Hi!
Iam trying to run a small example program using scalapack routines. This example is from tutorials on lam site. The program does a matrix vector mulitplication. Iam able to compile the code but not able to run it.
If i run it as
mpirun -v -np 4 mv_acml
iam getting the following error message...
1669 mv_acml running on n0
1413 mv_acml running on n1 (o)
1304 mv_acml running on n2
1288 mv_acml running on n3
-----------------------------------------------------------------------------
One of the processes started by mpirun has exited with a nonzero exit
code. This typically indicates that the process finished in error.
If your process did not finish in error, be sure to include a "return
0" or "exit(0)" in your C code before exiting the application.
PID 1669 failed on node n0 (10.5.0.1) due to signal 11.
-----------------------------------------------------------------------------
i guess signal 11 is a segmentation fault.Iam not sure why iam getting this.
I really appreciate your help.
Regards
Srinivasa Patri
Here is the code for reference
program parallel_mv
! global arrays
real a(16,16)
real b(16),y(16)
! variables for BLACS initialization and processor grid creation
integer iam,nprocs,ictxt,nprow,npcol,myrow,mycol
!variables needed for distributing global arrays across the proc grid
integer desca(9), descb(9),descy(9),m,n,mb,nb,rsrc,csrc
integer llda,lldb,info
! local arrays
real la(8,8)
real lb(8),ly(8)
!Initializing the BLACS library (STEP 2)
c write (*,*) 'Iam here before call to blacs_pinfo'
call blacs_pinfo (iam,nprocs)
c write (*,*) 'Iam here after call to blacs_pinfo'
call blacs_get(-1,0,ictxt)
! Creating and using the processor grid (STEP 3)
nprow=2; npcol=2
call blacs_gridinit(ictxt,'r',nprow,npcol)
call blacs_gridinfo(ictxt,nprow,npcol,myrow,mycol)
! Making the array descriptor vectors (STEP 4)
m=16; n=16
mb=8; nb=8
rsrc=0; csrc=0
llda=8
call descinit(desca,m,n,mb,nb,rsrc,csrc,ictxt,llda,info)
n=1; nb=1; lldb=8
call descinit(descb,m,n,mb,nb,rsrc,csrc,ictxt,lldb,info)
call descinit(descy,m,n,mb,nb,rsrc,csrc,ictxt,lldb,info)
! Filling the global arrays A,b
open(unit=12,file="a.data")
read(12,*) a
open(unit=13,file="b.data")
read(13,*) b
! Each processor fills in its local arrays with correct elements
! from the global arrays (STEP 5)
if(myrow.eq.0.and.mycol.eq.0) then
do i_loc=1,8
do j_loc=1,8
la(i_loc,j_loc)=a(i_loc,j_loc)
end do
lb(i_loc)=b(i_loc)
end do
end if
if(myrow.eq.1.and.mycol.eq.0) then
do i_loc=1,8
do j_loc=1,8
la(i_loc,j_loc)=a(i_loc+llda,j_loc)
end do
lb(i_loc)=b(i_loc+lldb)
end do
end if
if(myrow.eq.0.and.mycol.eq.1) then
do i_loc=1,8
do j_loc=1,8
la(i_loc,j_loc)=a(i_loc,j_loc+llda)
end do
end do
end if
if(myrow.eq.1.and.mycol.eq.1) then
do i_loc=1,8
do j_loc=1,8
la(i_loc,j_loc)=a(i_loc+llda,j_loc+llda)
end do
end do
end if
! Call the ScaLAPACK routine (STEP 6)
n=16
call psgemv('n',m,n,1.0,la,l,l,desca,lb,l,l,descb,1,0.0,ly,l,l,
& descy,l)
! Each processor prints out its part of the product vector y (STEP 7)
if(myrow.eq.0.and.mycol.eq.0) then
do i=1,8
print *,'PE:',myrow,mycol,' y(',i,')=',ly(i)
end do
end if
if(myrow.eq.1.and.mycol.eq.0) then
do i=1,8
print *,'PE:',myrow,mycol,' y(',i+lldb,')=',ly(i)
end do
end if
! Release the proc grid and BLACS library (STEP 8)
call blacs_gridexit(ictxt)
call blacs_exit(0)
end program parallel_mv
|
