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

Groups > comp.programming > #3813 > unrolled thread

About my scalable distributed Lock...

Back to article view | Back to comp.programming

Contents

  About my scalable distributed Lock... aminer <aminer@toto.net> - 2013-09-30 17:16 -0700

#3813 — About my scalable distributed Lock...

Hello,


I have thought more about my scalable distributed Lock and
i have done some benchmarks and more tests, and i have found that
it has something better than the Windows critical section and better
than a Spinlock with an exponential backoff, if  the locked region
becomes bigger the backoff of the Windows critical section will
not be good cause it will allow the same thread to reenter the locked
region many many times and this is not good i think, cause i think
the Windows critical section must balance more and give each thread
a chance to run, so i think that the Windows critical section and also 
the Spinlock with exponential backoff are not suited for locked regions 
that are bigger, they are suited for small locked regions, this problem 
does not happen with my scalable distributed lock , cause if you take
a look at the source code, the first Spinlock(with an exponential 
backoff) inside the fifo queue that i am using have a small critical 
section so my scalable distributed Lock will balance more and give each 
thread a chance to run , this is where also  my scalable distributed 
Lock is more efficient than the Windows critical section and the 
Spinlock with an exponential backoff, and my scalable distributed Lock 
is best suited also for bigger locked regions, the Windows
critical section is not. This is why also it is very usefull and good.


Amine Moulay Ramdane.

[toc] | [standalone]

Back to top | Article view | comp.programming

csiph-web