Arrays vs tuples / lists

Thomas Lindgren thomasl_erlang@REDACTED
Sat May 22 19:02:12 CEST 2004

--- Thomas Johnsson <thomas@REDACTED> wrote:
> > As you can see, tuple accesses are independent of
> data size,
> > as the data already resides on the process heap.
> With ets,
> > there is always copying, so as the object size
> increases, so
> > does the cost of a lookup.
> Hm... might one suspect that the real villain is the
> copying of data into
> the asking process, rather than the ets lookup in
> itself?

Yes, probably so in the standard runtime system. 

On the other hand, the shared-heap implementation
could conceivably avoid copying. I haven't checked
whether it actually does so.

Thomas (another one)

Do you Yahoo!?
Yahoo! Domains – Claim yours for only $14.70/year 

More information about the erlang-questions mailing list