Efficient Video Decoding Migration For Multiple Graphics Processor Systems

• US 20090160865A1
• Filed: 12/19/2007
• Published: 06/25/2009
• Est. Priority Date: 12/19/2007
• Status: Active Grant

• Current Assignee: Advanced Micro Devices, Inc.
• Original Assignee: Advanced Micro Devices, Inc.
• Inventors: Grossman, Mark S.
• Granted Patent: US 8,330,762 B2
• Time in Patent Office: Days
• US Class Current: 345/502
• CPC Class Codes: H04N 19/127 Prioritisation of hardware or computational resources
• CPC Class Codes: H04N 19/159 Prediction type, e.g. intra-frame, inter-frame or bidirectional frame prediction
• CPC Class Codes: H04N 19/172 the region being a picture, frame or field
• CPC Class Codes: H04N 19/44 Decoders specially adapted therefor, e.g. video decoders which are asymmetric with respect to the encoder
• CPC Class Codes: H04N 9/87 Regeneration of colour television signals H04N9/80 takes precedence

Abstract

Embodiments of the invention as described herein provide a solution to the problems of conventional methods as stated above. In the following description, various examples are given for illustration, but none are intended to be limiting. Embodiments include a frame processor module in a graphics processing system that examines the intra-coded and inter-coded frames in an encoded video stream and initiates migration of decoding and rendering functions to a second graphics processor from a first graphics processor based on the location of intra-coded frames in a video stream and the composition of intermediate inter-coded frames.

Claims

1. An apparatus comprising:

a plurality of graphics processing units each providing rendered graphics data to one or more users;

an input circuit receiving video stream data from a data store, the data including one or more individual encoded video streams, each encoded video stream of the one or more individual encoded video streams organized into groups of frames including intra-coded frames that are decoded without information from other frames within a group, and inter-coded frames that are decoded using information from at least one other frame within the group; and

a frame processing component examining a received video stream for a first graphics processing unit and initiating decoding of the received video stream on a second graphics processing unit upon receipt of a data migration signal.

2. The apparatus of claim 1 wherein the data migration signal is transmitted upon a triggering event selected from the group consisting of:

failure of the first graphics processing unit, scheduled maintenance of the first graphics processing unit, and excessive workload sustained by the first graphics processing unit.

3. The apparatus of claim 2 wherein the frame processing component compares an amount of time between the triggering event and a successive intra-coded frame of the video stream with a defined threshold time value.

4. The apparatus of claim 3 wherein if the time between the triggering event and the successive intra-coded frame does not exceed the defined threshold time value, the video stream is decoded and rendered by the second graphics processing unit upon receipt of the successive intra-coded frame.

5. The apparatus of claim 4 wherein if the time between the triggering event and the successive intra-coded frame exceeds the defined threshold time value, the video stream is rendered by the first graphics processor, all frames from the first intra-coded frame are decoded by the second graphics processing unit, and the video stream is decoded and rendered by the second graphics processing unit when the second graphics engine and the first graphics engine simultaneously decode the same frame.

6. The apparatus of claim 4 wherein if the time between the triggering event and the successive intra-coded frame exceeds the defined threshold time value, the video stream is transmitted to both the first graphics processor and the second graphics processor, and wherein the video stream is decoded and rendered by the second graphics processing unit upon receipt of a next successive inter-coded frame if the inter-coded frame is of a first type, and upon receipt of a second successive inter-coded frame if the inter-coded frame is of a second type.

7. The apparatus of claim 2 wherein the frame processing component determines the identity of reference frames required by the successive inter-coded frame, and the relative time difference an amount of copy bandwidth available for migration to the second graphics processing unit.

8. The apparatus of claim 7 wherein if the successive inter-coded frame is a bi-directional type frame, a portion of the video stream is copied to the second graphics processing unit if the copy time is faster than a decode time on the second graphics processing unit by a defined factor.

9. The apparatus of claim 4 wherein the compressed video stream data comprises MPEG compressed video stream data.

10. The apparatus of claim 9 wherein the intra-coded frame comprises an MPEG I-frame, and wherein the inter-coded frames comprise MPEG P-frames and B-frames.

11. The apparatus of claim 10 wherein the graphics processing units are coupled to the host CPU over an interface bus in a graphics processing subsystem of a computing device, and wherein the computing device is selected from the group consisting of:

a personal computer, a workstation, a handheld computing device, a digital television, a media playback device, and a game console.

12. A method for migrating digital video stream decoding from a first graphics processing unit (GPU) to a second GPU, comprising:

receiving a notification that migration from the first GPU to the second GPU is required;

analyzing a video stream and rendered by the first GPU and organized into groups of frames including intra-coded frames that are decoded without information from other frames within a group, and inter-coded frames that are decoded using information from at least one other frame within the group, to determine an amount of time between a migration trigger event and a next intra-coded frame; and

initiating decoding of the received video stream on a second graphics processing unit upon receipt of a data migration signal upon occurrence of the migration trigger event.

13. The method of claim 12 wherein the trigger event is selected from the group consisting of:

failure of the first graphics processing unit, scheduled maintenance of the first graphics processing unit, and excessive bandwidth consumption by the first graphics processing unit.

14. The method of claim 13 further comprising:

comparing an amount of time between the triggering event and a successive intra-coded frame of the video stream with a defined threshold time value;

determining an identity of reference frames required by a successive inter-coded frame of the video stream;

determining an amount of copy bandwidth available to copy at least a portion of the received video stream from the first GPU to the second GPU; and

determining a relative difference between a time required to copy the portion of the received video stream to the second GPU and a time required to decode a next frame of the portion of the received video stream in the second GPU.

15. The method of claim 14 wherein if the time between the triggering event and the successive intra-coded frame does not exceed the defined threshold time value, the video stream is decoded and rendered by the second graphics processing unit upon receipt of the successive intra-coded frame.

16. The method of claim 15 wherein if the time between the triggering event and the successive intra-coded frame exceeds the defined threshold time value, the video stream is rendered by the first graphics processor, all frames from the first intra-coded frame are decoded by the second graphics processing unit, and the video stream is decoded and rendered by the second graphics processing unit when the second graphics engine and the first graphics engine simultaneously decode the same frame.

17. The method of claim 15 wherein if the time between the triggering event and the successive intra-coded frame exceeds the defined threshold time value, the video stream is transmitted to both the first graphics processor and the second graphics processor, and wherein the video stream is decoded and rendered by the second graphics processing unit upon receipt of a next successive inter-coded frame if the inter-coded frame is of a first type, and upon receipt of a second successive inter-coded frame if the inter-coded frame is of a second type.

18. The method of claim 15 wherein the compressed video stream data comprises MPEG compressed video stream data.

19. The method of claim 18 wherein the intra-coded frame comprises an MPEG I-frame, and wherein the inter-coded frames comprise MPEG P-frames and B-frames.

20. The method of claim 19 wherein the graphics processing units are coupled to the host CPU over an interface bus in a graphics processing subsystem of a computing device, and wherein the computing device is selected from the group consisting of:

a personal computer, a workstation, a handheld computing device, a digital television, a media playback device, and a game console.

21. A method for migrating an MPEG-encoded digital video stream decoding from a first graphics processing unit (GPU) to a second GPU, comprising:

receiving a notification that migration from the first GPU to the second GPU is required;

analyzing the video stream as organized into groups of frames including I-frames, P-frames and B-frames, to determine an amount of time between a migration trigger event and a next I-frame; and

determining if the frame type of the next frame to be processed by the first GPU, and initiating decoding of the received video stream on the second graphics processing unit and migration of the data stream to the second GPU based on an amount of time between the triggering event and the next I-frame, the amount of copy bandwidth available, and the identity of reference frames required by the next frame.

22. The method of claim 21 wherein, if the next frame is a P-frame, the method comprises determining if the time to the next I-frame is less than a defined threshold time.

23. The method of claim 22 wherein, if the time to the next I-frame is less than the defined threshold time, the method comprises:

decoding a remainder of the portion of the video stream on the first GPU; and

migrating the remainder of the entire video stream to the second GPU starting with the next I-frame.

24. The method of claim 22 wherein, if the time to the next I-frame is greater than the defined threshold time, the method comprises:

copying a previously decoded P-frame to the second GPU; and

decoding the previously decoded P-frame on the second GPU.

25. The method of claim 21 wherein, if the next frame is a B-frame, the method further comprises:

determining if the B-frame is dependent on an I-frame or a P-frame; and

determining whether a time required to copy a remaining portion of the video stream is less than the time required to decode a block of the portion of the video stream in the second GPU by a defined threshold value.

26. The method of claim 25 wherein, if the B-frame is not dependent on an I-frame, and the time required to copy the remaining portion of the video stream is less than the threshold value, the method further comprises:

migrating the video stream to the second GPU starting with a second reference frame; and

decoding the B-frame on the second GPU.

27. The method of claim 25 wherein, if the B-frame is not dependent on an I-frame, and the time required to copy the remaining portion of the video stream is greater than or equal to the threshold value, the method further comprises:

copying a first decoded reference frame to the second GPU;

migrating the video stream to the second GPU starting with a second reference frame; and

decoding the second reference frame and the B-frame on the second GPU.