Well fill a table with a few million rows, and compare the results. If they both come back with the same rough times then you've learned they're both equally good :)
Sometimes its good to ask people "Is this true of XX?", but but if you're curious it almost takes less time to expore and test than wait for somebody to give you an answer - an answer that might actually be wrong. (Some things change across DB-releases, as I've learned to my cost. If I'm ever unsure of whether one approach is faster than the other IN MY SPECIFIC SITUATION I would test. Random benchmarks are often not as useful as real-world tests.)
