Control of predication across clock domains

Patent metadata

Field	Value
Publication number	US-10133300-B2
Application number	US-201414150554-A
Country	US
Kind code	B2
Filing date	Jan 8, 2014
Priority date	Jan 8, 2014
Publication date	Nov 20, 2018
Grant date	Nov 20, 2018

Abstract

Embodiments are disclosed for a method of executing instructions in a processing core of a microprocessor. In one embodiment, the method comprises, in a first clock domain, receiving an input from a second clock domain external to the first clock domain, the input comprising an indication from the second clock domain regarding whether to execute an instruction in the first clock domain. The method further comprises synchronizing the input from the second clock domain with the first clock domain, if the instruction is a predicatable instruction and the indication matches a predicate condition that indicates not to perform the instruction, then not performing the instruction, and otherwise performing the instruction.

First claim

The invention claimed is:

1. A method of executing instructions in a processing core of a microprocessor, the method comprising:
in a first clock domain, receiving an input from a signal source in a second clock domain external to the first clock domain, the input comprising an indication from the second clock domain of a condition that determines whether an instruction is executed in the first clock domain;
synchronizing the input from the signal source in the second clock domain with the first clock domain;
if the instruction is determined to be an invalid instruction, then not performing the instruction;
if the instruction is a predicatable instruction and the indication matches a predicate condition that indicates not to perform the instruction, then not performing the instruction; and
otherwise performing the instruction.

2. The method of claim 1 , wherein determining if the instruction is a predicatable instruction is further based upon a signal from a decoder of the processing core indicating whether the instruction is able to be predicated.

3. The method of claim 1 , wherein the processing core of the microprocessor is a first processing core of the microprocessor and the signal source in the second clock domain external to the first clock domain is a second processing core of the microprocessor.

4. The method of claim 1 , wherein the microprocessor is a first microprocessor of a computing device and the signal source in the second clock domain external to the first clock domain is a second microprocessor of the computing device.

5. The method of claim 4 , wherein the first microprocessor is a graphics processing unit and the second microprocessor is a central processing unit.

6. The method of claim 1 , wherein not performing the instruction includes turning an instruction valid indicator for the instruction off, the instruction valid indicator causing an execution unit controller of the processing core to interpret the instruction as a no-op.

7. A processor comprising:
a clock domain crossing block in a first clock domain the clock domain crossing block including two or more flip flops, and configured to receive input from an external signal source in a second clock domain, synchronize the input with the first clock domain, and output a synchronized indication in the first clock domain that determines whether an instruction is executed in the first clock domain, the input being based at least on a result of processing performed in the second clock domain;
a predication verification block comprising a first logical AND gate configured to receive input from the clock domain crossing block comprising the synchronized indication and from a decoder comprising a decoded indication regarding whether the instruction is of a type that is able to be predicated; and
an instruction verification block comprising a second logical AND gate configured to receive input including an inverted output of the first logical AND gate and output from the decoder comprising a validation indication regarding whether the instruction is a valid instruction, the instruction verification block configured to output a signal configured to indicate whether the instruction is to be executed based on the output of the second logical AND gate.

8. The processor of claim 7 , wherein the clock domain crossing block includes a first D flip flop, configured to receive the input from the external signal source and a clock signal from the first clock domain, and a second D flip flop, configured to receive an output of the first D flip flop and the clock signal from the first clock domain.

9. The processor of claim 7 , wherein the output of the first logical AND gate is directed to an input of a D flip flop prior to being directed to an inverter, the inverter forming the inverted output of the first logical AND gate.

10. The processor of claim 7 , wherein the processor comprises a processing core of a central processing unit and the external signal source includes a processing core of a graphics processing unit.

11. The processor of claim 7 , wherein the processor comprises a first processing core of a central processing unit and the external signal source includes a second processing core of the central processing unit.

12. The processor of claim 7 , wherein the output of the first logical AND gate is a logical 1 if the instruction is able to be predicated and the synchronized indication represents a condition indicating not to execute the instruction in the first clock domain.

13. The processor of claim 12 , wherein the output of the second logical AND gate is a logical 0 if the instruction is invalid or if the instruction is able to be predicated and the synchronized indication represents a condition indicating not to execute the instruction in the first clock domain.

Assignees

• Microsoft Technology Licensing Llc

Inventors

• Tubbs Matthew Ray
• Shearer Robert Allen
• Haraden Ryan

Classifications

• G06F1/12 Primary

  Synchronisation of different clock signals {provided by a plurality of clock generators} · CPC title

• G06F9/30145

  Instruction analysis, e.g. decoding, instruction word fields · CPC title

• G06F15/17325

  Synchronisation; Hardware support therefor (intertask synchronisation G06F9/52) · CPC title

• G06F9/30072 Primary

  to perform conditional operations, e.g. using predicates or guards · CPC title

• G06F9/3869

  Implementation aspects, e.g. pipeline latches; pipeline synchronisation and clocking · CPC title