In cuda-gdb, the command “set cuda break_on_launch application” sets a breakpoint at every kernel launch. But how to debug a kernel for a particular thread. For example, if 500 threads executing a kernel, then how to check values for thread 0 ?
Thanks