Hi Waleed,
It is highly recommended to upgrade to latest OFED. Meanwhile, Can you try
latest OMPI release (v1.8.4), where this warning is ignored on older OFEDs
-Devendar
On Sun, Dec 28, 2014 at 6:03 AM, Waleed Lotfy
wrote:
> I have a bunch of 8 GB memory nodes in a cluster who were lately
>
dear All,
sorry for your time. I have found the solution:
icount=1
DO iCPU=0,MPI%nCPU-1
IF(iCPU.NE.MPI%rank)THEN
iTag=iCPU
CALL
MPI_ISEND(Ndata2send(iCPU),1,MPI_INTEGER,iCPU,iTag,MPI_COMM_WORLD,send_request(icount),MPI%iErr)
icount=icount+1
ENDIF
ENDDO
ic
Dear all,
I have the following problem:
In my program each rank has a vector, where the position indicates where I
have to send the data.
For example for rank 0, I have:
*ALLOCATE(Ndata2send(0,MPI%nCPU-1)*
*Ndata2send(:) = 0,10,10,16*
where MPI%nCPU is the number of cPU, in my case 4
Thins m
Do you mean OMPI did not issue such a warning ?
If there was no such warning, could you please detail
How much ram on your system
The value returned by ulimit -l
The ofed stack you are running
The kernel modules you loaded
The model of your ib card
Cheers
Gilles
Saliya Ekanayake さんのメール:
>I mean
I meant it works now, sorry for the confusion.
Running the test revealed a warning on memory registration, which we fixed
by setting unlimited in ulimit -l. Then running OMPI sample worked too.
Thank you,
saliya
On Sun, Dec 28, 2014 at 11:18 PM, Ralph Castain wrote:
> So you are saying the t
