libsmctrl.git/libsmctrl_test_next_mask_override.c, branch ecrts25-ae Library to enable intra-context SM/TPC partitioning on NVIDIA GPUs Add test that higher-granularity masks override lower-granularity ones 2024-02-14T20:36:25+00:00 Joshua Bakita jbakita@cs.unc.edu 2024-02-14T20:36:25+00:00 b5281f5fc01fc925898c9323edab41b817df8661 Stream-level masks should always override globally-set masks. Next-kernel masks should always override both stream-level masks and globally-set masks. Tests reveal an issue with the next-kernel mask not overriding the stream mask on CUDA 11.0+. CUDA appears to apply the per-stream mask to the QMD/TMD after `launchCallback()` is triggered, making it impossible to override as currently implemented. 
Stream-level masks should always override globally-set masks.
Next-kernel masks should always override both stream-level masks
and globally-set masks.

Tests reveal an issue with the next-kernel mask not overriding the
stream mask on CUDA 11.0+. CUDA appears to apply the per-stream
mask to the QMD/TMD after `launchCallback()` is triggered, making
it impossible to override as currently implemented.