Re: [OMPI users] Concerning infiniband support

[Bowen Zhou]

On 01/20/2011 07:57 AM,

On Jan 20, 2011, at 7:51 AM, Bowen Zhou wrote:


 Besides all these advices have been given, you may need to use --prefix in 
configure script to override default installation directory since you don't 
have root account. Also you might want to look at MVAPICH as an alternative, an 
variant of MPICH2 that supports infiniband.


Ouch.  Such blasphemy on the Open MPI list hurts my poor little eyes...

;-)

(yes, that's a joke; all of us MPI people know each other.  Heck, we see each 
other every 2 months at the ongoing MPI-3 Forum meetings! :-) )

Haha, my bad. I found people in this list are equally capable of 
answering technical questions and making jokes. That's something unique. :-)

Re: [OMPI users] Concerning infiniband support

[Bowen Zhou]

Hi,

Besides all these advices have been given, you may need to use --prefix 
in configure script to override default installation directory since you 
don't have root account. Also you might want to look at MVAPICH as an 
alternative, an variant of MPICH2 that supports infiniband.


good luck,

Bowen Zhou
On 01/20/2011 01:59 AM,

Dear all,

I want to use infiniband, I am from a University in the US, my
University’s high performance center don’t have Gcc compiled openmpi
that support infiniband, so I want to compile myself.

But I have a few questions,

1.Is it ok to compile openmpi myself with infiniband support, if I don’t
have the root privilege? Is it going to work?

2.If it is ok, how can I find out where is the infiniband installation
directory, any shell command to show it?

3.Which configuration is correct? For example, using
“--with-openib=/usr/include/infiniband” as told in openmpi FAQ, or I
need to use "--with-openib=/usr/include/infiniband
--with-openib-libdir=/usr/lib64" both?


Thanks so much.

Daniel Wei

---

University of Notre Dame



___
users mailing list
us...@open-mpi.org
http://www.open-mpi.org/mailman/listinfo.cgi/users

Re: [OMPI users] MPI_Send doesn't work if the data >= 2GB

[Bowen Zhou]

On 12/05/2010 10:13 PM,

hi,

I met a question recently when I tested the MPI_send and MPI_Recv
functions.  When I run the following codes, the  processes hanged and I
found there was not data transmission in my network at all.

BTW: I finished this test on two X86-64 computers with 16GB memory and
installed Linux.

  1 #include 
  2 #include 
  3 #include 
  4 #include 
  5
  6
  7 int main(int argc, char** argv)
  8 {
  9   int localID;
  10 int numOfPros;
  11 size_t Gsize = (size_t)2 * 1024 * 1024 * 1024;
  12
  13 char* g = (char*)malloc(Gsize);
  14
  15 MPI_Init(, );
  16 MPI_Comm_size(MPI_COMM_WORLD, );
  17 MPI_Comm_rank(MPI_COMM_WORLD, );
  18
  19 MPI_Datatype MPI_Type_lkchar;
  20 MPI_Type_contiguous(2048, MPI_BYTE, _Type_lkchar);
  21 MPI_Type_commit(_Type_lkchar);
  22
  23 if (localID == 0)
  24 {
  25 MPI_Send(g, 1024*1024, MPI_Type_lkchar, 1, 1, MPI_COMM_WORLD);
  26 }
  27
  28 if (localID != 0)
  29 {
  30 MPI_Status status;
  31 MPI_Recv(g, 1024*1024, MPI_Type_lkchar, 0, 1, \
  32 MPI_COMM_WORLD, );
  33 }
  34
  35 MPI_Finalize();
  36
  37 return 0;
  38 }

Thanks
Xianjun



___
users mailing list
us...@open-mpi.org
http://www.open-mpi.org/mailman/listinfo.cgi/users

Hi,

It might be a problem with the Linux TCP stack. The similar issue 
happens with MPICH2 too. Can you try with message size (2G-1) bytes?

http://trac.mcs.anl.gov/projects/mpich2/ticket/1080

Best,
Bowen Zhou

Re: [OMPI users] How to run different versions of application in the same run?

[Bowen Zhou]

Thanks a lot! That's exactly what I was looking for.




See here:
http://www.open-mpi.org/faq/?category=running#mpmd-run



On Tue, 2010-10-12 at 22:21 -0400, Bowen Zhou wrote:

Greetings,

I'm doing software fault injection in a parallel application to evaluate 
the effect of hardware failures to the execution. My question is how to 
execute the faulty version of the application on one node and the 
fault-free version on all other nodes in the same run?


I understand that mpirun or mpiexec would require a globally accessible 
path to the same executable mounted with NFS or some other file system. 
So is there any way to specify different pathnames in different nodes?


Many thanks,

Bowen Zhou
___
users mailing list
us...@open-mpi.org
http://www.open-mpi.org/mailman/listinfo.cgi/users


___
users mailing list
us...@open-mpi.org
http://www.open-mpi.org/mailman/listinfo.cgi/users

[OMPI users] How to run different versions of application in the same run?

[Bowen Zhou]

Greetings,

I'm doing software fault injection in a parallel application to evaluate 
the effect of hardware failures to the execution. My question is how to 
execute the faulty version of the application on one node and the 
fault-free version on all other nodes in the same run?


I understand that mpirun or mpiexec would require a globally accessible 
path to the same executable mounted with NFS or some other file system. 
So is there any way to specify different pathnames in different nodes?


Many thanks,

Bowen Zhou

Re: [OMPI users] "self scheduled" work & mpi receive???

[Bowen Zhou]

Hi All:

I’ve written an openmpi program that “self schedules” the work.  

The master task is in a loop chunking up an input stream and handing off 
jobs to worker tasks.  At first the master gives the next job to the 
next highest rank.  After all ranks have their first job, the master 
waits via an MPI receive call for the next free worker.  The master 
parses out the rank from the MPI receive and sends the next job to this 
node.  The jobs aren’t all identical, so they run for slightly different 
durations based on the input data.


 

When I plot a histogram of the number of jobs each worker performed, the 
lower mpi ranks are doing much more work than the higher ranks.  For 
example, in a 120 process run, rank 1 did 32 jobs while rank 119 only 
did 2.  My guess is that openmpi returns the lowest rank from the MPI 
Recv when I’ve got MPI_ANY_SOURCE set and multiple sends have happened 
since the last call.


 


Is there a different Recv call to make that will spread out the data better?

 
How about using MPI_Irecv? Let the master issue an MPI_Irecv for each 
worker and call MPI_Test to get the list of idle workers, then choose 
one from the idle list by some randomization?




THANXS!

amb

 





___
users mailing list
us...@open-mpi.org
http://www.open-mpi.org/mailman/listinfo.cgi/users

Re: [OMPI users] OpenMPI load data to multiple nodes

[Bowen Zhou]

Since each node has its own memory in a distributed memory system,
there is no such thing as "global variable" that can be accessed by all
processes. So you need to use MPI to scatter the input from rank 0
process to all the other processes explicitly.

From: dtustud...@hotmail.com
To: us...@open-mpi.org
List-Post: users@lists.open-mpi.org
Date: Mon, 12 Jul 2010 20:33:04 -0600
Subject: Re: [OMPI users] OpenMPI load data to multiple nodes








thanks very much !!!
May I use global variable to do that ? 
It means that all nodes have the same global variable, such as globalVector. 
In the initialization, only node 0 load data from files and assign values to 
the globalVector. 
After that, all other nodes can get the same data by accessing the 
globalVector. 
Does it make sense ? 

Any help is appreciated. 
Jack
July 12  2010
> Date: Mon, 12 Jul 2010 21:44:34 -0400
> From: g...@ldeo.columbia.edu
> To: us...@open-mpi.org
> Subject: Re: [OMPI users] OpenMPI load data to multiple nodes
> 
> Hi Jack/Jinxu
> 
> Jack Bryan wrote:
> > Dear All,
> > 
> > I am working on a multi-computer Open MPI cluster system. 
> > 
> > If I put some data files in /home/mypath/folder, is it possible that all 
> > non-head nodes can access the files in the folder ? 
> >
> 
> Yes, possible, for instance, if the /home/mypath/folder directory is
> NFS mounted on all nodes/computers.
> Otherwise, if all disks and directories are local to each computer,
> you need to copy the input files to the local disks before you
> start, and copy the output files back to your login computer after the
> program ends.
> 
> > I need to load some data to some nodes, if all nodes can access the 
> > data, I do not need to load them to each node one by one. 
> > 
> > If multiple nodes access the same file to get data, is there conflict ? 
> > 
> 
> To some extent.
> The OS (on the computer where the file is located)
> will do the arbitration on which process gets the hold of the file at 
> each time.
> If you have 1000 processes, this means a lot of arbitration,
> and most likely contention.
> Even for two processes only, if the processes are writing data to a 
> single file, this won't ensure that they write
> the output data in the order that you want.
> 
> > For example, 
> > 
> > fopen(myFile) by node 1, at the same time fopen(myFile) by node 2. 
> > 
> > Is it allowed to do that on MPI cluster without conflict ? 
> > 
> 
> I think MPI won't have any control over this.
> It is up to the operational system, and depends on
> which process gets its "fopen" request to the OS first,
> which is not a deterministic sequence of events.
> That is not a clean technique.
> 
> You could instead:
> 
> 1) Assign a single process, say, rank 0,
> to read and write data from/to the file(s).
> Then use, say, MPI_Scatter[v] and MPI_Gather[v],
> to distribute and collect the data back and forth
> between that process (rank 0) and all other processes.
> 
> That is an old fashioned but very robust technique.
> It avoids any I/O conflict or contention among processes.
> All the data flows across the processes via MPI.
> The OS receives I/O requests from a single process (rank 0).
> 
> Besides MPI_Gather/MPI_Scatter, look also at MPI_Bcast,
> if you need to send the same data to all processes,
> assuming the data is being read by a single process.
> 
> 2) Alternatively, you could use the MPI I/O functions,
> if your files are binary.
> 
> I hope it helps,
> Gus Correa
> 
> > Any help is appreciated. 
> > 
> > Jinxu Ding
> > 
> > July 12  2010
> > 
> > 
> > The New Busy think 9 to 5 is a cute idea. Combine multiple calendars 
> > with Hotmail. Get busy. 
> > 
> > 
> > 
> > 
> > 
> > ___
> > users mailing list
> > us...@open-mpi.org
> > http://www.open-mpi.org/mailman/listinfo.cgi/users
> 
> ___
> users mailing list
> us...@open-mpi.org
> http://www.open-mpi.org/mailman/listinfo.cgi/users

The New Busy is not the old busy. Search, chat and e-mail from your inbox. Get 
started.   
_
Your E-mail and More On-the-Go. Get Windows Live Hotmail Free.
https://signup.live.com/signup.aspx?id=60969

[OMPI users] Any example for noncrashing bugs in MPI applications

[Bowen Zhou]

Hi,

I am developing a runtime detector for noncrashing 
bugs in MPI applications. It is capable of collecting
MPI communication and diagnose if a node is abnormal 
based on its communication with other nodes.

Therefore, it is highly desirable to have a couple of 
noncrashing bugs from real-life MPI application 
 that would manifest themselves in abnormal 
communication to test my tool. Or if someone 
experienced in development with MPI could give 
advice on the frequency of such bugs happening 
in the past?

Thank you,

Bowen

_
Hotmail: Free, trusted and rich email service.
https://signup.live.com/signup.aspx?id=60969