1, Initialize the registers before using them is the best practice.
Though the use case here doesn't care whether the registers are
initialized or not, some emulators complain the "read_before_write"
behavior. Initialize the registers used to silence these complaints.
2, Update s_wait stuff for gfx12.
Change-Id: I462b2b0b5017dd2876a5954169d3b6b2f1c2a75b
Signed-off-by: Lang Yu <lang.yu@amd.com>
[ROCm/ROCR-Runtime commit: fe5f12342d]