gomp-discuss
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Gomp-discuss] Start of an OpenMP Implementation Spec


From: Scott Robert Ladd
Subject: Re: [Gomp-discuss] Start of an OpenMP Implementation Spec
Date: Thu, 26 Aug 2004 11:35:49 -0400
User-agent: Mozilla Thunderbird 0.7.3 (X11/20040812)

Biagio Lucini wrote:
many thanks for your letter and welcome among us. I think we *all* are interested in moving gomp forward. The problem is that the project has big pushes when new and enthusiast members like you join it, but they won't last a forthnight. What we really need is someone who says "look: I'm here, ready to hack something" or better "I did this and that: what do you think"? Discussing theoretical proposals so far has brought us nowhere.

Perhaps "new and enthusiastic members" feel ignored and unappreciated.

Several months back, I posted a set of code examples, showing how basic OpenMP constructs could be realized using several parallelization models. I asked for comments and suggestions; the result was resounding silence from this group.

When GOMP began, I delivered an initial boilerplate for the support library, yet I've received very little feedback from this community.

I could finish much of the support library in a few short weeks -- but would that really move the project forward? Or would the code languish?

One initial "decision" was to implement for Linux, and then port later to other architectures. I disagree; in keeping with the principles of GNU and GCC, OpenMP should be based on an abstract model with platform-specific implementations. Opinions?

Lacking feedback on the concrete proposals I have made, I can only conclude:

1) No one here is really interested in realizing GOMP

2) My proposals are so awful that they are beyond comment

I'd be more than happy to lead, follow, or get out of the way, if I felt that any of those paths would move GOMP along. I've tried providing code and asking for feedback, without receiving much help; I've tried waiting for others to move ahead, and they have not.

I do not have all the knowledge required to implement OpenMP, nor do I have the financial resources to devote myself to OpenMP development.

Collaborative design is an iterative process; someone throws out an idea, others make suggestions, and a design evolves for practical implementation.

If you expect a fully-realized OpenMP design and implementation, I'm afraid that isn't going to happen without funding and some group effort. If you're waiting for someone to "hack something" in their spare time, you may wait a long time.

..Scott

--
Scott Robert Ladd
Coyote Gulch Productions (http://www.coyotegulch.com)
Software Invention for High-Performance Computing




reply via email to

[Prev in Thread] Current Thread [Next in Thread]