I want to view the CPU threads’ information while running a kernel on GPU (after calling clEnqueueNDRangeKernel). I try using gdb by setting breakpoint right after the call and run info threads. However, it seems like gdb is waiting until the gpu finishes running that kernel before displaying the info.
Basically, I’m curious if there is a background thread that OpenCL creates to do the “busy-wait” while the kernel is executed. I’ve heard that OpencCL does that on some gpu devices but not others, so I wonder what it is the case on my device. Thanks.