Spinning on an atomic load with acquire consistency vs. relaxed consistency

Question

Consider the code below:

// Class member initialization:
std::atomic ready = false;

...

// Core A:
while (!ready.load(std::memory_order_acquire)