Warp#
2026-10-01
2 min read time
Applies to Linux and Windows
This page covers the warp extended API, providing generalized warp shuffle operations for exchanging data of arbitrary size between threads within a warp.
Header |
Content |
CCCL Availability |
CUDA Toolkit Availability |
|---|---|---|---|
Warp shuffle from a specific lane |
CCCL 3.0.0 |
CUDA 13.0 |
|
Warp shuffle from original lane index - delta |
CCCL 3.0.0 |
CUDA 13.0 |
|
Warp shuffle from original lane index + delta |
CCCL 3.0.0 |
CUDA 13.0 |
|
Warp shuffle from original lane index xor mask |
CCCL 3.0.0 |
CUDA 13.0 |