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

Groups > comp.programming.threads > #2261 > unrolled thread

More about my algorithm of my concurrent FIFO queue

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

Contents

  More about my algorithm of my concurrent FIFO queue aminer <aminer@toto.net> - 2014-04-25 17:28 -0700
    Re: More about my algorithm of my concurrent FIFO queue aminer <aminer@toto.net> - 2014-04-25 17:42 -0700

#2261 — More about my algorithm of my concurrent FIFO queue

Hello,

Hello Chriss Thomasson, you have suggested to be to use Relay Detector,
but you have to know that a good reasonning also is efficient, it's
like formal proof, what you have to do is concentrate on the very 
important and difficult parts, but the easy parts are easy to proof...

What is the difficult parts to proof in my algorithm ?

Lokk inside the source code at:

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

The first difficult part is the lockfree mechanism...

So notice that my lockfree mechanism to not go over the "head", i have
just explain it to you that it's working perfectly, and since
i know very well what is a lockfree algorithms , you can be confident
with my lockfree mechanism that it is working perfectly, the other 
difficult part is when i have set fcount1^[newtemp1 and fMask].flag to 0 
inside the lockfree transaction before incrementing "tail" inside the 
pop() side, we can proof this one easily cause the push() side will not 
go beyong "fsize+magin" that means fsize+1000, the other part is the 
push() side and it's easy to proof and i will not proof it cause i know 
that it is working perfectly, so i think that you can be confident with 
my algorithm, i think that it's correct.


Thank you,
Amine Moulay Ramdane.

[toc] | [next] | [standalone]

#2262

On 4/25/2014 5:28 PM, aminer wrote:
>
> Hello,
>
> Hello Chriss Thomasson, you have suggested to be to use Relay Detector,
> but you have to know that a good reasonning also is efficient, it's
> like formal proof, what you have to do is concentrate on the very
> important and difficult parts, but the easy parts are easy to proof...
>
> What is the difficult parts to proof in my algorithm ?
>
> Lokk inside the source code at:
>
> http://pages.videotron.com/aminer/
>
> The first difficult part is the lockfree mechanism...
>
> So notice that my lockfree mechanism to not go over the "head", i have
> just explain it to you that it's working perfectly, and since
> i know very well what is a lockfree algorithms , you can be confident
> with my lockfree mechanism that it is working perfectly, the other
> difficult part is when i have set fcount1^[newtemp1 and fMask].flag to 0
> inside the lockfree transaction before incrementing "tail" inside the
> pop() side, we can proof this one easily cause the push() side will not
> go beyong "fsize+magin" that means fsize+1000, the other part is the

I mean "fsize+margin"...



> push() side and it's easy to proof and i will not proof it cause i know
> that it is working perfectly, so i think that you can be confident with
> my algorithm, i think that it's correct.
>
>
> Thank you,
> Amine Moulay Ramdane.
>
>
>
>

[toc] | [prev] | [standalone]

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

csiph-web