Granted Patent B2

US 12,498,993 · App. 18/753,989 · Granted Dec 16, 2025

Inventor: David Anthony Fontaine (Mountain View, CA)

Assignee: NVIDIA Corporation

G06F9/54G06F8/311G06F9/4494G06F9/543

Abstract

Claims (67)

1 . An acceleration processor unit (APU) comprising:

one or more core complexes, wherein the one or more core complexes include one or more central processing unit (CPU) cores;

one or more graphics complexes, wherein the one or more graphics complexes include one or more compute units (CUs);

an L2 cache;

one or more fabric interconnects;

a memory controller; and

one or more input/output (I/O) interfaces comprising a peripheral component interconnect express (PCIe) interface;

wherein:

the APU executes instructions to perform a stream get capture info (StreamGetCaptureInfo) application program interface (API) to query a stream's capture state;

wherein the StreamGetCaptureInfo API is to use:

a stream parameter to indicate the stream;

a captureStatus_out parameter to return a capture status;

an id_out parameter to indicate a unique identifier;

a graph_out parameter to return a graph being captured into;

a dependencies_out parameter to indicate a pointer to an array of nodes; and

a numDependencies_out parameter to indicate a size of the array of nodes returned in dependencies_out.

2 . The APU of claim 1 , wherein the stream is a stream in capture mode.

3 . The APU of claim 1 , wherein the captureStatus_out parameter indicates a location or a path to return the capture status.

4 . The APU of claim 1 , wherein the capture status indicates whether the stream is in capture mode.

5 . The APU of claim 1 , wherein the unique identifier is a unique identifier of a capture sequence.

6 . The APU of claim 1 , wherein the unique identifier is unique over a lifetime of an application.

7 . The APU of claim 1 , wherein the one or more CUs are to share the L2 cache.

8 . A system comprising:

memory; and

an acceleration processor unit (APU) comprising:

one or more core complexes, wherein the one or more core complexes include one or more central processing unit (CPU) cores;

one or more graphics complexes, wherein the one or more graphics complexes include one or more compute units (CUs);

an L2 cache;

one or more fabric interconnects;

a memory controller; and

one or more input/output (I/O) interfaces comprising a peripheral component interconnect express (PCIe) interface;

wherein:

the APU executes instructions to perform a stream get capture info (StreamGetCaptureInfo) application program interface (API) to query a stream's capture state;

wherein the StreamGetCaptureInfo API is to use:

a stream parameter to indicate the stream;

a captureStatus_out parameter to return a capture status;

an id_out parameter to indicate a unique identifier;

a graph_out parameter to return a graph being captured into;

a dependencies_out parameter to indicate a pointer to an array of nodes; and

a numDependencies_out parameter to indicate a size of the array of nodes returned in dependencies_out.

9 . The system of claim 8 , wherein the stream is a stream in capture mode.

10 . The system of claim 8 , wherein the captureStatus_out parameter indicates a location or a path to return the capture status.

11 . The system of claim 8 , wherein the capture status indicates whether the stream is in capture mode.

12 . The system of claim 8 , wherein the unique identifier is a unique identifier of a capture sequence.

13 . The system of claim 8 , wherein the unique identifier is unique over a lifetime of an application.

14 . The system of claim 8 , wherein the one or more CUs are to share the L2 cache.

15 . A method comprising:

performing, by an acceleration processor unit (APU), a stream get capture info (StreamGetCaptureInfo) application program interface (API) to query a stream's capture state;

wherein the StreamGetCaptureInfo API is to use:

a stream parameter to indicate the stream;

a captureStatus_out parameter to return a capture status;

an id_out parameter to indicate a unique identifier;

a graph_out parameter to return a graph being captured into;

a dependencies_out parameter to indicate a pointer to an array of nodes; and

a numDependencies_out parameter to indicate a size of the array of nodes returned in dependencies_out;

wherein the APU comprises:

one or more core complexes, wherein the one or more core complexes include one or more central processing unit (CPU) cores;

one or more graphics complexes, wherein the one or more graphics complexes include one or more compute units (CUs);

an L2 cache;

one or more fabric interconnects;

a memory controller; and

one or more input/output (I/O) interfaces comprising a peripheral component interconnect express (PCIe) interface.

16 . The method of claim 15 , wherein the stream is a stream in capture mode.

17 . The method of claim 15 , wherein the captureStatus_out parameter indicates a location or a path to return the capture status.

18 . The method of claim 15 , wherein the capture status indicates whether the stream is in capture mode.

19 . The method of claim 15 , wherein the unique identifier is a unique identifier of a capture sequence.

20 . The method of claim 15 , wherein the unique identifier is unique over a lifetime of an application.

Assignments (1)

Continuity (3)

References Cited (53)