Engine to enable high speed context switching via on-die storage

Patent metadata

Field	Value
Publication number	US-11210265-B2
Application number	US-202016869223-A
Country	US
Kind code	B2
Filing date	May 7, 2020
Priority date	Apr 1, 2017
Publication date	Dec 28, 2021
Grant date	Dec 28, 2021

Abstract

In an example, an apparatus comprises a plurality of execution units, and a first memory communicatively couple to the plurality of execution units, wherein the first shared memory is shared by the plurality of execution units and a copy engine to copy context state data from at least a first of the plurality of execution units to the first shared memory. Other embodiments are also disclosed and claimed.

First claim

The invention claimed is:

1. An apparatus, comprising:
a general-purpose graphics processing compute block comprising a plurality of processing resources to execute graphics instructions;
a first shared memory communicatively coupled to the plurality of processing resources;
a first communication fabric to communicatively couple the plurality of processing resources to the first shared memory; and
a copy engine to:
receive a signal from an instruction scheduler indicating an initiation of a preemption process;
stop an execution of an existing context on at least a first of the plurality of processing resources; and
copy context state data from the existing context on the at least a first of the plurality of processing resources to the first shared memory in parallel with executing a new context on the plurality of processing resources.

2. The apparatus of claim 1 , wherein the copy engine is to:
generate a signal to indicate that a context preemption process is complete; and
upload the context data from the first shared memory to a second memory, separate from the first shared memory.

3. The apparatus of claim 2 , further comprising:
a second communication fabric to communicatively couple the copy engine to the second memory.

4. The apparatus of claim 1 , wherein the copy engine is coupled to the first shared memory by the first communication fabric.

5. The apparatus of claim 1 , the copy engine to:
copy context state data from the existing context on at least a first of the plurality of processing resources to the first shared memory in parallel with executing a new context on the plurality of processing resources.

6. The apparatus of claim 5 , the copy engine to:
restore the first shared memory.

7. A method comprising:
receiving, in a copy engine of general-purpose graphics processing compute block comprising a plurality of processing resources to execute graphics instructions, a signal from an instruction scheduler indicating an initiation of a preemption process;
stopping an execution of an existing context on at least a first of the plurality of processing resources; and
copying context state data from the existing context on the at least a first of the plurality of processing resources to a first shared memory in parallel with executing a new context on the plurality of processing resources via a first communication fabric communicatively coupled to the plurality of processing resources to the first shared memory.

8. The method of claim 7 , further comprising:
generating a signal to indicate that a context preemption process is complete; and
uploading the context data from the first shared memory to a second memory, separate from the first shared memory.

9. The method of claim 7 , wherein the general-purpose graphics processing compute block further comprises:
a first communication fabric to communicatively couple the plurality of processing resources to the first shared memory.

10. The method of claim 9 , wherein the copy engine is coupled to the first shared memory by the first communication fabric.

11. The method of claim 7 , wherein the general-purpose graphics processing compute block further comprises:
a second communication fabric to communicatively couple the copy engine to a second memory.

12. The method of claim 7 , further comprising:
copying context state data from the existing context on at least a first of the plurality of processing resources to the first shared memory in parallel with executing a new context on the plurality of processing resources.

13. The method of claim 12 , further comprising:
restoring the first shared memory.

14. One or more non-transitory computer-readable media comprising a plurality of instructions stored thereon that, when executed by a processor, causes the processor to:
receive, in a copy engine of general-purpose graphics processing compute block comprising a plurality of processing resources to execute graphics instructions, a signal from an instruction scheduler indicating an initiation of a preemption process;
stop an execution of an existing context on at least a first of the plurality of processing resources; and
copy context state data from the existing context on the at least a first of the plurality of processing resources to a first shared memory in parallel with executing a new context on the plurality of processing resources via a first communication fabric communicatively coupled to the plurality of processing resources to the first shared memory.

15. The one or more computer-readable media of claim 14 , wherein the plurality of instructions further cause the processor to:
generate a signal to indicate that a context preemption process is complete; and
upload the context data from the first shared memory to a second memory, separate from the first shared memory.

16. The one or more computer-readable media of claim 14 , wherein the general-purpose graphics processing compute block further comprises:
a first communication fabric to communicatively couple the plurality of processing resources to the first shared memory.

17. The one or more computer-readable media of claim 16 , wherein the copy engine is coupled to the first shared memory by the first communication fabric.

18. The one or more computer-readable media of claim 14 , wherein the general-purpose graphics processing compute block further comprises:
a second communication fabric to communicatively couple the copy engine to the second memory.

19. The one or more computer-readable media of claim 14 , wherein the plurality of instructions further cause the processor to:
copy context state data from the existing context on at least a first of the plurality of processing resources to the first shared memory in parallel with executing a new context on the plurality of processing resources.

20. The one or more computer-readable media of claim 19 , wherein the plurality of instructions further cause the processor to:
restore the first shared memory.

Assignees

• Intel Corp

Inventors

• Koker Altug
• Surti Prasoonkumar
• Puffer David
• Maiyuran Subramaniam
• Lueh Guei-Yuan
• Appu Abhishek R
• Ray Joydeep
• Vembu Balaji
• Bar-On Tomer
• Lauritzen Andrew T
• Labbe Hugues
• Gierach John G
• Liktor Gabor

Classifications

• G06F9/461 Primary

  Saving or restoring of program or task context · CPC title

• G06F9/38

  Concurrent instruction execution, e.g. pipeline or look ahead · CPC title

• G06F12/1045

  associated with a data cache · CPC title

• G06F2201/84

  Using snapshots, i.e. a logical point-in-time copy of the data · CPC title

• G06F12/0831

  using a bus scheme, e.g. with bus monitoring or watching means · CPC title