# US 2008/0307205 A1 · Computationally efficient mathematical engine

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2008/0307205 A1
- Document ID: US20080307205A1
- Status: Published application
- Kind: A1
- Application number: 12/194,975
- Filed: 2008-08-20
- Published: 2008-12-11
- Earliest claimed: 2002-09-24
- Assignees: InterDigital Technology Corporation
- Inventors: Ryan Samuel Buchert; Chayil S. Timmerman; Stephan Shane Supplee
- Web page: https://patentgrep.com/patent/US20080307205A1

## CPC

- G06F 5/015: {having at least two separately controlled shifting levels, e.g. using shifting matrices (G06F5/012 takes precedence)}
- G06F 17/16: Matrix or vector computation {, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization (matrix transposition G06F7/78)}
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

A method and apparatus perform many different types of algorithms that utilizes a calculation unit capable of utilizing the same multipliers for different algorithms. The calculation unit preferably includes a processor that has a plural number of arithmetic logic unit circuits that are configured to process data in parallel to provide processed data outputs and an adder tree configured to add the processed data outputs from the arithmetic logic circuits. A shift register that has more parallel data outputs then the processor's inputs is controlled to selectively output data from the parallel outputs to the data inputs of the processor. A communication device preferably includes the calculation unit to facilitate processing of wireless communication signals.

## Claims

### Claim 1

A calculation unit comprising:
a processor having a plural number K of arithmetic logic unit circuits that are configured to process data in parallel to provide processed data outputs, where arithmetic logic unit circuits defines K data inputs of the processor, and having an adder tree configured to add the processed data outputs from the arithmetic logic circuits;
a shift register having a N parallel data outputs where N is greater than K; and
a shift register control circuit configured to control the shift register to selectively output data from no more than K of the N parallel outputs at one time to the K data inputs of the processor.

### Claim 2 (depends on 1)

The calculation unit of claim 1, further comprising a multiplexer configured to receive selectively output data from the shift register and to selectively provide such data from the shift register as input to the K data inputs of the processor.

### Claim 3 (depends on 2)

The calculation unit of claim 2 wherein the shift register has both serial and parallel inputs wherein the shift register control circuit is configured to an enable circuit configured to selectively enable the shift register's serial and parallel inputs.

### Claim 4 (depends on 3)

The calculation unit of claim 3 further comprising by an input memory associated with the shift register and the arithmetic logic unit circuits and a selection circuit configured to selectively enable input to the shift register and the arithmetic logic unit circuits from the input memory based on the type of calculation performed.

### Claim 5 (depends on 4)

The calculation unit of claim 4, further comprising a secondary input memory wherein the selection circuit is configured to selectively enable input from the secondary input memory to the processor based on the type of calculation performed.

### Claim 6 (depends on 1)

The calculation unit of claim 1 wherein the shift register is configured as a selectable memory for receiving input data from at least one input source and the shift register control circuit is configured to provide a selectable output via a plurality of folds, wherein each said fold has K data outputs and comprises at least one different position within the selectable memory.

### Claim 7 (depends on 1)

The calculation unit of claim 1 further including an accumulation circuit configured to receive and selectively accumulate each output from the adder tree based on the type of calculation performed.

### Claim 8 (depends on 1)

The calculation unit of claim 1 further comprising:
a memory configured to receive input data for complex resolution;
a store configured to store an operational factor for the complex function;
a multiplexer configured to selectively receive input from the shift register via the memory or the store;
a processing array circuit as the arithmetic logic unit circuits configured to process data from selected bit locations stored by the memory and data output from the multiplexer; and
an accumulator circuit configured to receive an output from the adder tree and provide an accumulated complex output.

### Claim 9 (depends on 8)

The calculation unit of claim 8 wherein the store is configured to provide a twiddle factor as the operational factor, for performing discrete Fourier transforms, wherein the multiplexer receives its input from the store when using the twidde factor.

### Claim 10 (depends on 9)

The calculation unit of claim 9 wherein said processor is configured to process data that includes both real and imaginary components.

### Claim 11 (depends on 10)

A communication device including the calculation unit of claim 10 configured to facilitate processing of wireless communication signals.

### Claim 12 (depends on 1)

The calculation unit of claim 1 wherein said processor is configured to process data that includes both real and imaginary components.

### Claim 13 (depends on 1)

A communication device including the calculation unit of claim 1 configured to facilitate processing of wireless communication signals.

### Claim 14

A method of processing data comprising:
providing a processor that has an adder tree and a plural number K of arithmetic logic unit circuits that define K data inputs of the processor;
providing a shift register that has N parallel data outputs, where N is greater than K;
receiving the data to be processed in the shift register;
controlling the shift register to selectively output data through no more than K of the N data outputs at one time to the K inputs of the processor;
using the arithmetic logic unit circuits of the processor to parallel process data output from the shift register and to provide processed data outputs; and
using the adder tree to add the processed data outputs from the arithmetic logic unit circuits.

### Claim 15 (depends on 14)

The method of claim 14 further comprising using a multiplexer to receive selectively output data from the shift register and to selectively provide such data from the shift register to the arithmetic logic unit circuits.

### Claim 16 (depends on 15)

The method of claim 15 wherein the shift register has both serial and parallel inputs further comprising selectively enabling the shift register's serial and parallel inputs based on the type of data processing computations to be performed.

### Claim 17 (depends on 14)

The method of claim 14 wherein an input memory is associated with the shift register and the processor arithmetic logic unit circuits, further comprising selectively enabling input to the shift register and the processor arithmetic logic unit circuits from the input memory based on the type of calculation performed.

### Claim 18 (depends on 17)

The method claim 17 further comprising selectively enabling input from a secondary input memory to the processor based on the type of calculation performed.

### Claim 19 (depends on 14)

The method of claim 14 wherein the data which is processed is data that includes both real and imaginary components.

### Claim 20 (depends on 14)

The method of claim 14 performed in a communication device to facilitate processing of wireless communication signals.
