Re: [Gomp-discuss] Re: Implementing OpenMP pragmas for the C frontend

[Lars Segerlund]

Steven Bosscher wrote:
> Op ma 10-02-2003, om 09:48 schreef Lars Segerlund:
 Sounded worse when I read it myself :-( ... sorry  ...
> I wouldn't have put it in such strong words ;-)
...[snip]...
> Maybe one of the reasons for this is because people involved in GCC
> development mostly are computer scientists, and that such people are not
> well known for understanding computational scientists?

 Here I would like to add a comment, computer scientists produce NO or 
shitty code with a few notable exeptions, ( IMHO ), what GNU has done 
the scientific community should have done long ago, also they dont 
understand SPEED, in all my experience with computers, anything that 
gives you more speed is good, it has NO relevance on which theoretical 
foundations it rests. Se the famous linux vs. minix debate, all thise 
CS's that said that micro kernels were the way to go are gone today, in 
the same way that AI people spent 30 years mumbling on computer chess 
when somebodu started counting moves, thus blowing the whole field and 
all their multi millon lisp machines away.

...[snip]...

> And those numerical experts saw a trend: Everybody goes multi-processor!
> So they identified a *need* for an easy-to-use interface to create
> explicitly parallel software for people who are not computer
> scientists.  OpenMP was born, and is now widely used and well
> established.

 Explicit parallellisation is hard to beat, and I think openMP looks 
fine, small, easy and common for the major languages that the scientific 
community use.

> If GCC does not want to be useful in a hpc environment, that's fine. 
> But I would like to hear that *before* we start an effort to implement
> OpenMP in GCC.  If the GCC community has an attitude like, "just put it
> in a branch and we'll see what we do with it," then I can think of more
> important things that I actually should put time in.
>
> Greetz
> Steven

 Now this I also fully agree with, on many 'open' projects you are 
ready to devote some time and make things better, and then people have 
'opinions' on what to do and not to do and whats important, it makes you 
want to do something else, and generally you do.

 You have to be nice to new people, strive for quality and speed, and 
most of all, code . If you don't code you don't count ! ( or in another 
way contribute with testing or so on ).

 Fact less debate is not constructive !

 I have been following the gcc list and a long time complaint on gcc is 
that it's slow ! Now some of the new optimizations will take care of the 
generated code, but the scanning/parsing ? It gets worse, and the first 
thing you hear in the debate is often, no there is nothing wrong with X 
even when X has been benchmarked to cause the slowdown ?

 All I want to say is really, that this group has the aim to extend gcc 
to handle parallellism, more specifically openMP, ( though I think we 
should keep the door open in the implementation ), in such a way as to 
have no impact on the regular gcc.

 I don't think we have to consider 'opinions' if not expressed as the 
consensus of another team, I think we ned to get the code done.

 Also, I think there already is a major interest in openMP, since 
articles have started appearing about the intel compiler.

 So to sum it up, there is a standard, and a strong userbase of openMP 
in existance now, and interest will probably increase.

 It's not likely to be liked by computer science people, ( neither is 
FORTRAN, C , C++ , Linux and so on ), and they are bound to do what they 
fdo best, which is complain if they remedied these problems by producing 
usable code it would be nice.

 Sorry for being a bit negative, but I got really upset, also 
appologies to those who feel in any way offended, I don't want to start 
a flame, just to say that let's do what we set out to.

 / Lars Segerlund