Granted Patent B1

US 12,639,054 · App. 18/114,870 · Granted May 26, 2026

Inventors: David Anthony Fontaine (Mountain View, CA); Steven Arthur Gurfinkel (San Jose, CA)

Assignee: NVIDIA Corporation

G06F8/433G06F9/52

Abstract

Claims (32)

1 . A computer-implemented method comprising:

performing an application programming interface (API) to indicate one or more graph node functions of one or more graph nodes to be added to a software graph based, at least in part, on a dependency type indicated by the API;

wherein the API is to indicate one or more graph node functions, based, at least in part, on a graph node type indicated by one or more node parameters;

wherein the dependency type includes one or more of a full execution dependency, a launch order dependency, a fast dependent launch dependency, or an anti-deadlock dependency; and

wherein the API is to receive a set of parameters comprising an identifier of the one or more graph nodes and dependency information corresponding to the dependency type.

2 . The computer-implemented method of claim 1 , wherein the API is to indicate one or more graph node functions, based, at least in part, on a graph node type indicated by one or more node parameters.

3 . The computer-implemented method of claim 1 , wherein the set of parameters comprises an indication of the software graph.

4 . The computer-implemented method of claim 1 , wherein the API is to cause the one or more graph nodes to be added to the software graph by identifying one or more nodes of the software graph dependent to or dependent on the one or more graph nodes.

5 . The computer-implemented method of claim 1 , wherein the dependency type indicates one or more constraints on scheduling a first operation corresponding to the one or more graph nodes and a second operation corresponding to a second graph node.

6 . The computer-implemented method of claim 1 , wherein the graph node functions include one or more of a memory operation, a kernel operation, a child graph operation, a null-operation, a variable operation, or an event operation.

7 . The computer-implemented method of claim 1 , wherein performing the API is to cause a second one or more nodes to be added to the software graph based on the set of parameters.

8 . A computer system comprising:

one or more processors and memory storing executable instructions that, if performed by the one or more processors, are to perform an application programming interface (API) to indicate one or more graph node functions of one or more graph nodes to be added to a software graph based, at least in part, on a dependency type indicated by the API;

wherein the API is to indicate one or more graph node functions, based, at least in part, on a graph node type indicated by one or more node parameters;

wherein the dependency type includes one or more of a full execution dependency, a launch order dependency, a fast dependent launch dependency, or an anti-deadlock dependency; and

wherein the API is to receive a set of parameters comprising an identifier of the one or more graph nodes and dependency information corresponding to the dependency type.

9 . The computer system of claim 8 , wherein the set of parameters comprises an indication of the software graph.

10 . The computer system of claim 8 , wherein the API is to further indicate a port identifier of a node associated with the one or more graph nodes.

11 . The computer system of claim 8 , wherein the API is to cause the one or more graph nodes to be added to the software graph by identifying one or more nodes of the software graph dependent to or dependent on the one or more graph nodes.

12 . The computer system of claim 8 , wherein the dependency type indicates one or more constraints on scheduling a first operation corresponding to the one or more graph nodes and a second operation corresponding to a second graph node.

13 . The computer system of claim 8 , wherein the graph node functions include one or more of a memory operation, a kernel operation, a child graph operation, a null-operation, a variable operation, or an event operation.

14 . One or more processors comprising:

circuitry to perform an application programming interface (API) to indicate one or more graph node functions of one or more graph nodes to be added to a software graph based, at least in part, on a dependency type indicated by the API;

wherein the API is to indicate one or more graph node functions, based, at least in part, on a graph node type indicated by one or more node parameters;

wherein the dependency type includes one or more of a full execution dependency, a launch order dependency, a fast dependent launch dependency, or an anti-deadlock dependency; and

wherein the API is to receive a set of parameters comprising an identifier of the one or more graph nodes and dependency information corresponding to the dependency type.

15 . The one or more processors of claim 14 , wherein the set of parameters comprises an indication of the software graph.

16 . The one or more processors of claim 14 , wherein the API is to further indicate a port identifier of a node associated with the one or more graph nodes.

17 . The one or more processors of claim 14 , wherein the API is to cause the one or more graph nodes to be added to the software graph by identifying one or more nodes of the software graph dependent to or dependent on the one or more graph nodes.

18 . The one or more processors of claim 14 , wherein the dependency type indicates one or more constraints on scheduling a first operation corresponding to the one or more graph nodes and a second operation corresponding to a second graph node.

19 . The one or more processors of claim 14 , wherein the graph node functions include one or more of a memory operation, a kernel operation, a child graph operation, a null-operation, a variable operation, or an event operation.

20 . The one or more processors of claim 14 , wherein performing the API is to cause a second one or more nodes to be added to the software graph based on the set of parameters.

Assignments (1)

References Cited (95)