# US 2023/0254269 A1 · Device and method with multi-stage electrical interconnection network

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

- Number: US 2023/0254269 A1
- Document ID: US20230254269A1
- Status: Published application
- Kind: A1
- Application number: 17/941,334
- Filed: 2022-09-09
- Published: 2023-08-10
- Earliest claimed: 2022-02-09
- Assignees: Samsung Electronics Co., Ltd.
- Applicants: Samsung Electronics Co., Ltd.
- Inventors: WONSEOK LEE; Alan Gara; YOUNG JUN HONG; WONYONG LEE; WOOSEOK CHANG
- Web page: https://patentgrep.com/patent/US20230254269A1

## CPC

- H04L 49/1515: Non-blocking multistage, e.g. Clos
- H04Q 3/52: using static devices in switching stages, e.g. electronic switching arrangements
- H04L 5/0001: {Arrangements for dividing the transmission path (two-way operation using the same type of signal, i.e. duplex H04L5/14)}
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)
- H04Q: SELECTING (switches, relays, selectors H01H; wireless communication networks H04W)

## Abstract

A device and method that implements a multi-stage electrical interconnection network is provided. The electronic device includes a plurality of computing devices and a plurality of switches grouped into a plurality of groups. Switches, of the plurality of switches, in a same group are configured to be fully connected to computing devices in the same group, each of switches of the plurality of switches included in a first group among the plurality of groups is configured to have a ono-to-one connection with any one of switches included in a second group, and a connection between the computing devices in the same group and the switches in the same group and a connection between switches in in the plurality of groups are electrical connections.

## Claims

### Claim 1

A device, comprising:
a plurality of computing devices and a plurality of switches respectively grouped into a plurality of groups,
wherein switches, of the plurality of switches, in a same group of the plurality of groups, are configured to be fully connected to computing devices, of the plurality of computing devices, in the same group,
wherein each of multiple switches of the plurality of switches, comprised in a first group among the plurality of groups is configured to have a one-to-one connection with any one of multiple switches of the plurality of switches comprised in a second group among the plurality of groups; and
wherein electrical connections are provided between the computing devices in the same group and the switches in the same group, and electrical connections are provided between respective switches in one or more other groups of the plurality of groups.

### Claim 2 (depends on 1)

The device of claim 1, wherein the switches in the same group are connected to the computing devices in the same group, and
the computing devices in the same group are connected to the switches in the same group.

### Claim 3 (depends on 1)

The device of claim 1, wherein the switches in the same group are connected to each other.

### Claim 4 (depends on 1)

The device of claim 1, wherein a total number of the switches in the same group is
less than or equal to a total number of lanes allocated to each of the computing devices in the same group.

### Claim 5 (depends on 4)

The device of claim 4, wherein the total number of the switches in the same group is
determined based on a result obtained by dividing, by an integer, the total number of lanes.

### Claim 6 (depends on 1)

The device of claim 1, wherein a same total number of lanes allocated to each of the computing devices in the same group is allocated to the switches in the same group.

### Claim 7 (depends on 1)

The device of claim 1, wherein a total number of the computing devices in the same group is
determined based on a total number of the switches in the same group, a total number of lanes allocated to each of the switches in the same group, and a total number of lanes allocated to each of the computing devices in the same group.

### Claim 8 (depends on 1)

The device of claim 1, wherein, for switch connections between groups of the plurality of groups, each of the plurality of groups use a mid-plane printed circuit board (PCB) and a connector connected through a cable.

### Claim 9 (depends on 1)

The device of claim 1, wherein a total number of respective computing devices in each of the plurality of groups is a same number.

### Claim 10 (depends on 1)

The device of claim 1, wherein a total number of respective switches in each of the plurality of groups is a same number.

### Claim 11 (depends on 1)

The device of claim 1, wherein each of the plurality of computing devices comprises:
a processor and/or a high bandwidth memory (HBM).

### Claim 12 (depends on 1)

The device of claim 1, wherein the connections between the computing devices in the same group and the switches in the same group, and the connections between the respective switches are implemented based on a link budget.

### Claim 13 (depends on 12)

The device of claim 12, wherein the link budget is a peripheral component interconnect express (PCIe) link budget.

### Claim 14

A device, comprising:
a plurality of computing devices and a plurality of switches respectively grouped into a plurality of groups,
wherein the plurality of groups are physically separated from each other in respective different circuitries; and
wherein connections between computing devices of the plurality of computing devices in a same group and switches of the plurality of switches in the same group and connections are provided between respective switches in the plurality of groups are based on an electrical interconnection network.

### Claim 15

A method, comprising:
partitioning data to be transmitted from a first computing device in a first group to a second computing device in a second group, and transmitting the partitioned data to first switches in the first group based on a full electrical connection between the first computing device and each of the first switches in the first group;
transmitting the partitioned data to second switches in the second group based on an electrical connection between the first switches in the first group and the second switches in the second group; and
transmitting the partitioned data from the second switches to the second computing device based on a full electrical connection between each of the second switches and the second computing device;
wherein each one of the first switches is configured to have a one-to-one connection with a corresponding one of the second switches.

### Claim 16 (depends on 15)

The method of claim 15, wherein switches in a same group are connected to computing devices in the same group, and
computing devices in the same group are connected to switches in the same group.

### Claim 17 (depends on 15)

The method of claim 15, wherein switches in a same group are not connected to each other.

### Claim 18 (depends on 15)

The method of claim 15, wherein a total number of switches in a same group is less than or equal to a total number of lanes allocated to each computing device in the same group.

### Claim 19 (depends on 18)

The method of claim 18, wherein a total number of the switches in the same group is
determined based on a result obtained by dividing, by an integer, the total number of the lanes.

### Claim 20 (depends on 15)

The method of claim 15, wherein a same total number of lanes allocated to each computing device in a same group is allocated to each switch in the same group.

### Claim 21 (depends on 15)

The method of claim 15, wherein a total number of computing devices in a same group is determined based on a total number of switches in the same group, a total number of lanes allocated to each of the switches, and a total number of lanes allocated to each of the computing devices.

### Claim 22 (depends on 15)

A non-transitory computer-readable storage medium storing instructions that, when executed by a processor, cause the processor to perform the method of claim 15.
