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

Groups > comp.programming.threads > #1736

Re: Spinlock and starvation...

Cross-posted to 3 groups.

Show all headers | View raw

And look at the graph of the "Lock comparison" up to 80 cores
the test&set with an exponential backoff have a good performance,
so no need for an MCS queue Lock, so i will return back to
the SpinLock with an exponential backoff cause it has a good performance.

Here is the graph bellow on this PDF page:

http://www.cs.rice.edu/~vs3/comp422/lecture-notes/comp422-lec19-s08-v1.pdf




Thank you,
Amine Moulay Ramdane.

On 9/23/2013 5:31 PM, aminer wrote:
>
> Hello,
>
> Finally , i have read the following, and it says this:
>
> "Starvation freedom is desirable, but not essential
> —practical locks: many permit starvation, although it is unlikely to occur"
>
> read here:
>
> http://www.cs.rice.edu/~vs3/comp422/lecture-notes/comp422-lec19-s08-v1.pdf
>
> So as you have just read , starvation is unlikely to occur with spinlock
> eith backoff, so the probability is
> very low that starvation occurs, so i will return back to the Spinlock
> with exponential backoff inside my SemaMonitor and SemaCondvar cause it
> has a decent performance.
>
>
>
> Thank you,
> Amine Moulay Ramdane.
>
>
>
>
>
>
>
>

Back to comp.programming.threads | Previous | Next — Previous in thread | Find similar

Thread

Spinlock and starvation... aminer <aminer@toto.net> - 2013-09-23 17:31 -0700
  Re: Spinlock and starvation... aminer <aminer@toto.net> - 2013-09-23 17:46 -0700

csiph-web