Charm++ parallel programming system

[Tom Quinn <trq AT astro.washington.edu>]

The delivery of the results is what I need prioritized.  I'll try 
"expedited".

Tom Quinn       Astronomy, University of Washington
Internet:       
trq AT astro.washington.edu
Phone:          206-685-9009

On Mon, 7 Nov 2016, Phil Miller wrote:

> Which part of the process needs to be prioritized?
> - The reduction
> - The delivery of the reduction's results
> - Both
>
> For the delivery of results, at least, you could mark the finishSmoothWalk() 
> entry
> method [expedited]. This is admittedly a heavy hammer, and may go further 
> than you'd
> like.
>
> We'd have to look into the possibility of increasing the priority of the 
> reduction
> itself.
>
> On Mon, Nov 7, 2016 at 1:07 PM, Tom Quinn 
> <trq AT astro.washington.edu>
>  wrote:
>       I sent this out a year ago, and I don't think I ever got a reply.  Any
>       ideas?
>
>       Tom Quinn       Astronomy, University of Washington
>       Internet:       
> trq AT astro.washington.edu
>       Phone:          206-685-9009
>
>       On Fri, 18 Sep 2015, Tom Quinn wrote:
>
>             This wasn't clear from the manual, and I think it might be a
>             good idea for one of ChaNGa's callbacks: can you raise the
>             priority of a callback? I'd like the finishSmoothWalk()
>             entry method to have high priority in the following code:
>
>             void TreePiece::markSmoothWalkDone()
>             {
>                    CkCallback cb =
>             CkCallback(CkIndex_TreePiece::finishSmoothWalk(),
>                                                pieces);
>                     // Use shadow array to avoid reduction conflict
>                     smoothProxy[thisIndex].ckLocal()->contribute(cb);
>                 }
>
>             void TreePiece::finishSmoothWalk()
>             {}
>
>             Tom Quinn       Astronomy, University of Washington
>             Internet:       
> trq AT astro.washington.edu
>             Phone:          206-685-9009