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

Groups > comp.programming.threads > #2146 > unrolled thread

I have come to another subject...

Back to article view | Back to comp.programming.threads

Contents

  I have come to another subject... aminer <aminer@toto.net> - 2014-04-11 21:31 -0700
    Re: I have come to another subject... aminer <aminer@toto.net> - 2014-04-11 21:36 -0700

#2146 — I have come to another subject...

Hello,


I have come to another subject...

Take a look at my scalable and relaxed MPMC and almost strict FIFO 
priority Queue version 1.07 here:

http://pages.videotron.com/aminer/


As i have told you, this queue is very important cause it's scalable,
that means if you have four threads on a Quadcore pushing at for example 
10000000 pushes per second each, you will be able to have
a total throughput of 4000000 pushs per second, so it will scale , and 
with more cores
it will scale more and you will have more throughtput than that...


But you have to be smart please, as you have noticed my scalable
almost strict FIFO queue is composed of multiple queues that runs in 
parallel, and it has to be used cleaverly to be scalable, so imagine 
that you have a Quadcore and you have two producers threads and
4 consumers threads and you want all the 4 cores to be utilized fully by 
the consumners threads , so you have to model my scalable queue as
follows:

You have to create 2 of my scalable and almost strict FIFO queue
each with 4 queues, and the first producer thread have to distribute the 
items in a round robin manner to the two queues..., and the second 
producer thread must do the same... this is how you can model a scenario 
with a number of producers smaller than the number
of consumers threads equal to the number of cores to be scalable.

The other scenario is the following: if you have more producers threads
than the numbers of cores that you want to utilize fully , you have
to create one of my scalable queue with the same number of queues as the 
number of producers, and the consumers have to enter a semaphore with a 
counter equal to the number of cores.


Hope you have understood my idea.


You can download my scalable and relaxed MPMC and almost strict FIFO 
priority Queue version 1.07 from:

http://pages.videotron.com/aminer/


Thank you,
Amine Moulay Ramdane.

[toc] | [next] | [standalone]

#2147

On 4/11/2014 9:31 PM, aminer wrote:
>
> Hello,
>
>
> I have come to another subject...
>
> Take a look at my scalable and relaxed MPMC and almost strict FIFO
> priority Queue version 1.07 here:
>
> http://pages.videotron.com/aminer/
>
>
> As i have told you, this queue is very important cause it's scalable,
> that means if you have four threads on a Quadcore pushing at for example
> 10000000 pushes per second each, you will be able to have

sorry i mean 1000000 pushes each...

> a total throughput of 4000000 pushs per second, so it will scale , and
> with more cores
> it will scale more and you will have more throughtput than that...
>
>
> But you have to be smart please, as you have noticed my scalable
> almost strict FIFO queue is composed of multiple queues that runs in
> parallel, and it has to be used cleaverly to be scalable, so imagine
> that you have a Quadcore and you have two producers threads and
> 4 consumers threads and you want all the 4 cores to be utilized fully by
> the consumners threads , so you have to model my scalable queue as
> follows:
>
> You have to create 2 of my scalable and almost strict FIFO queue
> each with 4 queues, and the first producer thread have to distribute the
> items in a round robin manner to the two queues..., and the second
> producer thread must do the same... this is how you can model a scenario
> with a number of producers smaller than the number
> of consumers threads equal to the number of cores to be scalable.
>
> The other scenario is the following: if you have more producers threads
> than the numbers of cores that you want to utilize fully , you have
> to create one of my scalable queue with the same number of queues as the
> number of producers, and the consumers have to enter a semaphore with a
> counter equal to the number of cores.
>
>
> Hope you have understood my idea.
>
>
> You can download my scalable and relaxed MPMC and almost strict FIFO
> priority Queue version 1.07 from:
>
> http://pages.videotron.com/aminer/
>
>
> Thank you,
> Amine Moulay Ramdane.
>
>
>
>
>
>
>
>
>
>
>
>

[toc] | [prev] | [standalone]

Back to top | Article view | comp.programming.threads

csiph-web