Re: NCL fortran->Nvidia GPU porting

From: Saji Hameed <saji.nh_at_nyahnyahspammersnyahnyah>
Date: Wed Aug 17 2011 - 16:38:31 MDT

Hi Rick and Elmer,

Very interesting, especially to hear about the OpenCL part. A friend
of mine, who is quite knowledgeable about these things, is helping me
with porting an atmospheric dispersion model (FLEXPART) into OpenCL.
What is of interest is that he is developing a few tools that could
help with similar efforts as yours. These include a utility to
optimize the Fortran code (taking care of global variables among other
things), and Fortran to C conversion and possibly automatic generation
of a large part of OpenCL code. Please checkout
https://github.com/wimvanderbauwhede/flx_wrf2 for more information.
We are looking for collaborators and contributors.

cheers,

saji

On Thu, Aug 18, 2011 at 12:34 AM, Rick Brownrigg <brownrig@ucar.edu> wrote:
> Hi Elmer,
>
> Those are some impressive gains, and yes, I have to think people would want more! ;-)
>
> I believe Parvis uses MPI and targets distributed compute nodes. In an orthogonal effort, this summer we began looking into how we might leverage GPUs found in commodity machines. We examined just a handful of potential operations that could be parallelized, and the results were positive (not nearly as dramatic as yours, however).
>
> One issue of concern for us is portability. To address that, our implementation leveraged OpenCL, which can target multiple vendors' GPUs as well as multicore CPUs.
>
> What are your thoughts in terms of sharing your efforts?  Do you have other functions/algorithms in mind that might be ripe for parallelizing?
>
> Thanks for letting us know about your efforts.
>
> Rick
>
> On Aug 16, 2011, at 10:05 AM, Elmer Joandi wrote:
>
>> Hi,
>> I ported in springtime for fun  cape 2d calculus from NCL fortran to GPU, got it about 100x faster (dual versus single precision, on singe corei7 CPU core versus casual Nvidia GPU, 300 secs versus 3 secs on 500x500 domain ).
>>
>> I wonder -
>>   a) what is the legal status of this code ?
>>   b) would anyone want it or more of such porting ?
>>   c) would ParVis have MPI or GPU or both   ?
>>
>>
>> --
>> Elmer Joandi
>> +3725185428
>> _______________________________________________
>> ncl-talk mailing list
>> List instructions, subscriber options, unsubscribe:
>> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>
> _______________________________________________
> ncl-talk mailing list
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>
_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Wed Aug 17 16:38:38 2011

This archive was generated by hypermail 2.1.8 : Wed Sep 07 2011 - 10:58:58 MDT