RE: Is a hash faster?

Any time I see "Is it faster to do X or Y", I reach for Benchmark.

Make two subs, one with this routine and the other as you described it, then run 'em both through Benchmark and see for yourself which is faster. Without numbers, it's all speculation.

Comment on RE: Is a hash faster?

Replies are listed 'Best First'.
RE: RE: Is a hash faster? by Adam (Vicar) on Oct 08, 2000 at 23:29 UTC
Without numbers, it's all speculation. That isn't entirely true. Most modern computer and math theory is based on 'formalism' which involves the use of symbols and variables in place of numbers. Why? Because benchmark only tells you the runtime for a given machine with a given set of external variables and a given input. If you trully want to know if something will be faster you must first look at it analytically (sp?) and determine which should be faster. Then explore worst case/best case scenarios. And then check your work with a couple benchmarks... but to some extent, you are right... it doesn't hurt to benchmark. I'm just saying its not the complete solution.	[reply]

Replies are listed 'Best First'.

RE: RE: Is a hash faster?
by Adam (Vicar) on Oct 08, 2000 at 23:29 UTC

Without numbers, it's all speculation.

That isn't entirely true. Most modern computer and math theory is based on 'formalism' which involves the use of symbols and variables in place of numbers. Why? Because benchmark only tells you the runtime for a given machine with a given set of external variables and a given input. If you trully want to know if something will be faster you must first look at it analytically (sp?) and determine which should be faster. Then explore worst case/best case scenarios.

And then check your work with a couple benchmarks... but to some extent, you are right... it doesn't hurt to benchmark. I'm just saying its not the complete solution.

[reply]