Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.programming.threads > #2265

Re: A new algorithm...

Cross-posted to 2 groups.

Show all headers | View raw

On 4/25/2014 4:26 PM, aminer wrote:
>
> Hello,
>
> Notice the backoff mechanism on the pop() side of my algorithm like this,
>
> t:=backoff.delay;
> for i:=0 to 40*t do asm pause end;
>
> this backoff mechanism is mandatory cause the lockfree mechanism do
> generate more cache-coherence traffic under high contention, hence
> with this backoff mechanism my algorithm will scale better on
> more and more cores.
>
>
> I have tested this backoff mechanism and it is working perfectly.


The backoff lowers also the contention.


>
>
>
>
> Thank you,
> Amine Moulay Ramsdane.
>
>
>
>

Back to comp.programming.threads | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread

Thread

A new algorithm... aminer <aminer@toto.net> - 2014-04-25 16:01 -0700
  Re: A new algorithm... aminer <aminer@toto.net> - 2014-04-25 16:26 -0700
    Re: A new algorithm... aminer <aminer@toto.net> - 2014-04-26 09:27 -0700
  Re: A new algorithm... "Chris M. Thomasson" <no@spam.invalid> - 2014-04-25 13:54 -0700

csiph-web