Granted Patent B2

US 12,445,528 · App. 18/515,082 · Granted Oct 14, 2025

Inventors: David Lawrence Rager (Austin, TX); Andrew Edward Brownsword (Bowen Island, CA); Guy Lewis Steele, Jr. (Lexington, MA)

Assignee: Oracle International Corporation

H04L67/306G06F9/3009G06F9/4881G06F9/52G06F16/2237G06F16/2246G06F16/258G06F16/9027G06Q30/0201G06Q30/0269G06T1/20G06Q50/01

Abstract

Claims (57)

1. A computer-implemented method comprising:

accessing a set of user profiles each user profile of the set of user profiles including one or more tags, each tag of the one or more tags characterizing a user based on an interaction between a user device and a web server or an application server;

determining that one or more characteristics of each user profile of the set of user profiles satisfies an offloading condition that is a condition to transfer the set of user profiles from a cloud service to a graphical processing unit (GPU);

in response to determining that the offloading condition is satisfied:

transforming, using the GPU, the set of user profiles into a plurality of subsets of user profiles;

storing, for each subset of the plurality of subsets of user profiles, the tags for the subset of user profiles in in a manner that accounts for the tags and a number of threads in a warp of the GPU, wherein a number of threads in the warp of the GPU equals a number of user profiles in the subset of user profiles;

determining, using the warps of the GPU, one or more user profiles of the set of user profiles that satisfy a logic of one or more logic trees, wherein the one or more user profiles are determined based on one or more tags of one or more specialized data structures associated with the subsets of the set of users and one or more operand values of the one or more logic trees; and

outputting the one or more user profiles.

2. The computer-implemented method of Claim 1 , further comprising:

creating at least one array, wherein the tags for the subset of user profiles are stored in or in accordance with the at least one array, wherein the at least one array stores tags for the subset of user profiles in a sorted order, wherein tags for a user profile are arranged sequentially in the at least one array according to identifiers.

3. The computer-implemented method of Claim 1 , further comprising:

creating at least one array, wherein the tags for the subset of user profiles are stored in or in accordance with the at least one array, wherein the at least one array stores tags for the subset of user profiles in an interleaved manner, wherein tags for a user profile are interleaved with tags for another user profile in the at least one array according to identifiers.

4. The computer-implemented method of claim 1 , wherein, for each subset of the plurality of subsets of user profiles, each thread of the number of threads in the warp is configured to execute instructions running on a core of the GPU.

5. The computer-implemented method of claim 1 , wherein determining, using the warps of the GPU, one or more user profiles of the set of user profiles that satisfy logic of one or more logic trees comprises reading a specialized data structure of the specialized data structures with a warp of the GPU.

6. The computer-implemented method of claim 1 , further comprising:

accessing at least one array, wherein the tags for the subset of user profiles are stored in or in accordance with the at least one array,

wherein determining, using the warps of the GPU, one or more user profiles of the set of user profiles that satisfy logic of one or more logic trees comprises:

retrieving tags from an array of the at least one array;

identifying one or more operand values of a logic tree of the one or more logic trees that references the retrieved tags; and

setting the logic tree of the one or more logic trees as true.

7. A system comprising:

one or more processors;

one or more non-transitory computer-readable media storing instructions, which, when executed by the system, cause the system to perform a set of actions including:

accessing a set of user profiles, each user profile of the set of user profiles including one or more tags, each tag of the one or more tags characterizing an interest of a user detected based on an interaction between a user device and a web server or an application server;

determining that one or more characteristics of each user profile of the set of user profiles satisfies an offloading condition that is a condition to transfer the set of user profiles from a cloud service to a graphical processing unit (GPU);

in response to determining that the offloading condition is satisfied;

transforming, using the GPU, the set of user profiles into a plurality of subsets of user profiles; and

storing, for each subset of the plurality of subsets of user profiles, the tags for the subset of user profiles in in a manner that accounts for the tags and a number of threads in a warp of the GPU, wherein a number of threads in the warp of the GPU equals a number of user profiles in the subset of user profiles;

determining, using the warps of the GPU, one or more user profiles of the set of user profiles that satisfy a logic of one or more logic trees wherein the one or more user profiles are determined based on one or more tags of one or more specialized data structures associated with the subsets of the set of users and one or more operand values of the one or more logic trees; and

outputting the one or more user profiles.

8. The system of Claim 7 , further comprising:

creating at least one array, wherein the tags for the subset of user profiles are stored in or in accordance with the at least one array, wherein the at least one array stores tags for the subset of user profiles in a sorted order, wherein tags for a user profile are arranged sequentially in the at least one array according to identifiers.

9. The system of Claim 7 , further comprising:

creating at least one array, wherein the tags for the subset of user profiles are stored in or in accordance with the at least one array, wherein the at least one array stores tags for the subset of user profiles in an interleaved manner, wherein tags for a user profile are interleaved with tags for another user profile in the at least one array according to identifiers.

10. The system of claim 7 , The computer-implemented method of claim 1 , wherein, for each subset of the plurality of subsets of user profiles, each thread of the number of threads in the warp is configured to execute instructions running on a core of the GPU.

11. The system of claim 7 , wherein determining, using the warps of the GPU, one or more user profiles of the set of user profiles that satisfy logic of one or more logic trees comprises reading a specialized data structure of the specialized data structures with a warp of the GPU.

12. The system of claim 7 , further comprising:

accessing at least one array, wherein the tags for the subset of user profiles are stored in or in accordance with the at least one array,

wherein determining, using the warps of the GPU, one or more user profiles of the set of user profiles that satisfy logic of one or more logic trees comprises:

retrieving tags from an array of the at least one array;

identifying one or more operand values of a logic tree of the one or more logic trees that references the retrieved tags; and

setting the logic tree of the one or more logic trees as true.

13. A computer-program product tangibly embodied in a non-transitory machine-readable storage medium, including instructions configured to cause one or more data processors to perform a set of actions including:

accessing a set of user profiles, each user profile of the set of user profiles including one or more tags, each tag of the one or more tags characterizing an interest of a user detected based on an interaction between a user device and a web server or an application server;

determining that one or more characteristics of each user profile of the set of user profiles satisfies an offloading condition that is a condition to transfer the set of user profiles from a cloud service to a graphical processing unit (GPU);

in response to determining that the offloading condition is satisfied:

transforming, using the GPU, the set of user profiles into a plurality of subsets of user profiles; and

storing, for each subset of the plurality of subsets of user profiles, the tags for the subset of user profiles in in a manner that accounts for the tags and a number of threads in a warp of the GPU, wherein a number of threads in the warp of the GPU equals a number of user profiles in the subset of user profiles;

determining, using the warps of the GPU, one or more user profiles of the set of user profiles that satisfy a logic of one or more logic trees, wherein the one or more user profiles are determined based on one or more tags of one or more specialized data structures associated with the subsets of the set of users and one or more operand values of the one or more logic trees; and

outputting the one or more user profiles.

14. The computer-program product of claim 13 , further comprising:

creating at least one array, wherein the tags for the subset of user profiles are stored in or in accordance with the at least one array,

wherein the at least one array stores tags for the subset of user profiles in a sorted order, wherein tags for a user profile are arranged sequentially in the at least one array according to identifiers.

15. The computer-program product of claim 13 , further comprising:

creating at least one array, wherein the tags for the subset of user profiles are stored in or in accordance with the at least one array, wherein the at least one array stores tags for the subset of user profiles in an interleaved manner, wherein tags for a user profile are interleaved with tags for another user profile in the at least one array according to interest identifiers.

16. The computer-program product of claim 13 , The computer-implemented method of claim 1 , wherein, for each subset of the plurality of subsets of user profiles, each thread of the number of threads in the warp is configured to execute instructions running on a core of the GPU.

17. The computer-program product of claim 13 , wherein determining, using the warps of the GPU, one or more user profiles of the set of user profiles that satisfy logic of one or more logic trees comprises reading a specialized data structure of the specialized data structures with a warp of the GPU.

Assignments (1)

Continuity (4)

References Cited (37)