On NVidia, there is the following limit: You can create maximum 32 processes where each one creates an OpenCL context.
What is the limit for AMD GPUs?
Tomer Gal, CTO at OpTeamizer
There is no such hard limit at OCL runtime level. The actual number may vary depending on OS, driver, system resources etc. It’s always recommended to check the API return status instead.
Yes, I am aware of the fact that this limit does not originate from the OCL runtime.
That's why I mentioned that I've hit this limit on NVidia GPU and am asking what is that limit on AMD GPU.
OS: Windows 10, 64 bit