Junyi <9jhzguy(a)gmail.com> writes:
For the duration of the kernel call, the cpu
busy-waits by default. I
changed the make_context() portion to include the SCHED_BLOCKING_SYNC flag,
but the kernel call just hangs. How should I trigger the release? Thanks!
Can you please supply a) a reproducing snippet of code (ideally short)
and b) some information about your system (GPU, OS, CUDA, Python