Nvidia patents cloud system that predicts which graphics code you need next
Nvidia describes a cloud-based system that helps devices load graphics instructions faster. When a device asks for one piece of code, a trained model predicts which other pieces will likely be needed soon and sends them along.
In the filing’s own words
Apparatuses, systems, and techniques for caching compiled shader programs in a cloud computing environment. A request for a compiled shader program associated with an initial rendering operation is received from a client device. Based on the request, at least one of a shader key representing a hardware or software state of the client device, or a shader value derived from source code of the compiled shader program is obtained. The shader key or shader value is provided as input to a trained model. From the trained model, an indication of additional compiled shader programs and corresponding confidence levels that each will be requested within a time window is obtained. The compiled shader program and additional compiled shader programs having confidence levels satisfying confidence criteria are retrieved from a shader cache and provided to the client device.
What exactly they claim (claim 1)
1. A system comprising: a memory; and a set of one or more processing devices coupled to the memory, the set of one or more processing devices configured to perform operations comprising: receiving, from a client device executing an application, a request for a compiled shader program associated with an initial rendering operation; obtaining, based on the request, at least one of: (i) a shader key representing one or more of a hardware state of the client device or a software state of the client device, or (ii) a shader value derived from a source code of the compiled shader program; providing the at least one of the shader key or the shader value as an input to a trained model; obtaining, from the trained model, an indication of one or more additional compiled shader programs and, for each additional compiled shader program of the one or more additional compiled shader program, a level of confidence that the respective additional compiled shader program will be requested within a time window associated with one or more subsequent rendering operations; retrieving, from a shader cache storing a plurality of compiled shader programs associated with the application, the compiled shader program, and at least one additional compiled shader program of the one or more additional compiled shader programs, the at least one additional compiled shader program having a level of confidence that satisfies one or more confidence criteria; and providing the compiled shader program and the at least one additional compiled shader program to the client device in accordance with the request.
Facts
- Company
- Nvidia (NVIDIA Corporation)
- Official title
- CACHING OF COMPILED SHADER PROGRAMS IN A CLOUD COMPUTING ENVIRONMENT
- Publication number
- US20260299906A1
- Application number
- 19701279
- Filed
- Jun 8, 2026
- Published
- Oct 1, 2026
- Inventors
- 5
- Claims
- 20
Classified under
G06F 8/44 Physics & computingG06F 8/60 Physics & computingG06F 8/71 Physics & computingG06F 16/2255 Physics & computingG06T 1/20 Physics & computingG06T 1/60 Physics & computingMore from Nvidia in AI & software
- Nvidia patents AI coach that analyzes your gameplay to suggest strategies Oct 13, 2026
- Nvidia patents AI system that writes and voices character conversations Sep 10, 2026
- Nvidia patents system that uses AI to find bugs in game videos Sep 17, 2026
- Nvidia patents AI method to make digital characters move realistically Sep 8, 2026
- Nvidia patents AI system that builds image generators from text prompts Sep 15, 2026
Summaries are simplified from the filing’s own abstract and first claim. A patent filing is not a product announcement.
Patent data and drawings: U.S. Patent and Trademark Office (USPTO). Nebula is not affiliated with or endorsed by the USPTO.