Hello,
Can someone tell me how to run an X10 2.0 program on places that reside on
different nodes of a cluster?
If I use the C++ backend, I should be using a launcher like
mpirunx10 -x X10_NTHREADS= -np --hostfile
What should I be doing if I am using the JAVA backend?
Thanks,
Kshitij
---
Hello, I am using x10 1.7 and the java backend. I am trying to parallelize
nested *for* loops.
This is the original loop:
for(var i:int = 1; i < n1;i++){
for(var j:int = 1; j < n2; j++){
for(var k:int = 1; khttp://p.sf.net/sfu/google-dev2dev
___
Thats exactly what I want. I want an array to be available/replicated in all
places.
Also, I am not clear why I get the classCastException.
Thanks, Kshitij
On Mon, Dec 7, 2009 at 5:20 PM, Igor Peshansky wrote:
> Kshitij Mehta wrote on 12/07/2009 05:53:30 PM:
>
> > Is there a w
Is there a way to make a variable global such that it is available on all
places? I want to avoid declaring it on one place and fetching it at other
places using a future.
Also, I get a classCastException at the foll.statement:
finish ateach(val p(t):Point(1) in Dist.makeUnique(Place.places))
example on how to fix it?
Please note that I cannot make the array a val object, it has to be a 'var'.
Thanks,
Kshitij Mehta
Graduate Student
Parallel Software Technologies Lab
Dept. of Computer Science
University of Houston
Texas, USA
www.cs.uh.e
I forgot to mention that we are running this in the Linux environment.
On Nov 12, 2009, at 2:03 PM, Kshitij Mehta wrote:
> Hello
> We are trying to convert a C code to x10 and parallelize it. The C code
> (sequential) takes 17 seconds to execute. We have converted the co
using the C++
backend takes ~12 minutes. We are using x10 v1.7.
Any ideas on why it could be taking so long?
Kshitij Mehta
Graduate Student
Parallel Software Technologies Lab
Dept. of Computer Science
University of Houston
USA
www.cs.uh.edu/~kmehta
>
> mpirunx10 -x X10_NTHREADS= -np --hostfile
>
>
>
> Igor Peshansky wrote:
>>
>> Kshitij Mehta wrote on 11/10/2009 01:08:14 PM:
>>
>>
>>> Hello
>>> Is there a way to set the no. of places in the Linux environment?
>>> Looks lik
ocess (corresponding to a place) is spawned on a different node.
In case of a shared memory machine (single node) can I be sure that
each place is mapped to a different core? Its ok if I cannot verify
that through the language.
On Nov 10, 2009, at 2:36 PM, Igor Peshansky wrote:
> Kshitij Meht
x27;javac ... -J-Xms256m -J-Xmx256m'
How can I do that in x10 ?
Thanks,
Kshitij Mehta
Graduate Student
Parallel Software Technologies Lab
Dept. of Computer Science
University of Houston
USA
www.cs.uh.edu/~kmehta
--
cluster, can I verify that the ateach block is indeed being run on
separate nodes ?
Something like:
ateach {
Console.OUT.println("I am place: " + here.id + " on node: " +
node_id);
}
Thanks,
Kshitij Mehta
Graduate Student
Parallel Software Technologies Lab
Dept. of
Hello
Is there a way to set the no. of places in the Linux environment?
Looks like the default value is 4, which was set by the x10 runtime
during installation. Please note that I am not using eclipse.
Thanks,
Kshitij Mehta
Graduate Student
Parallel Software Technologies Lab
Dept. of Computer
.
Is there another way to create a 5D array?
Thanks,
Kshitij Mehta
Graduate Student
Parallel Software Technologies Lab
Dept. of Computer Science
University of Houston
USA
www.cs.uh.edu/~kmehta
--
Come build with us! The
Hello
In the following snippet of code, what 'x:nat' in the second statement
mean?
val data: Rail[Int];
data = Rail.makeVar[Int](n, (x:nat)=>0);
Thanks,
Kshitij Mehta
Graduate Student
Parallel Software Technologies Lab
Dept. of Computer Science
University of Houston
USA
www.cs.u
14 matches
Mail list logo