Table 6.

Summary of scaling techniques

Scaling approach	Pros	Cons
Out-of-Core	• Concurrent computations	• Load balancing
Computation	• In-memory calculations	• Synchronization overhead
		• Limited by number of processor cores
Parallel	• Concurrent computations	• Inefficient for longer iteration steps
Computing	• Unlimited nodes	• Out-of-memory computation
		• Load balancing
		• Synchronization overhead
		• Communication overhead
Graphical	• Huge number of concurrent threads	• High thread divergence
Processing	• In-memory (GPU memory) calculations	• Data structures cannot be used
Units (GPUs)	• Low data transfer overhead compared to distributed computing	• Synchronization overhead