# US 2026/0195131 A1 · Preventing performance of conditional instructions

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

- Number: US 2026/0195131 A1
- Document ID: US20260195131A1
- Status: Published application
- Kind: A1
- Application number: 19/015,419
- Filed: 2025-01-09
- Published: 2026-07-09
- Earliest claimed: 2025-01-08
- Assignees: NVIDIA Corporation
- Applicants: NVIDIA Corporation
- Inventors: Vikram Dhar; Sean Jeffrey Treichler; Jack H. Choquette; Subhasmita Chakraborty; Yingda Wang; Kostantinos Kyriakopoulos; Bharath Vasudevan
- Web page: https://patentgrep.com/patent/US20260195131A1

## CPC

- G06F 9/3851: {from multiple instruction streams, e.g. multistreaming}
- G06F 9/38885: {Divergence aspects}
- G06F 9/30058: {Conditional branch instructions}
- G06F 9/30072: {to perform conditional operations, e.g. using predicates or guards}
- G06F 9/30076: {to perform miscellaneous control operations, e.g. NOP}
- G06F 9/30087: {Synchronisation or serialisation instructions}
- G06F 9/3009: {Thread control instructions}
- G06F 9/30094: {Condition code generation, e.g. Carry, Zero flag}
- G06F 9/38: Concurrent instruction execution, e.g. pipeline or look ahead
- G06F 9/3838: {Dependency mechanisms, e.g. register scoreboarding}
- G06F 9/3854: {Instruction completion, e.g. retiring, committing or graduating}
- G06F 9/3887: {controlled by a single instruction for multiple data lanes [SIMD]}
- G06F 9/38873: {Iterative single instructions for multiple data lanes [SIMD]}
- G06F 9/3888: {controlled by a single instruction for multiple threads [SIMT] in parallel}
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

Apparatuses, systems, and techniques to control the flow of instructions in parallel processing. In at least one embodiment, one or more instructions are conditional instructions a processor must wait for in order to move forward with the next set of instructions.

## Claims

### Claim 1

A processor, comprising:
one or more circuits to perform one or more first instructions to prevent one or more conditional instructions from being performed.

### Claim 2 (depends on 1)

The processor of claim 1, wherein the one or more circuits are to further perform one or more first instructions to track the completion of one or more branched instructions.

### Claim 3 (depends on 1)

The processor of claim 1, wherein the one or more conditional instructions are to be included in one or more second instructions performed in parallel.

### Claim 4 (depends on 1)

The processor of claim 1, wherein the one or more circuits are to further perform the one or more first instructions to identify if the one or more conditional instructions complete at a uniform register.

### Claim 5 (depends on 1)

The processor of claim 1, wherein the one or more first instructions comprise an instruction to cause the one or more circuits to track, by a warp-level register, an active mask value for a warp.

### Claim 6 (depends on 1)

The processor of claim 1, wherein the one or more first instructions comprise an instruction to prevent all threads of a warp from exiting a reliable region before the one or more conditional instructions are to be prevented from being performed.

### Claim 7 (depends on 1)

The processor of claim 1, wherein the one or more first instructions further comprise an instruction to release all threads of a warp comprising conditional instructions from a reliable region based, at least in part, on all threads being converged or cancelled.

### Claim 8

A system, comprising:
one or more processors to perform one or more first instructions to prevent one or more conditional instructions from being performed.

### Claim 9 (depends on 8)

The system of claim 8, wherein the one or more processors are to further perform the one or more first instructions to track completion of one or more branched instructions.

### Claim 10 (depends on 8)

The system of claim 8, wherein the one or more conditional instructions are to be included in one or more second instructions to be performed in parallel.

### Claim 11 (depends on 8)

The system of claim 8, wherein the one or more first instructions are to identify if the one or more conditional instructions complete at a uniform register.

### Claim 12 (depends on 8)

The system of claim 8, wherein the one or more first instructions comprise an instruction to track, by a warp-level register, an active mask value for a warp.

### Claim 13 (depends on 8)

The system of claim 8, wherein the one or more first instructions comprise an instruction to prevent all threads of a warp from exiting a reliable region before the one or more conditional constructions are to be prevented from being performed.

### Claim 14 (depends on 8)

The system of claim 8, wherein the one or more first instructions comprise an instruction to release all threads of a warp performing the one or more instructions, including the conditional instructions, from a reliable region, based on all threads being converged or cancelled.

### Claim 15

A method, comprising:
performing one or more first instructions to prevent one or more conditional instructions from being performed.

### Claim 16 (depends on 15)

The method of claim 15, further comprising performing one or more first instructions to track completion of one or more branched instructions.

### Claim 17 (depends on 15)

The method of claim 15, wherein the conditional instructions are to be included in one or more second instructions to be performed in parallel.

### Claim 18 (depends on 15)

The method of claim 15, further comprising performing one or more first instructions to identify whether the one or more conditional instructions complete at a uniform register.

### Claim 19 (depends on 15)

The method of claim 15, further comprising performing one or more first instructions to cause a warp-level register to be used to track an active mask value for a warp.

### Claim 20 (depends on 15)

The method of claim 15, further comprising performing the one or more first instructions to prevent all threads of a warp from exiting a reliable region before the one or more conditional constructions are prevented from being performed.
