# US 2026/0259715 A1 · Dynamic compiler techniques

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2026/0259715 A1
- Document ID: US20260259715A1
- Status: Published application
- Kind: A1
- Application number: 19/654,284
- Filed: 2026-04-21
- Published: 2026-09-03
- Earliest claimed: 2014-01-21
- Assignees: NVIDIA Corporation
- Applicants: NVIDIA Corporation
- Inventors: Vinod Grover; Thibaut Lutz
- Web page: https://patentgrep.com/patent/US20260259715A1

## CPC

- G06F 8/4443: {Inlining}
- G06F 8/47: {Retargetable compilers}
- G06F 9/4552: {Involving translation to a different instruction set architecture, e.g. just-in-time translation in a JVM}
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

Compiler techniques for inline parallelism and re-targetable parallel runtime execution of logic iterators enable selection thereof from the source code or dynamically during object-code execution.

## Claims

### Claim 1

One or more processors, comprising:
circuitry to:
during execution of machine code by a central processing unit (CPU), compile an intermediate representation embedded in the machine code into graphics processing unit (GPU) machine code; and
cause the GPU machine code to be executed by a GPU.

### Claim 2 (depends on 1)

The one or more processors of claim 1, wherein the embedded intermediate representation comprises an intermediate representation of a logic iterator.

### Claim 3 (depends on 1)

The one or more processors of claim 1, wherein the embedded intermediate representation includes a specifier indicating a processor on which code generated from the intermediate representation is to be executed.

### Claim 4 (depends on 1)

The one or more processors of claim 1, wherein the embedded intermediate representation includes a specifier indicating whether code generated from the intermediate representation is to be executed sequentially, executed in parallel, or selected for execution at runtime.

### Claim 5 (depends on 1)

The one or more processors of claim 1, wherein the circuitry is further to use a runtime library comprising one or more routines callable during execution of compiled code to compile the intermediate representation into the GPU machine code.

### Claim 6 (depends on 1)

The one or more processors of claim 1, wherein the GPU machine code is compiled for compute unified device architecture (CUDA) execution.

### Claim 7 (depends on 1)

The one or more processors of claim 1, wherein the embedded intermediate representation comprises an intermediate representation of a lambda expression, and wherein compiling the intermediate representation into the GPU machine code comprises compiling the lambda expression and one or more dependencies of the lambda expression into the GPU machine code.

### Claim 8

One or more processors, comprising:
circuitry to:
during execution of machine code by a central processing unit (CPU), compile an intermediate representation embedded in the machine code into graphics processing unit (GPU) machine code, wherein the intermediate representation includes a specifier indicating that code generated from the intermediate representation is to be executed by the GPU; and
cause the GPU machine code to be executed by a GPU.

### Claim 9 (depends on 8)

The one or more processors of claim 8, wherein the embedded intermediate representation comprises an intermediate representation of a logic iterator.

### Claim 10 (depends on 8)

The one or more processors of claim 8, wherein the specifier further indicates that code generated from the intermediate representation is to be compiled using just-in-time (JIT) compilation for execution by the GPU.

### Claim 11 (depends on 8)

The one or more processors of claim 8, wherein the specifier further indicates that execution of code generated from the intermediate representation is selectable during execution of the machine code by the CPU.

### Claim 12 (depends on 8)

The one or more processors of claim 8, wherein the circuitry is further to use a runtime library comprising one or more routines callable during execution of compiled code to compile the intermediate representation into the GPU machine code.

### Claim 13 (depends on 8)

The one or more processors of claim 8, wherein the embedded intermediate representation comprises an intermediate representation of a lambda expression, and wherein compiling the intermediate representation into the GPU machine code comprises compiling the lambda expression and one or more dependencies of the lambda expression into the GPU machine code.

### Claim 14 (depends on 8)

The one or more processors of claim 8, wherein the GPU machine code is compiled for compute unified device architecture (CUDA) execution.

### Claim 15

A method, comprising:
based, at least in part, on execution of machine code by a central processing unit (CPU), compiling an intermediate representation embedded in the machine code into graphics processing unit (GPU) machine code; and
causing the GPU machine code to be executed by a GPU.

### Claim 16 (depends on 15)

The method of claim 15, wherein the embedded intermediate representation comprises an intermediate representation of a logic iterator.

### Claim 17 (depends on 15)

The method of claim 15, wherein the embedded intermediate representation includes a specifier indicating a processor on which code generated from the intermediate representation is to be executed.

### Claim 18 (depends on 15)

The method of claim 15, wherein the embedded intermediate representation includes a specifier indicating whether code generated from the intermediate representation is to be executed sequentially, executed in parallel, or selected for execution at runtime.

### Claim 19 (depends on 15)

The method of claim 15, further comprising using a runtime library comprising one or more routines callable during execution of compiled code to compile the intermediate representation into the GPU machine code.

### Claim 20 (depends on 15)

The method of claim 15, wherein the embedded intermediate representation comprises an intermediate representation of a lambda expression, and wherein compiling the intermediate representation into the GPU machine code comprises compiling the lambda expression and one or more dependencies of the lambda expression into the GPU machine code.
