System and method for load balancing a video signal in a multi-core processor
View Patent ↗Sequential video data frames are encoded using cores including a first core and a second core. A first beginning frame is divided into slices. The first core is assigned to process a first slice. The second core is assigned to process a second slice. The first beginning frame is processed using the cores which results in a first ending frame in which the first slice was partitioned into a third slice and a fourth slice. The third slice was processed by the first core. The fourth slice and the second slice were processed by the second core. A second beginning frame, which immediately follows the first ending frame, is divided into a second plurality of slices. The first core is assigned to the third slice. The second core is assigned to a fifth slice which has a size equal to a sum of the second and fourth slices.
1. A method of encoding sequential video data frames using a plurality of cores comprising a first core and a second core, the method comprising:
dividing a beginning portion of a first frame into a first plurality of slices among the plurality of cores in which the first core is assigned to process a first slice of the plurality of slices and the second core is assigned to process a second slice of the plurality of slices;
processing the first frame using the plurality of cores;
partitioning the first slice into a third slice and a fourth slice, wherein the third slice is processed by the first core and the fourth slice and the second slice are processed by the second core; and
dividing a second frame, which immediately follows the first ending frame, into a second plurality of slices among the plurality of cores in which the first core is assigned to a fifth slice equivalent in size to the third slice and the second core is assigned to a sixth slice which has a size equal to a sum of the second and fourth slices.
2. The method of claim 1 , wherein the plurality of cores comprises a master core and a plurality of slave cores.
3. The method of claim 2 , wherein the step of processing the first frame is further characterized by the master core partitioning the slices according to a protocol.
4. The method of claim 3 , wherein the step of dividing the second frame is further characterized by each of the plurality of cores being assigned to a slice having a plurality of macroblocks equal in number to the number of macroblocks processed on the first frame.
5. The method of claim 1 , further comprising:
processing the second frame using the plurality of cores to produce a modified second frame.
6. The method of claim 5 , wherein the step of processing the second frame is further characterized as partitioning the second plurality of slices into a third plurality of slices greater in number than the second plurality of slices.
7. The method of claim 6 , wherein the step of processing the second frame is further characterized by each slice of the third plurality of slices being processed by one of the plurality of cores, wherein each core processes a slice having a size equal to a sum of the sizes of the slices that it processed at an end of the first frame.
8. The method of claim 1 , wherein the step of dividing the second frame is further characterized by the second plurality of slices being of a number of slices equal to a number of slices of the first plurality of slices.
9. The method of claim 1 , wherein the second slice and the fourth slice are non-adjacent.
10. A method of processing frames, comprising:
providing a processing unit comprising a plurality of cores of a first number;
dividing a first frame into a first plurality of slices of the first number, wherein each slice of the first plurality of slices has a size;
assigning each core of the plurality of cores to a slice of the first plurality of slices, wherein each slice of the first plurality of slices has a core of the plurality of cores assigned to it;
initially processing a beginning portion of the first frame with the plurality of cores processing the first plurality of slices as assigned according to the step of assigning each core of the plurality of cores to a slice of the first plurality of slices;
processing an ending portion of the first frame, wherein the ending portion of the first frame has a second plurality of slices of a second number, wherein the second number is greater than the first number, wherein each core has processed at least one slice of the second plurality of slices and at least one core has processed more than slice of the second plurality of slices, wherein each core has processed an amount of the first ending frame equal to a sum of the slices processed by it;
dividing a beginning portion of a second frame into a third plurality of slices of the first number;
assigning each core of the plurality of cores to a slice of the third plurality of slices, wherein each slice of the third plurality of slices has a core of the plurality of cores assigned to it, wherein each core is assigned to an amount of the frame equal to the amount it processed on the ending portion of the first frame; and
initially processing the beginning portion of the second frame with the plurality of cores processing the third plurality of slices as assigned according to the step of assigning each core of the plurality of cores to a slice of the third plurality of slices.
11. The method of claim 10 , further comprising:
finally processing the second frame to produce a modified second frame.
12. The method of claim 11 , wherein the step of finally processing the second frame is further characterized by an ending portion of the second frame comprising a fourth plurality of slices greater in number than the third plurality of slices.
13. The method of claim 10 , wherein the step of processing an ending portion the first frame is further characterized by a master core of the plurality of cores controlling the plurality of cores according to a protocol.
14. The method of claim 10 , wherein the method operates in compliance with standard H.264.
15. The method of claim 10 , wherein the first frame and the second frame comprise a plurality of macroblocks.
16. The method of claim 15 , wherein a size equal to a sum of the first plurality of slices is equal to a size equal to a sum of the plurality of macroblocks.
17. The method of claim 10 , further comprising initially processing a third frame with the cores processing sizes equal to the sizes processed at the ending portion of the second frame.
18. An encoder, comprising:
a plurality of slave cores, wherein each slave core is configured to process one or more slices of a frame;
a master core that directs operation of the slave cores and is configured to process one or more slices of a frame, wherein:
the master core is configured to assign itself and each slave core to a single slice of a first plurality of slices of a beginning portion of a first frame, wherein a number of slices of the first plurality of slices is equal to a number of slave cores plus one for the master core;
the master core is configured to partition the slices of the first plurality of slices to result in a second plurality of slices greater in number than the first plurality of slices according to a protocol based on relative processing complexity among the first plurality of slices;
the master core is configured to selectively assign itself and each slave core of the plurality of slave cores to process the second plurality of slices at an ending portion of the first frame, wherein a size of slice processed by each core is in accordance with the protocol; and
the master core is configured to assign itself and each slave core to a single slice of a third plurality of slices of a beginning portion of a second frame, wherein a number of slices of the third plurality of slices is equal to a number of slices of the first plurality of slices and less than a number of slices of the second plurality of slices; wherein each slave core is assigned to a slice having a size equal to the size processed by it during the ending portion of the first frame.
19. The encoder of claim 18 , further comprising a memory shared among the master core and the slave cores.
20. The encoder of claim 18 , wherein the beginning portion of the first frame comprises a plurality of macroblocks, wherein a size equal to a sum of the first plurality of slices is equal to a size equal to a sum of the plurality of macroblocks which is equal to a size of a sum of the second plurality of slices.