosspoint disposed at the intersection of said section of said third link connected to said instruction port of said first processor and said first link connected to said instruction memory corresponding to said first processor, said first crosspoint always enabled to provide connection, (2) a set of second crosspoints disposed at the intersections of said sections of said third link connected to said instruction port of said processors other than said first processor and said first link connected to said corresponding instruction memory, said second crosspoints disabled to prohibit connection when operating in the single instruction multiple data (SIMD) mode and enabled to provide connection when operating in the multiple instruction multiple data (MIMD) mode, and (3) a set of third crosspoints disposed at the intersections of said first links connected to said instruction memories other than said instruction memory corresponding to said first processor and said second links connected to said data ports of said plurality of processors, said set of third crosspoints disabled to prohibit connection when operating in the multiple instruction single data (MIMD) mode and enabled to provide connection when operating in single instruction multiple data (SIMD) mode.

5. The multi-processor system claimed in claim 3, further comprising:

a SIMD/MIMD mode register connected to said switch matrix and storing therein an indication of either the single instruction multiple data (SIMD) mode or the multiple instruction multiple data (MIMD) mode.

6. The multi-processor system claimed in claim 5, further comprising:

an external memory for storing instructions for each of said processors;

each processor further includes

a program counter having stored therein an address of a next instruction, and

a cache control logic connected to said program counter, said cache control logic

determining if an instruction corresponding to the address stored in said program counter is stored in said corresponding instruction memory,

accessing said corresponding instruction memory via said switch matrix to recall said instruction if the instruction corresponding to the address stored in said program counter is stored in said corresponding instruction memory when said SIMD/MIMD register indicates the multiple instruction multiple data (MIMD) mode,

requesting an instruction corresponding to the address stored in said program counter from said external memory if the instruction is not stored in said corresponding instruction memory when said SIMD/MIMD register indicates the multiple instruction multiple data (MIMD) mode,

said cache control logic of each processor except said first processor disabled when said SIMD/MIMD register indicates the single instruction multiple data (SIMD) mode.

7. The multi-processor system claimed in claim 3, wherein:

said plurality of processors, said plurality of memories and said switch matrix are formed on a single integrated circuit.

8. A multi-processor system operable in either a single instruction multiple data (SIMD) mode or in a multiple system comprising:

a SIMD/MIMD mode register storing therein an indication of either the single instruction multiple data (SIMD) mode or the multiple instruction multiple data (MIMD) mode for each of said plurality of processors;

a plurality of processors, each processors having a data port and an instruction portion and operating from instructions provided to said instruction port for controlling a process including movement of data to and from said data port;

a plurality of memories, said plurality of memories including at least one data memory corresponding to each of said processors and an instruction memory corresponding to each of said processors;

a switch matrix connected to said SIMD/MIMD mode register, to each of said plurality of processors and each of said plurality of memories, said switch matrix including

a set of first links, each first link connected to a corresponding memory,

a set of second links, each second line connected to said data ports of a corresponding processor,

a third link having a plurality of sections equal in number to the number of said processors, each section connected to said instruction port of a corresponding one of said processors,

a plurality of buffers disposed between adjacent sections of said third link forming a serial chain from a first processor to a last processor, each buffer connecting said adjacent sections of said third link when said SIMD/MIMD mode register indicates the single instruction multiple data (SIMD) mode for a corresponding processor, and isolating said adjacent sections of said third link when said SIMD/MIMD mode register indicates the multiple instruction multiple data (MIMD) mode for the corresponding processor, and

a plurality of crosspoints disposed at intersections between said first links and said second links at intersections between said first links and said sections of said third link, said plurality of crosspoints including (1) a first crosspoint disposed at the intersection of said section of said third link connected to said instruction port of a predetermined first processor and said first link connected to said corresponding instruction memory which is always enabled to provide connection, (2) a set of second crosspoints disposed at the intersection of said section of said third link connected to said instruction port of processors other than said predetermined first processor and said respective first link connected to said corresponding instruction memories which are disabled to prohibit connection when said SIMD/MIMD register indicates the single instruction multiple data (SIMD) mode for said processor corresponding to said instruction memory and enabled to provide connection when said SIMD/MIMD register indicates the multiple instruction multiple data (MIMD) mode for said processor corresponding to said instruction memory, (3) a set of third crosspoints disposed at the intersection of said first links connected to said instruction memories and said second links connected to said data ports of said plurality of processors, said set of third crosspoints disabled to prohibit connection when said SIMD/MIMD register indicates the single instruction multiple data (SIMD) mode for said processor corresponding to said instruction memory and enabled to provide connection when said SIMD/MIMD register indicates the multiple instruction multiple data (MIMD) mode for said processor corresponding to said instruction memory;

whereby said instruction port of each processor is connected to said instruction memory corresponding to the first processor in a serial chain in the single instruction multiple data (SIMD) mode, said instruction port of each processor in the multiple instruction multiple data (MIMD) mode is connected to said corresponding instruction memory, and access to said instruction memories via said data ports of said plurality of processors is prohibited when the corresponding processor is in the single instruction multiple data (SIMD) mode and permitted when the corresponding processor is in the multiple instruction multiple data (MIMD) mode.

9. The multiple-processor system claimed in claim 8, wherein:

each processor includes

a program counter having stored therein an address of a next instruction, and

an instruction fetch circuit for supplying to said instruction port said address stored in said program counter for fetching an instruction stored at said address; and

said switch matrix wherein

said buffers pass signals unidirectionally whereby when said buffers connect said adjacent sections of said third link in the single instruction multiple data (SIMD) mode only said program counter of a first processor in a serial chain of processors in the single instruction multiple data (SIMD) mode accesses its corresponding instruction memory.

10. The multi-processor system claimed in claim 8, further comprising:

an external memory for storing instructions for each of said processors;

each processor further includes

a program counter having stored therein an address of a next instruction, and

a cache control logic connected to said program counter, said cache control logic

determining if an instruction corresponding to the address stored in said program counter is stored in said corresponding instruction memory,

accessing said corresponding instruction memory via said switch matrix to recall said instruction if the instruction corresponding to the address stored in said program counter is stored in said corresponding instruction memory when said SIMD/MIMD register indicates the multiple instruction multiple data (MIMD) mode,

requesting an instruction corresponding to the address stored in said program counter from said external memory if the instruction is not stored in said corresponding instruction memory when said SIMD/MIMD register indicates the multiple instruction multiple data (MIMD) mode,

said cache control logic of each processor that is not a first processor in a serial chain of processors in said the single instruction multiple data (SIMD) mode being disable when said SIMD/MIMD register indicates the single instruction multiple data (SIMD) mode.

11. The multi-processor system claimed in claim 8, wherein:

said SIMD/MIMD mode register, said plurality of processors, said plurality of memories and said switch matrix are formed on a single integrated circuit.

12. A computer implemented method of operating a plurality of processors in either a single instruction multiple data (SIMD) mode or in a multiple instruction multiple data (MIMD) mode in conjunction with a plurality of memories supplying data and instructions to the processors, the method comprising the computer implemented steps of:

storing an indication of either the single instruction multiple data (SIMD) mode or the multiple instruction multiple data (MIMD) mode for each processor;

connecting through a switch matrix a first processor in a serial chain of processors operating in the single instruction multiple data (SIMD) mode to a corresponding instruction memory;

connecting through said switch matrix each processor in the serial chain of processors operating in the single instruction multiple data (SIMD) mode other than the first processor in the serial chain of processors operating in the single instruction multiple data (SIMD) mode to said instruction memory corresponding to the first processor in the serial chain of processors operating in the single instruction multiple data (SIMD) mode;

connecting each processor in the multiple instruction multiple data (MIMD) mode to a corresponding instruction memory so that each processor has its own corresponding construction memory;

prohibiting any other processor from accessing the instruction memory corresponding to a processor in the multiple instruction multiple data (MIMD) mode; and

providing access through said switch matrix to any other processor to the instruction any other processor to the instruction memories corresponding to the other processor other than the first processor in a serial chain of processors operating in the single instruction multiple data (SIMD) mode.

13. The computer implemented method as claimed in claim 12, further comprising the steps of:

determining at each processor if a next instruction to be executed is stored in said corresponding instruction memory;

accessing at each processor said corresponding instruction memory to recall the next instruction if the next instruction is stored in the corresponding instruction memory when either (1) the processor is in the multiple instruction multiple data (MIMD) mode or (2) the processor is in the single instruction multiple data (SIMD) mode and is the first processor in a serial chain of processors operating in the single instruction multiple data (SIMD) mode;

requesting at each processor the next instruction from an external memory if the instruction is not stored in said corresponding instruction memory when either (1) the processor is in the multiple instruction multiple data (MIMD) mode or (2) the processor is in the single instruction multiple data (SIMD) mode; and

disabling said steps of accessing and of requesting of each processor operating in the single instruction multiple data (SIMD) mode except a first processor in a serial chain of processors operating in the single instruction multiple data (SIMD) mode.