Variable length fractional bandwidth low-pass filtering

• US 6,058,406 A
• Filed: 03/23/1998
• Issued: 05/02/2000
• Est. Priority Date: 03/23/1998
• Status: Expired due to Term

• Current Assignee: ATI Technologies ULC (Advanced Micro Devices, Inc.)
• Original Assignee: ATI Technologies Incorporated (Advanced Micro Devices, Inc.)
• Inventors: Karanovic, Marinko
• Primary Examiner(s): Malzahn, David H.
• Application Number:
  US09/046,180
• Time in Patent Office: 771 Days
• Field of Search: 708/313
• US Class Current: 708/313
• CPC Class Codes: H03H 21/0012 Digital adaptive filters
• CPC Class Codes: H03L 7/1974 for fractional frequency division

Abstract

A variable length fractional bandwidth low-pass filter is implemented as a plurality of filter stages. Each filter stage performs fractional filtering of a plurality of input samples to form an output sample. The number of filter stages is determined based on the characteristic of the filter stage, and the load is suitably balanced to allow for a minimal number of filter stages. Load balancing is applied so as to utilize each of the filter stages only when required. The load balancing is applied so as to provide a substantially equal load to each filter stage, and the number of filter stages is determined so as to allow the filter to operate substantially at the input sample rate.

Claims

1. A variable length fractional filter comprising:

a plurality of filter stages operably coupled to perform fractional filtering on a plurality of input samples to produce a plurality of output samples; and

a load balancer operably coupled to balance the plurality of filter stages to each filter a substantially equal number of the plurality of input samples, based on a receive scaling factor data.

2. The filter of claim 1, wherein each filter stage of the plurality of filter stages is an accumulating finite impulse response (FIR) filter.

3. The filter of claim 1, wherein

each of the plurality of input samples occur at a first periodic interval, and

each of the plurality of filter stages produce an output sample, such that the output of the plurality of filter stages produces the plurality of output samples occurring at a second periodic interval which is substantially equal to a product of the scaling factor data and the first periodic interval.

4. The filter of claim 3, wherein the plurality of input samples corresponds to a video image at a first resolution, and the plurality of output samples correspond to the video image at a second resolution.

5. The filter of claim 1, also comprising

a coefficient generator operably coupled to the plurality of filter stages to provide coefficients based on phase differences between the input samples and the output samples.

6. The filter of claim 1, wherein the load balancer includes

a discrete time oscillator to provide a series of trigger pulses, and

a counter operably coupled to receive the series of trigger pulses and to produce a plurality of enable signals, each of the enable signals of the plurality of enable signals corresponding to one of the plurality of filter stages.

7. The filter of claim 6, wherein

the discrete time oscillator also provides a series of initial phase values corresponding to each trigger pulse;

and

each filter stage of the plurality of filter stages is operably coupled to receive the corresponding one of the enable signals, and to receive a corresponding initial phase value of the plurality of phase values in dependence upon the corresponding one of the enable signals.

8. The filter of claim 7, also comprising

a coefficient generator operably coupled to each of the plurality of filter stages to provide coefficients based on the scaling factor and the initial phase value corresponding to the each of the plurality of filter stages.

9. The filter of claim 8, wherein

the coefficient generator includes a plurality of predetermined coefficient values.

10. The filter of claim 1, wherein the load balancer is configured to allocate a substantially equal computational load to each of the plurality of filter stages.

11. A method for variable length fractional filtering comprising the steps of:

balancing the allocation of a plurality of filter stage processes to filter a substantially equal number of a plurality of input samples based on a received scaling factor data, and

performing fractional filtering on the plurality of input samples within each of the plurality of filter stage processes to produce a plurality of output samples.

12. The method of claim 11, wherein the plurality of input samples corresponds to a video image at a first resolution, and the plurality of output samples correspond to the video image at a second resolution.

13. The method of claim 11, wherein the step of performing fractional filtering includes the steps of:

generating coefficients based on phase differences between the input samples and the output samples corresponding to each of the substantially equal number of input samples, and

accumulating a product of each coefficient and the corresponding each of the substantially equal number of input samples to produce each output sample of the plurality of output samples.

14. The method of claim 13, wherein the step of balancing the allocation of the plurality of filter stage processes includes the steps of:

providing a series of trigger pulses corresponding to each output sample, and

processing the trigger pulses to produce a plurality of enable signals, each of the enable signals of the plurality of enable signals corresponding to one of the plurality of filter stage processes.

15. The method of claim 11, including the steps of:

determining an enable value and an initial phase value corresponding to each of the plurality of input samples;

filtering the substantially equal number of the plurality of input samples when the enable value is a first enable value, said filtering being based on the initial phase value.

16. The method of claim 15, wherein the step of filtering the substantially equal number of the plurality of input samples includes the steps of:

determining a plurality of coefficient values based on the initial phase value, each coefficient value of the plurality of coefficient values corresponding to one of the substantially equal number of the plurality of input samples;

multiplying each one of input samples by the corresponding each of the coefficient values to produce a plurality of products; and

accumulating each of the plurality of products to produce a filtered output sample.

17. The method of claim 15, wherein the step of determining the enable value and the initial phase value includes the steps of:

adding a predetermined increment value to an accumulated value;

setting the enable value to the first enable value when the accumulated value equals or exceeds a maximum value, and setting the enable value to a second enable value when the accumulated value is less than the maximum value;

subtracting the maximum value from the accumulated value when the accumulated value equals or exceeds the maximum value; and

setting the initial phase value equal to the accumulated value.

18. The method of claim 17, wherein the step of filtering the substantially equal number of the plurality of input samples includes the steps of:

determining a plurality of coefficient values based on the initial phase value, each coefficient value of the plurality of coefficient values corresponding to one of the plurality of input samples;

multiplying each of input samples by the corresponding each of the coefficient values to produce a plurality of products; and

accumulating each of the plurality of products to produce a filtered output sample.

19. An apparatus for fractionally filtering a plurality of input samples, comprising:

means for detennining an enable value to selectively enable a plurality of means for filtering and for detenmining an initial phase value corresponding to each of the plurality of input samples; and

a plurality of means, operably responsive to the means for determining the enable value, for filtering a predetermined number of the plurality of input samples when the enable value is a first enable value to produce an output sample, the filtering being based on the initial phase value.

20. The apparatus of claim 19, wherein each of the plurality of means for filtering the predetermined number of the plurality of input samples includes:

means for determining a plurality of coefficient values based on the initial phase value, each coefficient value of the plurality of coefficient values corresponding to each of the predetermined number of the plurality of input samples;

means for multiplying each of the predetermined number of the plurality of input samples by the corresponding each of the coefficient values to produce a plurality of products; and

means for accumulating each of the plurality of products to produce a filtered output sample.

21. The apparatus of claim 19, wherein the means for determining the enable value and the initial phase value includes:

means for adding a given increment value to an accumulated value corresponding to each of the plurality of input samples;

means operably coupled to the means for adding for setting the enable value to the first enable value when the accumulated value equals or exceeds a maximum value, and setting the enable value to a second enable value when the accumulated value is less than the maximum value;

means operably coupled to the means for adding for subtracting the maximum value from the accumulated value when the accumulated value equals or exceeds the maximum value; and

means operably coupled to the means for subtracting for setting the initial phase value equal to the accumulated value.

22. The apparatus of claim 21, wherein each of the plurality of means for filtering the predetermined number of the plurality of input samples includes:

means for determining a plurality of coefficient values based on the initial phase value, each coefficient value of the plurality of coefficient values corresponding to each of the predetermined number of the plurality of input samples;

means for multiplying each of the predetermined number of the plurality of input samples by the corresponding each of the coefficient values to produce a plurality of products; and

means for accumulating each of the plurality of products to produce a filtered output sample.