27 Nov
2007
27 Nov
'07
4:02 p.m.
Don Stewart wrote:
This is an FAQ. Unless you use the same algorithm and data types in benchmarks, you're not really benchmarking anything. And expecting one of the worst possible algorithms to be good is hoping for a little too much :)
Well, if I was comparing my Haskell against some expert-optimised C library then yeah. What puzzled me is that both programs are unoptimised and appear to be using *the same* algorithm, and yet the performance is incomparable. Now that Brent has spoken, I understand the error of my ways. ;-)
When you do use similar data types and algorithms, you get similar performance.
One would hope so, yes... :-)