# US 2003/0208671 A1 · Data flow processor

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

- Number: US 2003/0208671 A1
- Document ID: US20030208671A1
- Status: Published application
- Kind: A1
- Application number: 10/461,847
- Filed: 2003-06-13
- Published: 2003-11-06
- Earliest claimed: 2000-03-31
- Assignees: Intel Corporation
- Inventors: Randy R. Dunton
- Web page: https://patentgrep.com/patent/US20030208671A1

## CPC

- G06F 13/28: using burst mode transfer, e.g. direct memory access {DMA}, cycle steal (G06F13/32 takes precedence)
- G06F 9/30189: {according to execution mode, e.g. mode flag}
- Y02D 10/00: Energy efficient computing, e.g. low power processors, power management or thermal management
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- Y02D: CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE

## Abstract

A data flow processor includes a number of hardware units each having more than one mode. A plurality of hardware units may be connected together to implement a flow made up of a series of processes. The flows, initiated by a central processing unit, may proceed independently and substantially at their own pace. Thus, the flows may operate in parallel, independently with respect to one another. Each of the hardware units may be configured differently to operate with each of the different flows.

## Claims

### Claim 1

A method comprising:
programming at least two selectable functional modes for each of two reprogrammable digital signal processing devices; and
initiating at least two independent data flows for each of said devices, such that each of said flows use a different mode of at least one of said devices.

### Claim 2 (depends on 1)

The method of claim 1 including identifying a source of data to begin each data flow.

### Claim 3 (depends on 1)

The method of claim 1 including providing the data before initiating the flow.

### Claim 4 (depends on 1)

The method of claim 1 including accessing data in a random access memory and loading said data into channels to be utilized as source data.

### Claim 5 (depends on 4)

The method of claim 4 including enabling said devices to directly access source data.

### Claim 6 (depends on 5)

The method of claim 5 including providing a direct memory access controller and assigning a plurality of channels to provide source data for said flows.

### Claim 7 (depends on 6)

The method of claim 6 including assigning one of said direct memory access channels to respond to read requests from one of said devices.

### Claim 8 (depends on 6)

The method of claim 6 including providing a designated direct memory access channel for the final result of a data flow.

### Claim 9 (depends on 1)

The method of claim 1 including coupling said devices together through a common bus.

### Claim 10 (depends on 9)

The method of claim 9 including coupling said devices together over a packet bus.

### Claim 11 (depends on 10)

The method of claim 10 including providing device identifiers for devices on said bus and setting the output of one of said devices to the device identifier for the other of said devices.

### Claim 12 (depends on 9)

The method of claim 9 including providing an arbiter that is assigns priorities for accessing said bus.

### Claim 13 (depends on 12)

The method of claim 12 including assigning the bus access priorities before initiating said at least two independent data flows.

### Claim 14 (depends on 1)

The method of claim 1 including designating each of said devices to either send, receive or send and receive data.

### Claim 15 (depends on 1)

The method of claim 1 including receiving a string of memory addresses, and converting those addresses into a data structure having at least two dimensions.

### Claim 16 (depends on 15)

The method of claim 15 including forming a data structure having at least three dimensions and using said data structure to model a three-dimensional object.

### Claim 17 (depends on 15)

A method of claim 15 including converting a one dimensional data structure from a memory into a data structure having at least two dimensions for use by said devices and writing data back to said memory as a one dimensional data structure.

### Claim 18

An article comprising a medium storing instructions that cause a processor-based system to:
program at least two selectable functional modes for each of two reprogrammable digital signal processing devices; and
initiate at least two independent data flows for each of said devices, such that each of said flows use a different mode of at least one of said devices.

### Claim 19 (depends on 18)

The article of claim 18 wherein said processor-based system identifies a source of data to begin each data flow.

### Claim 20 (depends on 18)

The article of claim 18 further storing instructions that cause a processor-based system to provide the data before initiating the flow.

### Claim 21 (depends on 18)

The article of claim 18 further storing instructions that cause a processor-based system to access data in a random access memory and load said data into channels to be utilized as source data.

### Claim 22 (depends on 21)

The article of claim 21 further storing instructions that cause a processor-based system to assign a plurality of direct memory access channels to provide source data for said flows.

### Claim 23 (depends on 21)

The article of claim 21 further storing instructions that cause said processor-based system to convert a string of memory addresses into a data structure having at least two dimensions.

### Claim 24 (depends on 23)

The article of claim 23 further storing instructions that cause a processor-based system to form a data structure having at least three dimensions and use said data structure to model a three dimensional object.

### Claim 25 (depends on 23)

The article of claim 23 further storing instructions that cause a processor-based system to convert a one dimensional data structure from a memory into a data structure having at least two dimensions for use by said devices and to cause said data to be written back to said memory as a one dimensional data structure.

### Claim 26

A system comprising:
at least two digital signal processing devices;
a central processing unit;
a bus coupling said devices and said unit; and
each of said devices including a memory programmable by said unit to set each device in one of at least two functional modes.

### Claim 27 (depends on 26)

The system of claim 26 wherein said bus is a packetized bus.

### Claim 28 (depends on 26)

The system of claim 26 wherein said devices are linked to implement a data flow initiated by said unit.

### Claim 29 (depends on 26)

The system of claim 26 including a memory controller coupled to said bus, said memory controller converting addresses into a data structure having at least two dimensions.

### Claim 30 (depends on 29)

The system of claim 29 including a memory coupled to said memory controller, said memory providing a one dimensional data structure and said controller converting said one dimensional data structure into a data structure having at least two dimensions.
