graphdot.cuda.resizable_array module

class graphdot.cuda.resizable_array.ResizableArray(dtype, count=0, allocator='managed')[source]

Bases: object

Python version of std::vector. To be used together with CUDA managed memory to reduce kernel launch delay.

append(value)[source]
capacity
clear()[source]
data
reserve(count)[source]
resize(count)[source]
graphdot.cuda.resizable_array.managed_allocator(count, dtype)[source]
graphdot.cuda.resizable_array.numpy_allocator(count, dtype)[source]