Forum Discussion
Altera_Forum
Honored Contributor
7 years agoTo be honest OpenCL is not suitable for low-latency design since the compiler is actually designed to sacrifice latency and maximize throughput by using a very deep pipeline that efficiently absorbs stalls from external memory.
I haven't used the profiler in a very long time but from what I remember, the profiler should also show host to device transfers alongside with the kernel run time; this could be the reason for the gaps. Memory transfers between the FPGA and its external memory are included in the kernel run time. ...