Forum Discussion
1 Reply
- Altera_Forum
Honored Contributor
Hi,
I'm not sure what is the best way to do it without stalling the pipeline and/or not continuously reading from global memory and thus eat up your bandwidth, but I have a suggestion for you. You could add a second kernel aside your first which simply communicates with the second via an Altera channel or an OpenCL pipe. You launch this kernel from the host when you want the change to happen, which sends a signal to your first kernel. How does that sound? Carl