> I think the test was "rigged". Arrays in Delphi > work very different from arrays in other languages. > This test demonstrated that for this case Delphi's > arrays work really well. That's all.
AIUI from the follow up comments in the original post, the test conditions were set by the C# guys so it's hard to see how it could be claimed that things were deliberately rigged in Delphi's favour. > I think it is rather because the arrays work > vastly different, so we're comparing vastly > different things here. Let's make a test that > employs hash structures and let's see again. Well, not comparing *different* things so much as comparing something that doesn't do well in C#/.NET. I agree, it *would* be interesting to see how a problem involving hash structures would perform, as well as any number of other scenarios. But my concern is with the performance of code most likely to occur "naturally in the wild". I'd suggest that (in general) problems that are most intuitively solved with hash structures typically perhaps don't occur as frequently as problems involving arrays? In the case of the original benchmarking exercise, my point is simply this: The "simple but efficient" Delphi code that was posted seemed perfectly naturally written... the sort of code that any half competent developer would write without stopping to give too much thought to efficiency of the algorithm, and yet it performed as well as (in fact slightly better than, iirc) the most highly tuned C# effort (to date). If it takes an algorithm and techniques guru to extract even acceptable performance from a language than ime you won't get that anything like that performance in the majority of the application code that is written because most developers are not thinking about performance at all and even if they are, will not necessarily be as equipped to achieve in this respect as the sort of people submitting "solutions" to such benchmarking exercises. And in any event, the "highly tuned" Delphi code still left the highly tuned C# code waaaay behind. :) _______________________________________________ NZ Borland Developers Group - Delphi mailing list Post: delphi@delphi.org.nz Admin: http://delphi.org.nz/mailman/listinfo/delphi Unsubscribe: send an email to delphi-requ...@delphi.org.nz with Subject: unsubscribe
