# US 2008/0288950 A1 · Concurrent Management of Adaptive Programs

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

- Number: US 2008/0288950 A1
- Document ID: US20080288950A1
- Status: Published application
- Kind: A1
- Application number: 11/750,441
- Filed: 2007-05-18
- Published: 2008-11-20
- Earliest claimed: 2007-05-18
- Assignees: Intel Corporation
- Inventors: Matthew Hammer; Mohan Rajagopalan; Anwar Ghuloum
- Web page: https://patentgrep.com/patent/US20080288950A1

## CPC

- G06F 9/5066: {Algorithms for mapping a plurality of inter-dependent sub-tasks onto a plurality of physical CPUs (mappping at compile time, see G06F8/451)}
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

A method for concurrent management of adaptive programs is disclosed wherein changes in a set of modifiable references are initially identified. A list of uses of the changed references is next computed using records made in structures of the references. The list is next inserted into an elimination queue. Comparison is next made of each of the uses to the other uses to determine independence or dependence thereon. Determined dependent uses are eliminated and the preceding steps are repeated for all determined independent uses until all dependencies have been eliminated.

## Claims

### Claim 1

A method for an adaptive computer program comprising:
populating one or more modifiable references with task information;
employing idle resources to cooperate with said one or more modifiable references and process assigned tasks in a background mode using said task information; and
providing task results on an as-needed basis.

### Claim 2 (depends on 1)

The evaluation method of claim 1 wherein said one or more modifiable references are unpopulated or contain only ready-to-read values or values resulting from one or more previously assigned tasks immediately prior to said populating.

### Claim 3 (depends on 2)

The evaluation method of claim 2 further comprising;
re-populating one or more of said one or more modifiable references with continuing task information; and
providing continuing task results on an as-needed basis.

### Claim 4 (depends on 3)

The evaluation method of claim 3 wherein said idle resources are taken from the group including processors and hardware threads.

### Claim 5 (depends on 4)

The evaluation method of claim 4 wherein said one or more modifiable references comprise signatures based on destination-passing-style methods.

### Claim 6 (depends on 5)

The evaluation method of claim 5 further comprising:
tracking system state changes by said one or more modifiable references.

### Claim 7 (depends on 6)

The evaluation method of claim 6 further comprising:
abstracting out an execution trace as a binary tree of said one or more modifiable references, said trace describing nodes where data is created and consumed, and where different control and data-flow dependencies leading to said nodes are created and consumed.

### Claim 8 (depends on 7)

The evaluation method of claim 7 further comprising:
exposing said dependencies through sequence and split primitives.

### Claim 9 (depends on 8)

The evaluation method of claim 8 further comprising:
using said exposed dependencies to identify nodes where computations may be parallelized.

### Claim 10 (depends on 9)

The evaluation method of claim 9 further comprising:
parallelizing said computations.

### Claim 11

An article comprising an adaptive computer program that if executed enables a system to:
populate one or more modifiable references with task information;
employ idle resources to cooperate with said one or more modifiable references and process assigned tasks in a background mode using said task information; and
provide task results on an as-needed basis.

### Claim 12 (depends on 11)

The article of claim 11 wherein said one or more modifiable references are unpopulated or contain only ready-to-read values or values resulting from one or more previously assigned tasks immediately prior to said populating.

### Claim 13 (depends on 12)

The article of claim 12 wherein said computer program further enables said system to;
re-populate one or more of said one or more modifiable references with continuing task information; and
provide continuing task results on an as-needed basis.

### Claim 14 (depends on 13)

The article of claim 13 wherein said idle resources are taken from the group including processors and hardware threads.

### Claim 15 (depends on 14)

The article of claim 14 wherein said one or more modifiable references comprise signatures based on destination-passing-style methods.

### Claim 16 (depends on 15)

The article of claim 15 wherein said computer program further enables said system to:
track system state changes by said one or more modifiable references.

### Claim 17 (depends on 16)

The article of claim 16 wherein said computer program further enables said system to:
abstract out an execution trace as a binary tree of said one or more modifiable references, said trace describing nodes where data is created and consumed, and where different control and data-flow dependencies leading to said nodes are created and consumed.

### Claim 18 (depends on 17)

The article of claim 17 wherein said computer program further enables said system to:
expose said dependencies through sequence and split primitives.

### Claim 19 (depends on 18)

The article of claim 18 wherein said computer program further enables said system to:
using said exposed dependencies to identify nodes where computations are parallelized.

### Claim 20

A method for concurrent management of adaptive programs comprising:
identification of changes in a set of modifiable references;
computation of a list of uses of said changed references using records made in structures of said references;
insertion of said list into an elimination queue;
comparison of each of said uses of said list to other uses of said list to determine independence or dependence thereon;
elimination of said determined dependent uses; and
repetition of said identification, computation, insertion and comparison for all determined independent uses until all dependencies have been eliminated.
