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


Groups > comp.programming.threads > #2296 > unrolled thread

More about scalability...

Started byaminer <aminer@toto.net>
First post2014-04-28 17:37 -0700
Last post2014-04-28 22:04 +0000
Articles 2 — 2 participants

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


Contents

  More about scalability... aminer <aminer@toto.net> - 2014-04-28 17:37 -0700
    Re: More about scalability... Kaz Kylheku <kaz@kylheku.com> - 2014-04-28 22:04 +0000

#2296 — More about scalability...

Fromaminer <aminer@toto.net>
Date2014-04-28 17:37 -0700
SubjectMore about scalability...
Message-ID<ljmhmo$bl5$1@news.albasani.net>
Hello,


I have asked you a question just before but nobody have answered...


So i have found a paper that is interresting to read , here it is:

http://htor.inf.ethz.ch/publications/img/hoefler-ramos-hpdc13-cc_modeling.pdf


So if you have noticed , what i was doing just before is analysing the
concurrent FIFO queue algorithms using the simple MESI protocol,
but that's not the correct way to do it... since the new architectures
are using extended MESI protocols,  that are scalable, look at for 
exemple Xeon Phi processor in the above link , you will notice
that it is using an extended MESI protocal and in its arhitecture
its cores are arranged on a  bidirectional ring bus that
provides high scalability, but here scalability means
that the cache-to-cache communication is scalable , that
means we can have many bidirectional cache-to-cache communications in 
running in parallel , so that makes the things different , cause
the Chriss Thomasson concurrent FIFO queue or my concurrent FIFO queue
here: http://pages.videotron.com/aminer/CQueue2.htm
will be more scalable with this kind of extended MESI protocol and
with this kind of scalable architecture and those kind of scalable 
architectures will allow our scalable concurrent FIFO queues that are to 
scale better and to score more throughput, cause those scalable kind of 
scalable architecures will lower the contention and makes our
algorithms scalable.



Thank you,
Amine Moulay Ramdane.

[toc] | [next] | [standalone]


#2297

FromKaz Kylheku <kaz@kylheku.com>
Date2014-04-28 22:04 +0000
Message-ID<20140428150351.401@kylheku.com>
In reply to#2296
On 2014-04-29, aminer <aminer@toto.net> wrote:
>
> Hello,
>
> I have asked you a question just before but nobody have answered...

Check in 15 minutes and re-post. Repeat if necessary.

[toc] | [prev] | [standalone]


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


csiph-web