[erlang-questions] Execution delays

[Ashley Holman]

Argh, silly me!  I think I just figured out what's happening.  It's the
final multiplication in fac_multi_collect/2 that's having to run after the
third process returns.  It is multiplying some pretty massive numbers so
that would explain it.

This raises a more general question though - are there some profiling tools
out there so that I can more easily find out where the time is going and
when things are executing?

Thanks very much.

On Thu, Aug 9, 2012 at 11:02 PM, Ashley Holman <dscvlt@REDACTED> wrote:

> Hi,
>
> (code attached)
>
> I've just began learning Erlang over the last few days and so far a really
> enjoyed programming in it.  As a learning exercise, I decided to implement
> the classic "factorial" function: fac(N) = N*fac(N-1); F(0) = 1.  For
> calculating large factorials, this takes a long time (fac(200000) takes 42
> secs on my laptop).
>
> So, I made some optimisations (using condense_terms/1 and shuffle/1 - but
> you can basically ignore those).  However, it still was only using 100% of
> 1 cpu core, and I have 2 cores, so this was a good opportunity to learn an
> important feature of Erlang, spawning processes!  I partitioned the list
> into 3 sublists and spawned a process to multiply each sublist.  eg.
> [pseudo code] Proc1([1,2,3]) * Proc2([4, 5, 6]) * Proc3(7, 8, 9) should
> give me the result of fac(9) using 3 processes.
>
> This appears to be working but I'm completely confused about certain
> delays in execution I'm seeing - I'm not sure where it is spending some of
> its time in the code.  I put in some debugging, and it appears that the
> child processes (fac_multi_actor) are finishing their calculations and
> returning the results, but the receiver (fac_multi_collect) doesn't return
> for several seconds later.
>
> The whole point of this was to try to optimise it, so I would like to know
> where these few seconds are going (~30% of execution time).
>
> fac:fac(200000) is taking ~9.5s using the three processes, but based on
> the debugging it looks like it should be completing in 6.  Something else
> hangs around using 100% of one core for several seconds later.  I've
> attached the code in case anyone is able to take a look at this behaviour.
>
> PS. I realise it will take some effort for someone to actually bother
> reading the code and trying to reproduce my problem, so no problem if I
> don't get any answers.  In case anyone feels like helping though it would
> be much appreciated because I'm a bit confused at the moment!
>
> Thanks
> Ash
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://erlang.org/pipermail/erlang-questions/attachments/20120809/a05cbe1d/attachment.htm>