Re: Compiler support for multicore architectures

kamal <kamalpr@hp.com>
Tue, 18 Nov 2008 20:33:44 -0800 (PST)

          From comp.compilers

Related articles
Compiler support for multicore architectures gauravgautam123@gmail.com (gaurav) (2008-11-18)
Re: Compiler support for multicore architectures kamalpr@hp.com (kamal) (2008-11-18)
Re: Compiler support for multicore architectures ian@airs.com (Ian Lance Taylor) (2008-11-18)
Re: Compiler support for multicore architectures m.helvensteijn@gmail.com (2008-11-19)
Re: Compiler support for multicore architectures walter@bytecraft.com (Walter Banks) (2008-11-19)
Re: Compiler support for multicore architectures toby@telegraphics.com.au (toby) (2008-11-20)
Re: Compiler support for multicore architectures jatin.bhateja@gmail.com (Jatin Bhateja) (2008-11-21)
Re: Compiler support for multicore architectures kamalpr@hp.com (kamal) (2008-11-23)
[3 later articles]
| List of all articles for this month |

From: kamal <kamalpr@hp.com>
Newsgroups: comp.compilers
Date: Tue, 18 Nov 2008 20:33:44 -0800 (PST)
Organization: Compilers Central
References: 08-11-086
Keywords: parallel
Posted-Date: 19 Nov 2008 19:51:00 EST

On Nov 19, 2:27 am, gaurav <gauravgautam...@gmail.com> wrote:
> hi,
>
> There has been a lot of discussions going on about compiler support
> for multicore architectures.
>
> I have a few basic questions.
>
> What different things can a compiler do to make mutithreaded programs
> run better on multicore? Isnt it much depends on the threading
> library , OS and the programmar instead on compiler ?


If the program is already a multi-threaded application, then you do
not need anything more from the compiler. But the OS and threads
library will have to ensure that sufficient number of LWPs are created
(for M:N) and that they are efficiently deployed to utilize all
available cores. There has been some work on the OS front -to schedule
lwps based on affinity.
  The thing that differentiates multiple cores within the same
processor die with multiple processors is their use of a common cache
and possibly pipelines and also the fact that communication from core
to core within the same processor die is faster than across processor
dies. Previously, communication overhead used to significantly hamper
scheduling parts of a program across processors, and this is lower for
cores within the same processor.


> It will be great to have your thoughts on this.
>
Most of the activity has been on adapting sequential programs to take
advantage of multi-core architectures because Moore's law is being hit
i.e. processor speeds are not doubling and so manufacturers are forced
to expand sideways and add more cores instead of increasing clock
frequency. But the sequential programs once written will not show an
improvement in speed unless the additional cores are utilized and that
is where compiler writers are trying to find ways to show performance
gain.


regards
-kamal


Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.