# US 2021/0173732 A1 · Redundant processing node changing method and processor capable of changing redundant processing node

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

- Number: US 2021/0173732 A1
- Document ID: US20210173732A1
- Status: Published application
- Kind: A1
- Application number: 16/729,288
- Filed: 2019-12-27
- Published: 2021-06-10
- Earliest claimed: 2019-12-04
- Assignees: Industrial Technology Research Institute
- Applicants: Industrial Technology Research Institute
- Inventors: Li-Ren HUANG; Chi-Chun HSU
- Web page: https://patentgrep.com/patent/US20210173732A1

## CPC

- G06F 11/0751: {Error or fault detection not based on redundancy (power supply failures G06F1/30; network fault management H04L41/06)}
- G06F 11/0724: {in a multiprocessor or a multi-core unit (multiprocessors per se G06F15/80)}
- G06F 11/0772: {Means for error signaling, e.g. using interrupts, exception flags, dedicated error registers}
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

A processor capable of changing redundant processing node comprises a plurality of processing nodes and a plurality of comparators. The plurality of processing nodes comprises a first processing node, a second processing node, and a third processing node, wherein the first processing node performs a first computation, the second processing node selectively performs the first computation or a second computation, and the third processing node performs the second computation. The plurality of comparators comprises a first comparator and a second comparator, wherein the first comparator connects to the first and second processing nodes to compare whether the results of the first computation performed by the first and second processing nodes are identical, and the second comparator connects to the second and third processing nodes to compare whether results of the second computation performed by the second and third processing nodes are identical.

## Claims

### Claim 1

A redundant processing node changing method, wherein the method is configured to be adapted to a processor comprising a plurality of processing nodes and a plurality of comparators, and the method comprises:
selecting a first node pair from the plurality of processing nodes;
setting two processing nodes of the first node pair as a first processing node and a second processing node in a verification procedure and performing the verification procedure;
selecting a second node pair from the plurality of processing nodes; and
after the verification procedure is performed with the first node pair, setting two processing nodes of the second node pair as the first processing node and the second processing node in the verification procedure and performing the verification procedure;
wherein the verification procedure comprises:
preforming a computation by the first processing node to generate a first result;
preforming the computation by the second processing node to generate a second result;
comparing the first result and the second result by a comparator, wherein the comparator is one of the plurality of comparators that connects to the first processing node and the second processing node;
outputting an error signal by the comparator when the first result is different from the second result; and
outputting the first result or the second result when the first result is identical to the second result.

### Claim 2 (depends on 1)

The redundant processing node changing method of claim 1, wherein
before selecting the first node pair from the plurality of processing nodes, the method further comprises determining a level of parallelism;
selecting the first node pair from the plurality of processing nodes comprises selecting a first pair set from the plurality of processing nodes according to the level of parallelism, with the first pair set comprising the first node pair;
performing the verification procedure with the first node pair comprises performing the verification procedure concurrently by every pair of the first pair set;
selecting the second node pair from the plurality of processing nodes comprises selecting a second pair set from the plurality of processing nodes according to the level of parallelism, with the second pair set comprising the second node pair; and
performing the verification procedure with the second node pair comprises performing the verification procedure concurrently by every pair of the second pair set.

### Claim 3 (depends on 2)

The redundant processing node changing method of claim 2, wherein both of a number of pairs of the first pair set and a number of pairs of the second pair set equals to the level of parallelism.

### Claim 4 (depends on 2)

The redundant processing node changing method of claim 2, wherein the level of parallelism is associated with a fault tolerance time interval.

### Claim 5 (depends on 1)

The redundant processing node changing method of claim 1, before selecting the first node pair and the second node pair from the plurality of processing nodes, further comprising:
determining a running way, wherein the running way is a route along which the verification procedure is performed for a plurality of times.

### Claim 6 (depends on 5)

The redundant processing node changing method of claim 5, further comprising:
determining another running way after the plurality of times of the verification procedure is finished.

### Claim 7 (depends on 1)

The redundant processing node changing method of claim 1, wherein at least one of the two processing nodes of the second node pair is different from the two processing nodes of the first node pair.

### Claim 8 (depends on 1)

The redundant processing node changing method of claim 1, wherein the comparator connecting to the first node pair is identical to the comparator connecting to the second node pair.

### Claim 9 (depends on 1)

The redundant processing node changing method of claim 1, wherein the comparator connecting to the first node pair is different from the comparator connecting to the second node pair.

### Claim 10 (depends on 1)

The redundant processing node changing method of claim 1, wherein the plurality of processing nodes of the processor are arranged in a form of a matrix and the processing nodes of the first node pair and the second node pair are in a single row or in a single column of the matrix.

### Claim 11 (depends on 1)

The redundant processing node changing method of claim 1, wherein the plurality of processing nodes of the processor are arranged in a form of a matrix, the two processing nodes of the first node pair are in two rows and two columns of a sub-matrix, and the two processing nodes of the second node pair are in said two rows and said two columns of the sub-matrix.

### Claim 12

A processor capable of changing redundant processing node comprising:
a plurality of processing nodes comprising a first processing node, a second processing node, and a third processing node, wherein:
the first processing node is configured to perform a first computation;
the second processing node is configured to selectively perform the first computation or a second computation; and
the third processing node is configured to perform the second computation; and
a plurality of comparators comprising a first comparator and a second comparator; wherein:
the first comparator connects to the first processing node and the second processing node to compare whether a result of the first computation performed by the first processing node is identical to a result of the first computation performed by the second processing node; and
the second comparator connects to the second processing node and the third processing node to compare whether a result of the second computation performed by the second processing node is identical to a result of the second computation performed by the third processing node.

### Claim 13 (depends on 12)

The processor capable of changing the redundant processing node of claim 12, wherein the first comparator comprises a first output port and a second output port, the first output port is configured to output the result of the first computation when the results of the first computation performed by the first and second processing nodes are identical, and the second output port is configured to output an error signal when the results of the first computation performed by the first and second processing nodes are different.

### Claim 14 (depends on 13)

The processor capable of changing the redundant processing node of claim 13, wherein the first comparator and the second comparator have an identical architecture or an identical function.

### Claim 15 (depends on 12)

The processor capable of changing the redundant processing node of claim 12, wherein each of the comparators further comprises a n-to-2 multiplexer, wherein a number of the processing nodes connecting to the n-to-2 multiplexer is greater than two.

### Claim 16 (depends on 12)

The processor capable of changing the redundant processing node of claim 12, wherein a number of the plurality of comparators, a number of the plurality of processing nodes, and a connection type of the plurality of processing nodes and the plurality of comparators are associated with a level of parallelism, wherein the level of parallelism indicates a number of at least one computation performed by one or more redundant processing node at a time.

### Claim 17 (depends on 12)

The processor capable of changing the redundant processing node of claim 12, wherein a connection type of the plurality of processing nodes and the plurality of comparators is associated with a traversal policy, wherein the traversal policy is configured to change an order of the redundant processing node.

### Claim 18 (depends on 12)

The processor capable of changing the redundant processing node of claim 12, wherein
the plurality of processing nodes further comprises a fourth processing node connecting to the first comparator and the second comparator, and the fourth processing node is configured to perform a computation identical to the computation performed by the second processing node;
the first comparator is further configured to compare whether the results of the first computation respectively performed by the first processing node, the second processing node, and the fourth processing node are identical; and
the second comparator is further configured to compare whether the results of the second computation performed by the second processing node, the third processing node, and the fourth processing node are identical.
