Thank you very much guys,
I had this idea that "at" is not needed since I am 'already there' through
iteration.
I may come back for more stuff later but this definitely clears much of my
confusion.
Thank you again.
Haris
On Sun, Mar 28, 2010 at 9:00 PM, Igor Peshansky <ig...@us.ibm.com> wrote:
> This is a typical X10 beginner mistake: you specify types that
> are too imprecise. You cannot index into an Array with a Point
> unless it can be statically proven that the rank of that Point
> is exactly equal to the rank of the Array's region.
>
> The best rule of thumb is to let X10 infer the types for you.
> Barring that, you should look carefully at the methods you call
> to ensure that the rank is propagated correctly. Incidentally,
> to get efficient iteration, you'd want to propagate other
> pr0operties of the Array's region (e.g., the fact that it's
> rectangular).
>
> Another problem is that array elements have to be accessed in
> the place where they live (as specified by the distribution).
> This is not something that can be checked statically, but you
> will get a runtime exception if this rule is violated.
>
> So, you should rewrite your code as follows:
>
> public static def main(args:Rail[String]!){
> val R = 1..4; // The inferred type here is Region(1){rect}
> val D = Dist.makeCyclic(R); // The inferred type here is
> Dist{region==R)
>
> val data_squares = Array.make[Int](D, ((i):Point(1)) => i*i);
>
> for (val p in D) { // The inferred type here is Point(1)
> Console.OUT.println( "Array index: " + p + " distributed place :" +
> D(p));
>
> // this is the problem part, but this is how I understood arrays
>
> Console.OUT.println( "Array element: " + at (D(p)) data_squares(p));
>
> // how can I access e.g. data_squares(3) and its value and place.
> // using 2.0 java back end on eclipse
> }
> }
>
> Vijay already covered most of these points, but one that I wanted
> to bring up is that, unless you know exactly what the type is,
> better let the type system infer the most precise type. In
> particular, if you'd left the type of D as just Dist, the iteration
> would have been woefully inefficient (which would have hurt your
> peformance for larger arrays).
>
> As Vijay said, we are working on relaxing that requirement.
> Igor
>
> Haris H <hha...@gmail.com> wrote on 03/28/2010 03:57:47 PM:
>
> > Hi guys,
> > can somebody clear this up for me please.
> > I can not get my head around this element manipulation in distributed
> arrays
> > in 2.0.
> >
> > Here is the overly simplified code but represents my problem,
> > so please tell me what am I doing wrong and what it should be.
> >
> > I am having trouble understanding accessing elements of the array and
> their
> > distributed location.
> >
> > Thank you.
> >
> > public static def main(args:Rail[String]!){
> >
> > R: Region = 1..4;
> > D: Dist = Dist.makeCyclic(R);
> >
> > val data_squares = Array.make[Int](D, ((i):Point) => i*i);
> >
> > for (val p: Point in D) {
> > Console.OUT.println( "Array index: " + p + " distributed place :"
> +
> > D(p));
> >
> > // this is the problem part, but this is how I understood arrays
> >
> > Console.OUT.println( "Array element: " + data_squares(p));
> >
> > // how can I access e.g. data_squares(3) and its value and place.
> > // using 2.0 java back end on eclipse
> > }
> > }
> --
> Igor Peshansky (note the spelling change!)
> IBM T.J. Watson Research Center
> X10: Parallel Productivity and Performance (http://x10-lang.org/)
> XJ: No More Pain for XML's Gain (http://www.research.ibm.com/xj/)
> "I hear and I forget. I see and I remember. I do and I understand" --
> Confucius
>
>
>
> ------------------------------------------------------------------------------
> Download Intel® Parallel Studio Eval
> Try the new software tools for yourself. Speed compiling, find bugs
> proactively, and fine-tune applications for parallel performance.
> See why Intel Parallel Studio got high marks during beta.
> http://p.sf.net/sfu/intel-sw-dev
> _______________________________________________
> X10-users mailing list
> X10-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/x10-users
>
------------------------------------------------------------------------------
Download Intel® Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev
_______________________________________________
X10-users mailing list
X10-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/x10-users
