# ros2\_cuda\_ipc: Zero-copy GPU data sharing between ROS 2 processes

**URL:** <https://discourse.openrobotics.org/t/ros2-cuda-ipc-zero-copy-gpu-data-sharing-between-ros-2-processes/57157>\
**Category:** Projects\
**Tags:** release, ros2, gpu\
**Created:** [August 4, 2026, 3:47pm UTC](https://discourse.openrobotics.org/t/ros2-cuda-ipc-zero-copy-gpu-data-sharing-between-ros-2-processes/57157 "2026-08-04T15:47:34Z")\
**Posts on this page:** 1\
**Showing post:** 9

<div class="post-metadata">

**Author:** ![ZhenshengLee](https://sea2.discourse-cdn.com/flex022/user_avatar/discourse.openrobotics.org/zhenshenglee/32/10103_2.png) [@ZhenshengLee](https://discourse.openrobotics.org/u/ZhenshengLee)\
**Post date:** [August 5, 2026, 2:30am UTC](https://discourse.openrobotics.org/t/ros2-cuda-ipc-zero-copy-gpu-data-sharing-between-ros-2-processes/57157/9 "2026-08-05T02:30:13Z")

</div>

Does this lib directly depend on CUDA IPC API?

As far as I know, CUDA Inter-Process Communication (IPC) memory-sharing APIs are not supported on the NVIDIA Jetson platform running Jetson Linux (L4T).

[CUDA IPC vs NVSHMEM for shared memory between applications - CUDA / CUDA Programming and Performance - NVIDIA Developer Forums](https://forums.developer.nvidia.com/t/cuda-ipc-vs-nvshmem-for-shared-memory-between-applications/237071)

Do you consider using different backend like nvsci/nvstream?

Anyway, I think NITROS have done this already.

---

_[View the full topic](https://discourse.openrobotics.org/t/ros2-cuda-ipc-zero-copy-gpu-data-sharing-between-ros-2-processes/57157)._