MantisBT - ParaView
> Hi Joe,
> this is great! thanks for getting Berk involved.
> Re: PARALLEL - i think we will not have issues or may be only minor ones,
> and sequential version will be built relatively easy.
> Re BLAS/LAPACK:
> we can replace calls to BLAS level 1,2 and 3 with some simple routines that
> will do linear algebra.
> with LAPACK routines we will have to build LU factorization functions,
> linear solvers, eigenvalue solvers. this is possible but i do not think it
> is a good idea mainly for two reasons:
> a) BLAS and LAPACK are free libraries that can be downloaded and installed
> easily.
> b) this will limit (or make difficult) developing future versions of our
> plug-in as we would like to utilize as much as we can what is already
> existing in Nektar. also Nektar's users (potential customers) can create
> they own utilities and keep the same style as used in Nektar already and
> assume that BLAS and LAPACK are available.
> lets see what Berk can suggest, it might be that they also have some subset
> of BLAS/LAPACK built-in.
> Leopold
>
>
>
>
> On Thu, Mar 10, 2011 at 6:42 PM, joseph insley <insley@mcs.anl.gov> wrote:
>>
>> Hey, Leo,
>>
>> I have made a PV_NEKTAR tarball, which I believe includes the minimum
>> necessary code from Nektar-PARALLEL, and the actual reader plugin code. And
>> wrote a short README file explaining how to build the various components.
>> While doing this I took note of a few issues that will need to be addressed
>> at some point. (Sooner rather than later, will be better)
>>
>> External dependencies:
>> Blas
>> Lapack
>> gfortran
>>
>> I don't know whether these should be included in the distribution, or
>> users will need to have these installed separately, or what. I'm hoping
>> Berk can provide some guidance on that.
>>
>> MPI:
>> Currently I have only been building all of the components (both Nektar and
>> the plugin itself) assuming that it will be parallel. This means, I
>> believe, that the reader can only work in client/server mode. Also, I
>> believe that binary distributions of PV do not include MPI builds. Users
>> will need to compile from source in order to get the parallel version. If
>> we want users to be able to run standalone ParaView, on their laptop or
>> local workstation, for example, and use the Nektar reader plugin, I think we
>> are going to need to support a non-parallel version. I remember early on in
>> our development, while you were visiting ANL if I remember correctly, we
>> touched on this issue briefly, and decided to punt on it initially, and
>> worry about it later. I think now may be the time that we need to start
>> worrying about it. I will try building everything without PARALLEL=1 and
>> see how far I get. But I suspect that I will run into problems before too
>> long. When (if?) I do get stuck, I will poke you for help in resolving. I
>> just wanted to give you a heads-up about it.
>>
>> joe.
>
>
PV_NEKTAR.tar.gz (1,709,358) 2011-10-15 11:21https://www.vtk.org/Bug/file/9046/PV_NEKTAR.tar.gz