Digital signal processor having distributed register file

• US 6,510,510 B1
• Filed: 12/22/1998
• Issued: 01/21/2003
• Est. Priority Date: 01/25/1996
• Status: Expired due to Term

• Current Assignee: Analog Devices, Inc.
• Original Assignee: Analog Devices, Inc.
• Inventors: Garde, Douglas
• Primary Examiner(s): Follansbee, John A.
• Application Number:
  US09/218,346
• Time in Patent Office: 1,491 Days
• Field of Search: 712/35, 712/218
• US Class Current: 712/218
• CPC Class Codes: G06F 15/7857 using interleaved memory addressing G06F12/0607
• CPC Class Codes: G06F 9/30036 Instructions to perform operations on packed data, e.g. vector, tile or matrix operations
• CPC Class Codes: G06F 9/3802 Instruction prefetching
• CPC Class Codes: G06F 9/3806 using address prediction, e.g. return stack, branch history buffer
• CPC Class Codes: G06F 9/3814 Implementation provisions of instruction buffers, e.g. prefetch buffer; banks
• CPC Class Codes: G06F 9/382 Pipelined decoding, e.g. using predecoding
• CPC Class Codes: G06F 9/3824 Operand accessing
• CPC Class Codes: G06F 9/383 Operand prefetching cache prefetching G06F12/0862
• CPC Class Codes: G06F 9/3861 Recovery, e.g. branch miss-prediction, exception handling error detection or correction G06F11/00
• CPC Class Codes: G06F 9/3885 using a plurality of independent parallel functional units
• CPC Class Codes: Y02D 10/00 Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

A computation block for use in a digital signal processor includes a register file for storage of operands and results and one or more computation units for executing digital signal computations. A first digital signal computation is performed with one of the computation units, and an intermediate result is produced. The intermediate result is transferred from a result output of the computation unit to an intermediate result input of one or more of the computation units without first transferring the intermediate result to the register file. A second digital signal computation is performed using the intermediate result to produce a final result or a second intermediate result.

Claims

1. A computation block for performing digital signal computations, comprising:

a register file for storage of operands and results of said digital signal computations;

first and second computation units for executing said digital signal computations using said operands and producing said results;

one or more operand buses each coupled between an operand output of said register file and an operand input of said first and second computation units; and

one or more result buses each coupled to a result output of said first and second computation units, to an intermediate result input of said first and second computation units and to a result input of said register file, wherein an intermediate result of a digital signal computation may be transferred directly from the result output of one of said computation units to the intermediate result inputs of one or both of said first and second computation units for use in a subsequent computation without first transferring the intermediate result to said register file, wherein each of said computation units comprises a first latch having inputs coupled to each of said operand buses and each of said result buses, a first multiplexer having inputs coupled to said first latch for selecting a first operand from one of said operand buses or one of said result buses in response to a first control signal, a second latch having inputs coupled to each of said operand buses and each said result buses, a second multiplexer having inputs coupled to said second latch for selecting a second operand from one of said operand buses or one of said result buses in response to a second control signal, a computation circuit for receiving said first and second operands and executing said digital signal computation, and an output latch having an input coupled to said computation circuit and an output coupled to one or more of said result buses.

2. A computation block as defined in claim 1 wherein said first computation unit comprises a multiplier for performing multiplication operations and said second computation unit comprises an ALU for performing arithmetic operations.

3. A computation block as defined in claim 2 further comprising a third computation unit, wherein each of said operand buses is coupled to an operand input of said third computation unit and wherein said result bus is coupled to a result output and an intermediate result input of said third computation unit, said third computation unit comprising a shifter for performing shifting operations.

4. A computation block as defined in claim 1 wherein said register file comprises a plurality of registers, an operand latch having inputs coupled to said registers and outputs coupled to each of said operand buses, and a result latch having inputs coupled to each of said result buses and outputs coupled to said registers.

5. A computation block as defined in claim 4 wherein the first and second latches in each of said computation units and the result latch in said register file are controlled in response to a first phase of a system clock and wherein the operand latch in said register file and the output latch in each of said computation units are controlled in response to a second phase of the system clock.

6. A computation block as defined in claim 1 comprising two or more result buses each coupled to the result outputs of one or more of said first and second computation units, to the intermediate result inputs of said first and second computation units and to the result inputs of said register file.

7. The computation block of claim 1, wherein the first and second latches of the first computation unit constitute a first local portion of the register file and the first and second latches of the second computation unit constitute a second local portion of the register file.

8. The computation block of claim 1, wherein the first latch comprises a first operand latch and a first intermediate result latch.

9. The computation block of claim 1, wherein the second latch comprises a second operand latch and a second intermediate result latch.

10. A computation block for performing digital signal computations, comprising:

a register file for storage of operands and results of said digital signal computations, said register file having an operand output and a result input;

at least one computation unit for executing said digital signal computations using said operands and producing said results, said computation unit having an operand input, an intermediate result input and a result output;

at least one operand bus coupled between the operand output of said register file and the operand input of said computation unit; and

at least one result bus coupled to the result output of said computation unit, to the intermediate result input of said computation unit and to the result input of said register file, wherein an intermediate result of a digital signal computation may be transferred directly from the result output of said computation unit to the intermediate result input of said computation unit for use in a subsequent computation without first transferring the intermediate result to said register file, wherein said computation unit comprises a first latch coupled to said operand bus and said result bus, a first multiplexer having inputs coupled to said first latch for selecting a first operand from said operand bus or said result bus in response to a first control signal, a second latch having inputs coupled to said operand bus and said result bus, a second multiplexer having inputs coupled to said second latch for selecting a second operand from said operand bus or said result bus in response to a second control signal, a computation circuit receiving said first and second operands from said first and second multiplexers, respectively, and an output latch having an input coupled to said computation circuit and having an output coupled to said result bus.

11. A computation block as defined in claim 10 wherein said register file comprises a plurality of registers, an operand latch having an input coupled to said registers and an output coupled to said operand bus, and a result latch having an input coupled to said result bus and an output coupled to said registers.

12. A computation block as defined in claim 11 wherein said first and second latches and said result latch are controlled in response to a first phase of a system clock and wherein the operand latch in said register file and the output latch in said computation unit are controlled in response to a second phase of the system clock.

13. A computation unit for performing digital signal computations, comprising:

a computation circuit having a first input for receiving a first operand used in digital signal computations and a second input for receiving a second operand used in digital signal computations;

a first operand latch having an input coupled to at least one operand bus for storing operands passed on the at least one operand bus from an external register file;

a first intermediate result latch having an input coupled to at least one result bus for storing operands from the at least one result bus;

a second operand latch having an input to coupled to the at least one operand bus for storing operands passed on the at least one operand bus from the external register file;

a second intermediate result latch having an input coupled to the at least one result bus for storing operands from the at least one result bus;

a first multiplexer for selecting a first operand from one of the first operand latch and first intermediate result latch and supplying the first operand to the first input of the computation circuit; and

a second multiplexer for selecting a second operand from one of the second operand latch and second intermediate result latch and supplying the second operand to the second input of the computation circuit.

14. The computation unit of claim 13, further comprising a result latch coupled between the computation circuit and the at least one result bus for passing results output by the computation circuit to the at least one result bus.

15. The computation unit of claim 14, wherein the operand stored by the first intermediate result latch is passed to the at least one result bus by the result latch.

16. The computation unit of claim 14, wherein the operand stored by the first intermediate result latch is passed to the at least one result bus by a second computation unit.

17. The computation unit of claim 13, wherein the first and second intermediate result latches constitute a local portion of a distributed register file associated with the computation unit.