share data between different kernels

Dear experts:

I am new to cuda. I have a question regarding sharing data between different kernels. If I launch several kernels (like with streams), is that possible to share data between these kernels? Is the inter kernel communication supported by cuda?



AFAIK, you can share data using global memory but that’s all ‘communication’ you can get. But since this is the driver forum, you’d better ask again in the cuda forum: