On Tue, Mar 19, 2013 at 8:54 PM, Matt <[email protected]> wrote:
> I created a simple script to test cluster's performance sending messages to
> workers.
>
> My problem is it slows down the send() rate significantly as I add workers,
> starting at 1 worker being the fastest. I would expect the send() rate to be
> completely untied to the number of workers (at least until you hit the
> number of CPUs). My code does no more "work" in the master if there is 1
> worker or 20, so it's not making any sense to me.
>
> Tested on Mac (Core i7), EC2 and Linux (Xeon X5675).
>
> This occurs even after using taskset to set process affinity to using unique
> cores for the master and each worker, in case someone's first thought was
> hyperthreading being a problem.
>
> Any reason for this slow down? This was node 0.8.17.
>
> Code: https://gist.github.com/baudehlo/5199506

What numbers are you seeing?  large_number / number_of_workers is what
you should expect to see: a fairly constant total throughput with
per-worker throughput inversely proportional to the number of workers.

Re: hyperthreading, if that's enabled test with num_workers = num_cpus
/ 2 (and pin each worker to a separate physical CPU.  Recent kernels
do a pretty good job of managing that themselves, though.)

-- 
-- 
Job Board: http://jobs.nodejs.org/
Posting guidelines: 
https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines
You received this message because you are subscribed to the Google
Groups "nodejs" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to
[email protected]
For more options, visit this group at
http://groups.google.com/group/nodejs?hl=en?hl=en

--- 
You received this message because you are subscribed to the Google Groups 
"nodejs" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/groups/opt_out.


Reply via email to