Nvidia patents code method to cut data sent between CPU and GPU
Nvidia describes a system that generates code for computers to talk between the main processor (CPU) and graphics chip (GPU). It works by removing certain fixed data from the message sent between them, as the receiving chip already has that information stored locally.
In the filing’s own words
In various examples, a compiled code having a host program and a device program is generated with a definition for an application binary interface (ABI). The ABI facilitates communication of data/instructions between a CPU and a GPU while sending only data dynamic data over the ABI. Systems and methods identify a structure of data to be communicated over the ABI. A function for a second coding language is selected to generate the structure without static variables. The function for the second coding language can be used to generate host and device instructions for execution on a host device. When data is communicated over the ABI without the static variables, it is combined with the missing data of the static variable within the destination. Explicit elimination of certain data transmitted over the API can have significant impact over the number of computations performed by the compiler.
What exactly they claim (claim 1)
1. One or more processors comprising processing circuitry to: identify, in a first code, a first structure comprising one or more parameters defining an operation to be performed, wherein the one or more parameters comprise a static variable; select a function for a second code to generate a second structure comprising the one or more parameters of the first structure without the static variable; generate, using the function, a first program to transfer the second structure to a second program; and generate the second program using the static variable, wherein the second program is to receive the second structure and perform the operation without receiving the static variable at a time the second program is being executed.
Facts
- Company
- Nvidia (NVIDIA Corporation)
- Official title
- SYSTEMS AND METHODS FOR CODE GENERATION WITH AN APPLICATION BINARY INTERFACE
- Publication number
- US20260277552A1
- Application number
- 19081551
- Filed
- Mar 17, 2025
- Published
- Sep 17, 2026
- Inventors
- 1
- Claims
- 20
Classified under
G06F 8/30 Physics & computingG06F 8/41 Physics & computingG06F 11/3604 Physics & computingMore from Nvidia in Chips & hardware
- Nvidia patents method to skip zero data in neural network math Oct 13, 2026
- Nvidia patents hardware to speed up motion blur in ray tracing Oct 13, 2026
- Nvidia patents method to sync data across split processor chips Sep 17, 2026
- Nvidia patents chip math trick to speed up AI calculations Oct 8, 2026
- Nvidia patents data center cooling that uses cryogenic fluids Oct 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.