Allocating processing resources for multiple instances of a software component
View Patent ↗Methods and corresponding systems for allocating processing resources for a number of instances (N) of a software component include determining an average processing cost (μ) and a variance (σ 2 ) for the software component. Then a processing cost for the software component is estimated as a function of N, the average processing cost (μ), and the variance (σ 2 ), and processing resources are allocated in response to the estimated processing cost. The software component can be partitioned into a number of blocks (L), wherein the L blocks include a required block and one or more optional blocks. In some embodiments in response to a total estimated processing cost exceeding an available processing value, selected optional blocks can be disabled to reduce the total estimated processing cost to a value equal to or less than the available processing value. The optional blocks can be prioritized and disabled in order of priority.
1. A method for allocating processing resources for a number of instances (N) of a software component and a probability of completion (Po) comprising:
in a processing unit;
determining an average processing cost (μ) and a variance (σ 2 ) for the software component;
estimating a processing cost for the software component as a function of N, the probability of completion (Po), the average processing cost (μ), and the variance (σ 2 ); and
allocating processing resources in response to the processing cost.
2. The method for allocating processing resources according to claim 1 wherein the estimating a processing cost for the software component, and the allocating processing resources, comprises:
partitioning the software component into a number of blocks (L blocks), wherein the L blocks include at least one required block and one or more optional blocks;
determining an average processing cost (μ i ) and a variance (σ i 2 ) for each of the L blocks;
estimating an estimated processing cost (m i ) for each of the L blocks as a function of N, and the respective average processing cost (μ i ) and variance (σ i 2 ) of the L blocks; and
allocating processing resources in response to the estimated processing cost m i for each of the L blocks.
3. The method for allocating processing resources according to claim 2 wherein the estimating an estimated processing cost (m i ) for each of the L blocks comprises:
computing a scaling factor (α) that is a function of N and the probability of completion (P 0 ) of the L blocks; and
estimating an estimated processing cost (m i ) for each of the L blocks as a function of N, the probability of completion (P 0 ), and the respective average processing cost (μ i ) and variance (σ i 2 ) for each of the L blocks according to the formula m i =μ i +α·σ i .
4. The method for allocating processing resources according to claim 3 further comprising selecting the probability of completion (P 0 ) to achieve a quality of service (QoS) level.
5. The method for allocating processing resources according to claim 3 wherein the computing a scaling factor (α) that is a function of N and the probability of completion (P 0 ) comprises computing the scaling factor (α) according to the formula
α
=
2
erf
-
1
(
2
·
P
0
-
1
)
N
.
6. The method for allocating processing resources according to claim 2 wherein the allocating processing resources comprises:
adding the estimated processing cost m i for each of the L blocks in each of the N software components to compute a total estimated processing cost for the N software components;
comparing the total estimated processing cost to an available processing value; and
in response to the total estimated processing cost exceeding the available processing value, disabling selected optional blocks to reduce the total estimated processing cost to a value equal to or less than the available processing value.
7. The method for allocating processing resources according to claim 6 wherein disabling selected optional blocks comprises:
prioritizing the optional blocks; and
disabling a selected lower priority optional block before disabling a selected higher priority optional block.
8. The method for allocating processing resources according to claim 2 wherein the partitioning the software component into a number of blocks (L blocks), wherein the L blocks include at least one required block and one or more optional blocks comprises partitioning an echo canceller software component into L blocks, wherein the L blocks includes a required echo cancelling block and one or more optional quality-enhancing blocks.
9. The method for allocating processing resources according to claim 2 further comprising updating the average processing cost (μ i ) and variance (σ i 2 ) for one or more of the L blocks based upon measurements in a preceding time window.
10. A system for allocating processing resources for a number of instances (N) of a software component and a probability of completion (Po) comprising:
a data memory for storing data associated with the N instances of the software component; and
a processing unit coupled to the data memory, wherein the processing unit and the data memory are cooperatively operable for executing the N instances of the software component, and for:
estimating an estimated processing cost for the software component as a function of N, the probability of completion (Po), an average processing cost (μ) and a variance (σ 2 ) for the software component; and
allocating processing resources in response to the estimated processing cost.
11. The system for allocating processing resources according to claim 10 wherein the software component is partitioned into a number of blocks (L blocks), wherein the L blocks include at lest one required block and one or more optional blocks, and wherein the processing unit and the data memory are further cooperatively operable for:
estimating a processing cost (m i ) for each of the L blocks as a function of N, and a respective average processing cost (μ i ) and a variance (σ i 2 ) of each of the L blocks; and
allocating processing resources in response to the estimated processing cost m i of the L blocks.
12. The system for allocating processing resources according to claim 11 wherein the processing unit and the data memory are further cooperatively operable for:
computing a scaling factor (α) that is a function of N and the probability of completion (P 0 ) of the L blocks; and
estimating an estimated processing cost (m i ) for each of the L blocks as a function of N, the probability of completion (P 0 ), and the respective average processing cost (μ i ) and variance (σ i 2 ) for each of the L blocks according to the formula m i =μ i +α·σ i .
13. The system for allocating processing resources according to claim 12 wherein the probability of completion (P 0 ) is selected to achieve a quality of service (QoS) level.
14. The system for allocating processing resources according to claim 12 wherein the processing unit and the data memory are further cooperatively operable for computing the scaling factor (α) according to the formula
α
=
2
erf
-
1
(
2
·
P
0
-
1
)
N
.
15. The system for allocating processing resources according to claim 11 wherein the processing unit and the data memory are further cooperatively operable for:
adding the estimated processing cost m i for each of the L blocks in each of the N software components to compute a total estimated processing cost for the N software components;
comparing the total estimated processing cost to an available processing value; and
in response to the total estimated processing cost exceeding the available processing value, disabling selected optional blocks to reduce the total estimated processing cost to a value equal to or less than the available processing value.
16. The system for allocating processing resources according to claim 15 wherein the processing unit and the data memory are further cooperatively operable for:
prioritizing the optional blocks; and
disabling a selected lower priority optional block before disabling a selected higher priority optional block.
17. The system for allocating processing resources according to claim 11 wherein the software component comprises an echo cancelling software component, and wherein the L blocks include a required echo cancelling block and one or more optional quality-enhancing blocks.
18. The system for allocating processing resources according to claim 11 wherein the processing unit and the data memory are further cooperatively operable for updating the average processing cost (μ i ) and variance (σ i 2 ) for one or more of the L blocks based upon measurements in a preceding time window.
19. A method for allocating processing resources for a number of instances (N) of a software component and a probability of completion (Po) comprising:
in a processing unit;
partitioning the software component into a number of blocks (L blocks), wherein the L blocks include at least one required block and one or more optional blocks;
determining an average processing cost (μ i ) and a variance (σ i 2 ) for each of the L blocks;
computing a scaling factor (α) that is a function of N and the probability of completion (P 0 ) of the L blocks;
estimating an estimated processing cost (m i ) for each of the L blocks as a function of α, and the respective average processing cost (μ i ) and variance (σ i 2 ) of the L blocks; and
allocating processing resources in response to the estimated processing cost m i for each of the L blocks.
20. The method for allocating processing resources according to claim 19 comprising:
adding the estimated processing cost m i of each of the L blocks in each of the N software components to compute a total estimated processing cost for the N software components;
comparing the total estimated processing cost to an available processing value; and
in response to the total estimated processing cost exceeding the available processing value, disabling selected optional blocks to reduce the total estimated processing cost to be equal to or less than the available processing value.