# US 2023/0276301 A1 · System and method for gpu-initiated communication

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

- Number: US 2023/0276301 A1
- Document ID: US20230276301A1
- Status: Published application
- Kind: A1
- Application number: 17/854,522
- Filed: 2022-06-30
- Published: 2023-08-31
- Earliest claimed: 2022-02-27
- Assignees: NVIDIA Corporation
- Applicants: Nvidia Corporation
- Inventors: Sreeram POTLURI; Davide ROSSETTI; Elena AGOSTINI; Pak MARKTHUB; Daniel MARCOVITCH; Sayantan SUR
- Web page: https://patentgrep.com/patent/US20230276301A1

## CPC

- H04W 28/06: Optimizing {the usage of the radio link}, e.g. header compression, information sizing {, discarding information (system modifying transmission characteristic according to link quality by modifying frame length H04L1/0007; dynamic adaptation of the packet size for flow control or congestion control H04L47/365)}
- G06F 9/3836: {Instruction issuing, e.g. dynamic instruction scheduling or out of order instruction execution}
- G06F 9/3877: {using a secondary processor, e.g. coprocessor (peripheral processor G06F13/12)}
- G06T 1/20: Processor architectures; Processor configuration, e.g. pipelining
- H04W 24/02: Arrangements for optimising operational condition
- H04W 72/56: based on priority criteria
- H04W: WIRELESS COMMUNICATION NETWORKS (broadcast communication H04H; communication systems using wireless links for non-selective communication, e.g. wireless extensions H04M1/72)
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL

## Abstract

A computer based system and method for sending data packets over a data network may include: preparing data packets and packet descriptors on one or more graphical processing units (GPUs); associating packets with a packet descriptor, which may determine a desired transmission time of the packets associated with that descriptor; receiving an indication of a clock time; and physically transmitting packets via an output interface, at a clock time corresponding to the desired transmission time. A computer based system and method for GPU-initiated communication over a 5G data network may include allocating one or more memory buffers in GPU memory; performing at least one 5G signal processing procedure by a GPU; preparing descriptors for a plurality of packets, where each packet includes allocated memory buffers, and where the descriptors provide scheduling instructions for the packets; and triggering the sending of packets over the network based on prepared descriptors.

## Claims

### Claim 1

A method for sending data packets over a data network, the method comprising:
preparing, by a graphical processing unit (GPU), one or more descriptors for one or more packets, each packet comprising one or more memory buffers;
storing, by the GPU, one or more of the descriptors and the packets in GPU memory; and
triggering, by the GPU, the sending of one or more of the packets, the sending based on one or more of the descriptors.

### Claim 2 (depends on 1)

The method of claim 1, wherein triggering comprises updating one or more registers of a network adapter.

### Claim 3 (depends on 2)

The method of claim 2, comprising receiving, by the GPU, one or more incoming packets and storing the incoming packets in GPU memory; and
performing, by the network adapter, a read operation on the incoming packets to initiate a processing procedure of the incoming packets by the GPU.

### Claim 4 (depends on 1)

The method of claim 1, wherein one or more of the descriptors comprises scheduling instructions for the one or more packets, the scheduling instructions comprising at least one of: send at a specific time, and do not send until one or more of the packets include predefined data.

### Claim 5 (depends on 4)

The method of claim 4, comprising transmitting, using the network adapter, one or more of the packets over the network based on the scheduling instructions.

### Claim 6 (depends on 5)

The method of claim 5, wherein the transmission of the one or more packets is determined by predefined priorities.

### Claim 7 (depends on 1)

The method of claim 1, wherein the sending of the one or more packets comprises mapping memory addresses between different memory components using a dedicated driver.

### Claim 8 (depends on 1)

The method of claim 1, comprising issuing one or more memory barriers between one or more procedures including at least one of: packet preparation, receiving of packets, processing of packets, packet descriptor preparation, and packet transmission.

### Claim 9

A system comprising:
a memory;
a GPU memory unit; and
one or more graphical processing units (GPUs) to:
prepare one or more descriptors for one or more packets, each packet comprising one or more memory buffers;
store one or more of the descriptors and the packets in the GPU memory unit; and
trigger the sending of the one or more packets, the sending based on one or more of the descriptors.

### Claim 10 (depends on 9)

The system of claim 9, wherein the one or more of the GPUs is to update one or more registers of a network adapter.

### Claim 11 (depends on 9)

The system of claim 9, further comprising:
a network adapter;
wherein one or more of the GPUs is to receive one or more incoming packets and store the incoming packets in the GPU memory unit; and
wherein the network adapter is to perform a read operation on the incoming packets to initiate a processing procedure of the incoming packets by the GPU.

### Claim 12 (depends on 9)

The system of claim 9, wherein one or more of the descriptors comprise scheduling instructions for one or more of the packets, the scheduling instructions comprising at least one of: send at a specific time, and do not send until the one or more packets include predefined data.

### Claim 13 (depends on 12)

The system of claim 12, further comprising:
a network adapter;
wherein the network adapter is to transmit the one or more packets over the network based on the scheduling instructions.

### Claim 14 (depends on 13)

The system of claim 13, wherein the network adapter is to determine the transmission of one or more of the packets by predefined priorities.

### Claim 15 (depends on 9)

The system of claim 9, wherein the sending of the one or more packets comprises mapping memory addresses between different memory components using a dedicated driver.

### Claim 16 (depends on 9)

The system of claim 9, wherein the GPU is to issue one or more memory barriers between one or more procedures including at least one of: packet preparation, receiving of packets, processing of packets, packet descriptor preparation, and packet transmission.

### Claim 17

A method for creating and transmitting data packets over a data network, the method comprising:
creating, by one or more graphical processing units (GPUs), one or more data packets, each packet comprising one or more memory buffers;
associating, by one or more of the GPUs, one or more of the packets with one or more descriptors; and
triggering, by one or more of the GPUs, the transmission of one or more of the packets, the transmission based on one or more of the descriptors.

### Claim 18 (depends on 17)

The method of claim 17, wherein the triggering comprises setting one or more registers of a network adapter.

### Claim 19 (depends on 17)

The method of claim 17, wherein one or more of the descriptors comprises scheduling instructions for the one or more packets, the scheduling instructions comprising at least one of: transmit at a specific time, and do not transmit until one or more of the packets include predefined data.

### Claim 20 (depends on 19)

The method of claim 19, comprising scheduling, using the network adapter, the transmission of one or more of the packets over the network based on the scheduling instructions.
