Josh originally wrote this to the chapel bugs and info lists, and there was
some discussion there on January 6 and 8, 2016, which I assume folks can
find using AltaVista or similar.
In any case, I wanted to let folks know that I have recently written a few
new ports or the Parallel Research Kernels (
https://github.com/ParRes/Kernels) that address Josh's issue.
New ports (of stencil, transpose, and sometimes synch_p2p) include:
- Rust. These are serial implementations and they are not only the first
but only Rust code I have ever written. They closely resemble the C++
implementation of the PRKs, and I don't claim to be a particularly good C++
programmer either.
- C++ w/ serial, OpenMP (host and target), OpenCL, TBB, and Cilk so far.
My TBB stencil code is incorrect, but there is a reasonable chance that it
will be fixed by the time anyone bothers to look.
- Fortran 2008 w/ serial, pretty (intrinsics), OpenMP (host and target),
and an accelerator directive programming model that shall not be named.
- Octave/Matlab.
- Python - both naive and Numpy.
These are in addition to other implementations written by a large group of
amazing volunteers and a few Intel employees:
- Chapel, the ports of which were recently enhanced by students at GWU,
after the initial port by a Cray intern (I think)
- Julia
- C w/ MPI-1, MPI-3 RMA, MPI-3 SHM, MPI-1 + OpenMP, OpenSHMEM
- C w/ MPI-1 that work with AMPI, FG-MPI and Fenix
- UPC
- Fortran 2008 coarrays
Anyways, I'm sorry it took so long for me to start working on the TBB port,
but please know that if you ask for something on the internet, there is a
non-zero probability that I will implement it some day :-)
Best,
Jeff
On Tue, Jan 5, 2016 at 8:32 PM, Joshua Olson <[email protected]>
wrote:
> Someone in the OpenVDB <http://www.openvdb.org> forum wanted to know if
> there's a recent comparison with Intel TBB. It seems there is not. Second,
> I just found out where to report issues, and it's nothing as nice as GitHub
> or JIRA.
>
> As someone wanting something more mature than Julia, Rust, or Nim
> <http://nim-lang.org> to port the sparse data structure behind much of 3D
> animation simulation, I am disappointed to find these problems. Does the
> Chapel team have a justification for only benchmarking against itself or
> using disparate SF lists riddled with spam?
>
> I'd appreciate some fear alleviation,
> Josh Olson
>
>
>
--
Jeff Hammond
[email protected]
http://jeffhammond.github.io/
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Chapel-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/chapel-users
