[erlang-questions] benchmarks game harsh criticism

Jack Orenstein <>
Wed Nov 28 20:58:25 CET 2007


Going off on a hopefully interesting tangent ...

On Wed, November 28, 2007 1:45 pm, Isaac Gouy wrote:
>
> I do remember a comment from Ulf Wiger to the effect that the Erlang VM
> seemed to reach peak performance after 24 hours of constant load, which
> he thought very appropriate given the intended uses.

Can someone confirm this and explain the cause?

I have worked in Java for many years and learning Erlang now. There
is a similar phenomenon in Java as the just-in-time compiler compiles
frequently used code paths. But for the applications I've worked on,
this tends to get done within the first few minutes or hours.

Is it the same sort of thing going on in Erlang? If not, then what
is the cause of the speedup (and why 24 hours)?

Jack Orenstein





More information about the erlang-questions mailing list