Architectural support for execution control of prologue and eplogue periods of loops in a VLIW processor

• US 5,794,029 A
• Filed: 10/18/1996
• Issued: 08/11/1998
• Est. Priority Date: 08/07/1996
• Status: Expired due to Term

• Current Assignee: Elbrus International Limited
• Original Assignee: Elbrus International Limited
• Inventors: Gruzdov, Feodor A., Sakhin, Yuli Kh., Babaian, Boris A., Gorokhov, Valeri G., Volkonski, Vladimir Yu.
• Primary Examiner(s): Lim, Krisna
• Application Number:
  US08/733,480
• Time in Patent Office: 662 Days
• Field of Search: 395/709, 395/588
• US Class Current: 712/241
• CPC Class Codes: G06F 8/452 Loops
• CPC Class Codes: G06F 9/325 for loops, e.g. loop detection or loop counter

Abstract

For certain classes of software pipelined loops, prologue and epilogue control is provided by loop control structures, rather than by predicated execution features of a VLIW architecture. For loops compatible with two simple constraints, code elements are not required for disabling garbage operations during prologue and epilogue loop periods. As a result, resources associated with implementation of the powerful architectural feature of predicated execution need not be squandered to service loop control. In particular, neither increased instruction width nor an increased number of instructions in the loop body is necessary to provide loop control in accordance with the present invention. Fewer service functions are required in the body of a loop. As a result, loop body code can be more efficiently scheduled by a compiler and, in some cases, fewer instructions will be required, resulting in improved loop performance. Loop control logic includes a loop control registers having an epilogue counter field, a shift register, a side-effects enabled flag, a current loop counter field, a loop mode flag, and side-effects manual control and loads manual control flags. Side-effects enabling logic and load enabling logic respectively issue a side-effects enabled predicate and a loads enabled predicate to respective subsets of execution units. Software pipelined simple and inner loops are supported.

Claims

1. An apparatus comprising:

a prologue distinguishing register complex responsive to physical iterations of loop body code, wherein the state of the prologue distinguishing register complex advances toward an end of prologue state in correspondence with the physical iterations of the loop body code;

an epilogue distinguishing register responsive to the physical iterations of the loop body code, wherein the state of the epilogue distinguishing register advances toward a beginning of epilogue state in correspondence with the physical iterations of the loop body code;

side-effects enabling logic coupled to the prologue distinguishing register complex and responsive to the state thereof, wherein the side-effects enabling logic supplies a side-effects enabled predicate except during a prologue period of the loop body; and

loads enabling logic coupled to the epilogue distinguishing register and responsive to the state thereof, wherein the loads enabling logic supplies a loads enabled predicate except during an epilogue period of the loop body.

2. An apparatus, as recited in claim 1, wherein the prologue distinguishing register complex comprises:

an N-bit shift register wherein an indication at a first particular one of the N bits corresponds to the end of prologue state and wherein the state of the N-bit shift register is advanced by shifting.

3. An apparatus, as recited in claim 2,

wherein the prologue distinguishing register complex further includes a side-effects enabled flag coupled to the N-bit shift register, wherein the side-effects enabled flag comprises a sticky-bit representation of the first particular one of the N bits of the shift register;

and

wherein the side-effects enabling logic is coupled to the side-effects enabled flag and responsive to the state thereof, wherein the side-effects enabling logic supplies the side-effects enabled predicate in response to an indication at the side-effects enabled flag.

4. An apparatus, as recited in claim 1,

wherein the prologue distinguishing register complex comprises a prologue counter field, wherein a count value thereof corresponds to the end of prologue state, and wherein the state prologue counter is advanced by counting in accordance with a preselected count direction;

and

wherein the side-effects enabling logic is coupled to the prologue counter field and responsive to the state thereof, wherein the side-effects enabling logic supplies the side-effects enabled predicate in response to the count value.

5. An apparatus, as recited in claim 3, further comprising:

an initialization operation in code for execution prior to the loop body code, wherein the initialization operation sets an Mth bit of the N-bit shift register, M being equal to a number of overlapped logical iterations minus one (1) in the loop body code, and wherein the initialization operation initializes the epilogue distinguishing register with an indication corresponding to a number of logical iterations in the loop body code.

6. An apparatus, as recited in claim 4, further comprising:

an initialization operation in code for execution prior to the loop body code, wherein the initialization operation initializes the prologue counter with an initialization value equal to a number of overlapped logical iterations minus one (1) in the loop body code, and wherein the initialization operation initializes the epilogue distinguishing register with an indication corresponding to a number of logical iterations in the loop body code.

7. An apparatus, as recited in claim 1, further comprising:

a first execution unit coupled to the side-effects enabling logic, wherein the first execution unit executes a first-type operation from the loop body code when the side-effects enabled predicate is supplied and inhibits execution of the first-type operation from the loop body code otherwise; and

a second execution unit coupled to the loads enabling logic, wherein the second execution unit executes a second-type operation from the loop body code when the loads enabled predicate is supplied and inhibits execution of the second-type operation from the loop body code otherwise.

8. An apparatus, as recited in claim 7,

wherein the first-type operation is an operation that causes side-effects;

and

wherein the second-type operation is a load operation.

9. An apparatus, as recited in claim 8, further comprising

a third execution unit for executing a third-type operation from the loop body code, wherein the third operation is not an operation that causes side-effects and not a load operation, and wherein the third execution unit is unaffected by the side-effects enabled predicate and the loads enabled predicate.

10. An apparatus, as recited in claim 9, further comprising a memory coupled to the first and second execution units as an operand source for the load operation and as a result destination for a store operation.

11. The apparatus of claim 1 in a loop control unit of a pipelined processor.

12. The apparatus of claim 1 in a loop control unit of a Very Long Instruction Word (VLIW) processor.

13. An apparatus, as recited in claim 1, wherein the loop body code comprises software pipelined loop body code corresponding to a simple loop.

14. An apparatus, as recited in claim 1, wherein the loop body code comprises software pipelined loop body code corresponding to an innermost loop nested within one or more levels of outer loops.

15. A method of controlling execution of software pipelined loop body code, the method comprising the steps of:

initializing a prologue distinguishing register complex with an indication corresponding to a number of overlapped logical iterations minus one (NOVL-1) in the loop body code;

initializing an epilogue distinguishing register with an indication corresponding to a number of logical iterations in the loop body code;

advancing the state of the prologue distinguishing register complex toward an end of prologue state in correspondence with physical iterations of the loop body code;

advancing the state of the epilogue distinguishing register toward a beginning of epilogue state in correspondence with physical iterations of the loop body code;

after the prologue distinguishing register complex reaches the end of prologue state, supplying a side-effects enabled predicate to a first execution unit providing execution support for operations of the loop body code that cause side-effects; and

until the epilogue distinguishing register reaches the beginning of epilogue state, supplying a loads enabled predicate to a second execution unit providing execution support for load operations of the loop body code.

16. The method of claim 15, wherein the prologue distinguishing register complex state advancing step includes:

shifting an N-bit shift register;

setting a sticky-bit side-effects enabled flag in correspondence with shifting an indication into a particular one of the N bits of the shift register corresponding to the end of prologue state; and

supplying the side-effects enabled predicate based on the state of the sticky-bit side-effects enabled flag.

17. The method of claim 15, wherein the prologue distinguishing register complex state advancing step includes:

advancing a counter in accordance with a preselected count direction; and

supplying the side-effects enabled predicate based on a value of the counter corresponding to the end of prologue state.

18. The method of claim 15, further comprising the steps of:

executing a first operation of a physical iteration of the loop body code at the first execution unit in accordance with the side-effects enabled predicate and otherwise inhibiting execution of the first operation;

executing a second operation of a physical iteration of the loop body code at the second execution unit in accordance with the loads enabled predicate and otherwise inhibiting execution of the second operation;

executing a third operation of a physical iteration of the loop body code at a third execution unit without regard to the side-effects enabled predicate or the loads enabled predicate.

19. A method for providing a computer system apparatus, the method comprising the steps of:

providing a prologue distinguishing register complex responsive to physical iterations of loop body code, wherein the state of the prologue distinguishing register complex advances toward an end of prologue state in correspondence with the physical iterations of the loop body code;

providing an epilogue distinguishing register responsive to the physical iterations of the loop body code, wherein the state of the epilogue distinguishing register advances toward a beginning of epilogue state in correspondence with the physical iterations of the loop body code;

providing side-effects enabling logic coupled to the prologue distinguishing register complex and responsive to the state thereof, wherein the side-effects enabling logic supplies a side-effects enabled predicate except during a prologue period of the loop body; and

providing loads enabling logic coupled to the epilogue distinguishing register and responsive to the state thereof, wherein the loads enabling logic supplies a loads enabled predicate except during an epilogue period of the loop body.