memory_estimates_dtype

nvmath.sparse.advanced.memory_estimates_dtype = dtype([('permanent_device_memory', '<u8'), ('peak_device_memory', '<u8'), ('permanent_host_memory', '<u8'), ('peak_host_memory', '<u8'), ('hybrid_min_device_memory', '<u8'), ('hybrid_max_device_memory', '<u8'), ('reserved', '<u8', (10,))])

NumPy dtype object that encapsulates the memory estimates in sparse.advanced.

Parameters:

• permanent_device_memory – permanent device memory

• peak_device_memory – peak device memory

• permanent_host_memory – permanent host memory

• peak_host_memory – peak host memory

• hybrid_min_device_memory – (if in hybrid memory mode) minimum device memory for the hybrid memory mode

• hybrid_max_device_memory – (if in hybrid memory mode) maximum host memory for the hybrid memory mode

• reserved – reserved for future use