The maximum number of threads per block, beyond which a launch of the function would fail. This number depends on both the function and the device on which the function is currently loaded.


Generated by Doxygen for NVIDIA CUDA Library  NVIDIA