Did you try researching this at all by yourself? There are a plethora of blog posts explaining exactly what you are asking. Or you could go read the original paper. This would be easy enough to write a test script as well to empirically discover the answer.
With python and going from memory, bcrypt set for 12 rounds takes about .3 seconds on my laptop to compute a hash. I can compute hundreds of thousands of SHA512 hashes in that same amount of time.
Edit: And theoretically, you could make a SHA hash function that ran quite slowly, but why not just use bcrypt at that point.
I wasn't trying to make the bitcoin community to do my searching, no far from it. There is alot of files on google for encryption methods but most are pretty outdated as far as performance goes and I like to hear reliable information, so if someone mentions a factual article here that I've seen it online, That tells me that they probably did their research like I did, Validating that the article I read must be somewhat true. Scince another person with different type of perspective saw that article valid. In other words I don't go by hear say, or just "read an article" online and believe it, I must cross reference and figure out others opinions before I can see it as the truth.
and to g-max well thanks that's a very use full and helpful article about entropy and securing short passwords