[erlang-questions] ot: Swedish article featuring Erlang Solutions CTO

Isaac Gouy <>
Tue Jun 14 20:39:34 CEST 2011


On Mon, Jun 13, 2011 at 10:41, Robert Virding wrote:

> What is perhaps more interesting than the article itself are 
> all the comments. "I know nothing about language X but I know
> that it is 10-15 times slower than language Y".

I did see a comment like that, but at first glance I didn't read it correctly.

I think it actually said something more like - Haskell is X times slower and Erlang is Y times slower than C.



On Mon, Jun 13, 2011 at 2:46, Jesper Louis Andersen

> He is considering the "Computer Language benchmarks game",
> specifically the "fasta" benchmark on a Quad Core

Fasta? What a strange choice!

Oh! Could it be that none of the tests before Fasta show a C program as fastest :-)


> You can make the opposite conclusion by considering 
> the thread-ring benchmark

Another sequential program ;-)

(At least spectral-norm keeps all the cores working.)


-snip-
> The danger of concluding something from a synthetic benchmark 
> is that you are going to conclude the wrong thing.

I doubt the person who made that comment on the Computer Sweden website actually was concluding something from looking at the benchmarks game web pages.

It seems more likely that they went through the web pages looking for some measurement they could use to demonstrate what they already believed.

The salutary lesson is that we all seem to have a weakness for ignoring contradictions to our beliefs and seeking confirmation for our beliefs.





More information about the erlang-questions mailing list