# Share CUDA kernel across contexts

**URL:** <https://forums.developer.nvidia.com/t/share-cuda-kernel-across-contexts/339210>\
**Category:** CUDA Programming and Performance\
**Tags:** kernel, cuda\
**Created:** [July 16, 2025, 6:33am UTC](https://forums.developer.nvidia.com/t/share-cuda-kernel-across-contexts/339210 "2025-07-16T06:33:36Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![wuxinyue.wxy](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@wuxinyue.wxy](https://forums.developer.nvidia.com/u/wuxinyue.wxy)\
**Post date:** [July 16, 2025, 6:33am UTC](https://forums.developer.nvidia.com/t/share-cuda-kernel-across-contexts/339210/1 "2025-07-16T06:33:36Z")

</div>

I see device memory can be shared across CUDA contexts by leveraging CUDA IPC. Can CUDA kernel also be shared across CUDA contexts, just like what a CPU-side shared library do?

In my case, there are multiple (at least two) processes using GPU, and they launches exactly same kernels. I want to avoid multiple kernel copies in GPU side and make them share same kernel code.