Re: [Boost-users] Boost & OpenMP

Subject: Re: [Boost-users] Boost & OpenMP
From: Raymond Wan (r.wan_at_[hidden])
Date: 2010-04-07 06:43:24

Hi Robert,

Robert Jones wrote:
> - In some ways this is an alternative to Boost thread pools.
> - Using explicit pragmas is an intermediate step, and ultimately it might be
> hoped that a
>   compiler would be able to do all this without external direction.
> - The whole OpenMP notion is too specific for a general library like Boost

I, too, have used OpenMP and noticed significant improvements by using it.  It would be nice if 
Boost supported it.  I'm not very knowledgeable in OpenMP or threads -- but I won't let that stop me 
from giving an opinion.  :-)

I have never used Boost.thread before, but looking at its documentation now, it looks like its 
purpose is to allow developers to create multithreaded software from scratch.  So, various functions 
are needed to manage communication between threads, etc.  On the other hand, OpenMP's strength is to 
allow users to take a sequential program and parallelize it by indicating which variables [in a 
loop, etc.] do not conflict with each other.  So, as for point #1, I don't think it is an 
alternative to threads -- their motivations are slightly different.

As for your second point, I don't know what compiler research is nowadays, but I would think it 
would be difficult to determine automatically if two variables will never conflict.  In fact, doing 
it by hand with OpenMP, I got it wrong once or twice -- poor eyesight + poor variable names.  :-)

Putting this together, I think OpenMP isn't the same as Boost.threads or even Boost.mpi and I'm not 
so sure how Boost could fit with OpenMP -- it seems more suitable as pragmas to the compiler. 
However, I don't mind being proved wrong as something more Boost-like might be more intuitive to 
users than the #pragmas...

Ray