What is the rationale for such low priority for OpenMP support on FreeBSD?

estrabd

estrabd

I suppose I am late to this party.

So I see that clang is to become the base compiler in FreeBSD 10. I hope this greatly accelerates LLVM/clang's support of things like OpenMP. It saddens me to see FreeBSD continue to shy away from solid, modern support for OpenMP. From what I can see, LLVM/clang has OpenMP on its radar, but it's not going to mature for quite a while.

I suppose if anything is going to drive this need in clang, it'll be that it's the base compiler for an OS that is supposedly interested in high scalability on SMP. They've been stuck a base compiler (gcc 4.2.1) for ages, and it also happens to be the earliest gcc with GOMP integration.

Now they're going to one that has no OpenMP support; I guess I just don't understand it other than, like I said above, it'll drive clang features very rapidly.

Ultimately, I am continuously dismayed at some of these decisions FreeBSD makes that are providing less base-system access to the SMP power it's so obviously hungry for. :r
 
Even though this is an open source project, I see that things happen as much for business reasons (from my limited point of view). For example, the FreeBSD community wan't wants a new installer, wants a new BSD licensed compiler, and wants better multi-processing/parallel programming. However, Apple also wants BSD Licensed compiler and iXSystems (PC-BSD) wants a new installer. So for business reasons those two things are getting done. Other projects are done as organization decisions by the FreeBSD Foundation, such as the new graphic work being done.

I think that OpenMP need to be pushed into the community by some entity, person, foundation, business.

Is this question coming from you or from your business? Can you be this entity to push it in?
 
You must log in or register to reply here.
Back
Top