www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - Memory reordering explained by example

reply deadalnix <deadalnix gmail.com> writes:
http://preshing.com/20120515/memory-reordering-caught-in-the-act

And this is why shared is important for D.
May 16 2012
parent reply Gor Gyolchanyan <gor.f.gyolchanyan gmail.com> writes:
The problem is, that ancient processor architectures are used for modern
processors and software.
The correct solution to the concurrency problems would be a new
architecture, designed to naturally deal with concurrency.

On Wed, May 16, 2012 at 12:44 PM, deadalnix <deadalnix gmail.com> wrote:

 http://preshing.com/20120515/**memory-reordering-caught-in-**the-act<http://preshing.com/20120515/memory-reordering-caught-in-the-act>

 And this is why shared is important for D.

-- Bye, Gor Gyolchanyan.
May 16 2012
next sibling parent deadalnix <deadalnix gmail.com> writes:
Le 16/05/2012 12:59, Gor Gyolchanyan a écrit :
 The problem is, that ancient processor architectures are used for modern
 processors and software.
 The correct solution to the concurrency problems would be a new
 architecture, designed to naturally deal with concurrency.

 On Wed, May 16, 2012 at 12:44 PM, deadalnix <deadalnix gmail.com
 <mailto:deadalnix gmail.com>> wrote:

     http://preshing.com/20120515/__memory-reordering-caught-in-__the-act
     <http://preshing.com/20120515/memory-reordering-caught-in-the-act>

     And this is why shared is important for D.




 --
 Bye,
 Gor Gyolchanyan.

It have a very high cost to do so. It would prevent many thead local optimization and, in practice, most data are thread local. It isn't a good idea. We need languages and frameworks that deal with that.
May 16 2012
prev sibling parent Sean Cavanaugh <WorksOnMyMachine gmail.com> writes:
On 5/16/2012 5:59 AM, Gor Gyolchanyan wrote:
 The problem is, that ancient processor architectures are used for modern
 processors and software.
 The correct solution to the concurrency problems would be a new
 architecture, designed to naturally deal with concurrency.

We have them, they are called GPUs
May 18 2012