fdurastante/psblas3
1
0
Fork 0
mirror of https://github.com/sfilippone/psblas3.git synced 2026-10-16 02:49:03 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fix sphalo to have calls to CSGET with full output.

Browse Source
This commit is contained in:
Salvatore Filippone
2019-04-11 12:19:58 +01:00
parent 8fc4f82ff9
commit a019bfcf33
5 changed files with 145 additions and 85 deletions
Download Patch File Download Diff File Expand all files Collapse all files
Changelog
+9 -1
View File
@@ -1,6 +1,14 @@
Changelog. A lot less detailed than usual, at least for past
history.
2018/11/23: Reimplement hash function.
2019/01/16: In bldext, implement ALL_TO_ALLV by hand for certain
compilers.
2018/10/10: New ICTXT argument in prec%init
2018/07/30: Improved implementations for TRIL/TRIU
2018/04/15: Fix internals to have tmp_ovrlap in local indexing. Change
default in CDALL with VL to no global checks.
2018/03/22: Fix defaults for matrix/vector internals
2018/02/03: Accept 'GMRES' as synonim with 'RGMRES'.
2018/11/23: Reimplement hash function.
2018/10/10: new ICTXT argument to prec%init method.
2018/09/04: Modified vector class get_vect method
2018/08/10: Optional arguments in GETROW method.
base/tools/psb_csphalo.F90
+34 -21
View File
@@ -80,7 +80,7 @@ Subroutine psb_csphalo(a,desc_a,blk,info,rowcnv,colcnv,&
& n_el_send,k,n_el_recv,idx, r, tot_elem,&
& n_elem, j, ipx,mat_recv, iszs, iszr,idxs,idxr,nz,&
& irmin,icmin,irmax,icmax,data_,ngtz,totxch,nxs, nxr,&
& l1, err_act
& l1, err_act, nsnds, nrcvs, nrg, ncg
integer(psb_mpk_) :: icomm, minfo
integer(psb_mpk_), allocatable :: brvindx(:), &
& rvsz(:), bsdindx(:),sdsz(:)
@@ -187,7 +187,7 @@ Subroutine psb_csphalo(a,desc_a,blk,info,rowcnv,colcnv,&
call desc_a%get_list(data_,pdxv,totxch,nxr,nxs,info)
ipdxv = pdxv%get_vect()
! For all rows in the halo descriptor, extract and send/receive.
! For all rows in the halo descriptor, extract the row size
Do
proc=ipdxv(counter)
if (proc == -1) exit
@@ -196,8 +196,8 @@ Subroutine psb_csphalo(a,desc_a,blk,info,rowcnv,colcnv,&
n_el_send = ipdxv(counter+psb_n_elem_send_)
tot_elem = 0
Do j=0,n_el_send-1
idx = ipdxv(counter+psb_elem_send_+j)
n_elem = a%get_nz_row(idx)
idx = ipdxv(counter+psb_elem_send_+j)
n_elem = a%get_nz_row(idx)
tot_elem = tot_elem+n_elem
Enddo
sdsz(proc+1) = tot_elem
@@ -205,6 +205,12 @@ Subroutine psb_csphalo(a,desc_a,blk,info,rowcnv,colcnv,&
counter = counter+n_el_send+3
Enddo
!
! Exchange row sizes, so as to know sends/receives.
! This is different from the halo exchange because the
! size of the rows may vary, as opposed to fixed
! (multi) vector row size.
!
call mpi_alltoall(sdsz,1,psb_mpi_mpk_,&
& rvsz,1,psb_mpi_mpk_,icomm,minfo)
if (info /= psb_success_) then
@@ -212,6 +218,10 @@ Subroutine psb_csphalo(a,desc_a,blk,info,rowcnv,colcnv,&
call psb_errpush(info,name,a_err='mpi_alltoall')
goto 9999
end if
nsnds = count(sdsz /= 0)
nrcvs = count(rvsz /= 0)
if (debug_level >= psb_debug_outer_)&
& write(debug_unit,*) me,' ',trim(name),': Done initial alltoall',nsnds,nrcvs
idxs = 0
idxr = 0
@@ -224,10 +234,10 @@ Subroutine psb_csphalo(a,desc_a,blk,info,rowcnv,colcnv,&
n_el_send = ipdxv(counter+psb_n_elem_send_)
bsdindx(proc+1) = idxs
idxs = idxs + sdsz(proc+1)
idxs = idxs + sdsz(proc+1)
brvindx(proc+1) = idxr
idxr = idxr + rvsz(proc+1)
counter = counter+n_el_send+3
idxr = idxr + rvsz(proc+1)
counter = counter+n_el_send+3
Enddo
iszr=sum(rvsz)
@@ -237,7 +247,7 @@ Subroutine psb_csphalo(a,desc_a,blk,info,rowcnv,colcnv,&
& ' Send:',sdsz(:),' Receive:',rvsz(:)
mat_recv = iszr
iszs=sum(sdsz)
if (info == psb_success_) call psb_ensure_size(max(iszs,1),iasnd,info)
call psb_ensure_size(max(iszs,1),iasnd,info)
if (info == psb_success_) call psb_ensure_size(max(iszs,1),jasnd,info)
if (info == psb_success_) call psb_ensure_size(max(iszs,1),valsnd,info)
#if defined(IPK4) && defined(LPK8)
@@ -337,29 +347,33 @@ Subroutine psb_csphalo(a,desc_a,blk,info,rowcnv,colcnv,&
ipx = 1
counter=1
idx = 0
tot_elem=0
!
! Make sure to get all columns in csget.
! This is necessary when sphalo is used to compute a transpose,
! as opposed to just gathering halo for spspmm purposes.
!
ncg = huge(ncg)
tot_elem = 0
Do
proc=ipdxv(counter)
proc = ipdxv(counter)
if (proc == -1) exit
n_el_recv=ipdxv(counter+psb_n_elem_recv_)
counter=counter+n_el_recv
n_el_send=ipdxv(counter+psb_n_elem_send_)
n_el_recv = ipdxv(counter+psb_n_elem_recv_)
counter = counter+n_el_recv
n_el_send = ipdxv(counter+psb_n_elem_send_)
Do j=0,n_el_send-1
idx = ipdxv(counter+psb_elem_send_+j)
n_elem = a%get_nz_row(idx)
call a%csget(idx,idx,ngtz,iasnd,jasnd,valsnd,info,&
& append=.true.,nzin=tot_elem)
& append=.true.,nzin=tot_elem,jmax=ncg)
if (info /= psb_success_) then
info=psb_err_from_subroutine_
ch_err='psb_sp_getrow'
call psb_errpush(info,name,a_err=ch_err)
call psb_errpush(info,name,a_err='psb_sp_getrow')
goto 9999
end if
tot_elem=tot_elem+n_elem
tot_elem = tot_elem+ngtz
Enddo
ipx = ipx + 1
ipx = ipx + 1
counter = counter+n_el_send+3
Enddo
nz = tot_elem
@@ -368,8 +382,7 @@ Subroutine psb_csphalo(a,desc_a,blk,info,rowcnv,colcnv,&
if (colcnv_) call psb_loc_to_glob(jasnd(1:nz),desc_a,info,iact='I')
if (info /= psb_success_) then
info=psb_err_from_subroutine_
ch_err='psb_loc_to_glob'
call psb_errpush(info,name,a_err=ch_err)
call psb_errpush(info,name,a_err='psb_loc_to_glob')
goto 9999
end if
base/tools/psb_dsphalo.F90
+34 -21
View File
@@ -80,7 +80,7 @@ Subroutine psb_dsphalo(a,desc_a,blk,info,rowcnv,colcnv,&
& n_el_send,k,n_el_recv,idx, r, tot_elem,&
& n_elem, j, ipx,mat_recv, iszs, iszr,idxs,idxr,nz,&
& irmin,icmin,irmax,icmax,data_,ngtz,totxch,nxs, nxr,&
& l1, err_act
& l1, err_act, nsnds, nrcvs, nrg, ncg
integer(psb_mpk_) :: icomm, minfo
integer(psb_mpk_), allocatable :: brvindx(:), &
& rvsz(:), bsdindx(:),sdsz(:)
@@ -187,7 +187,7 @@ Subroutine psb_dsphalo(a,desc_a,blk,info,rowcnv,colcnv,&
call desc_a%get_list(data_,pdxv,totxch,nxr,nxs,info)
ipdxv = pdxv%get_vect()
! For all rows in the halo descriptor, extract and send/receive.
! For all rows in the halo descriptor, extract the row size
Do
proc=ipdxv(counter)
if (proc == -1) exit
@@ -196,8 +196,8 @@ Subroutine psb_dsphalo(a,desc_a,blk,info,rowcnv,colcnv,&
n_el_send = ipdxv(counter+psb_n_elem_send_)
tot_elem = 0
Do j=0,n_el_send-1
idx = ipdxv(counter+psb_elem_send_+j)
n_elem = a%get_nz_row(idx)
idx = ipdxv(counter+psb_elem_send_+j)
n_elem = a%get_nz_row(idx)
tot_elem = tot_elem+n_elem
Enddo
sdsz(proc+1) = tot_elem
@@ -205,6 +205,12 @@ Subroutine psb_dsphalo(a,desc_a,blk,info,rowcnv,colcnv,&
counter = counter+n_el_send+3
Enddo
!
! Exchange row sizes, so as to know sends/receives.
! This is different from the halo exchange because the
! size of the rows may vary, as opposed to fixed
! (multi) vector row size.
!
call mpi_alltoall(sdsz,1,psb_mpi_mpk_,&
& rvsz,1,psb_mpi_mpk_,icomm,minfo)
if (info /= psb_success_) then
@@ -212,6 +218,10 @@ Subroutine psb_dsphalo(a,desc_a,blk,info,rowcnv,colcnv,&
call psb_errpush(info,name,a_err='mpi_alltoall')
goto 9999
end if
nsnds = count(sdsz /= 0)
nrcvs = count(rvsz /= 0)
if (debug_level >= psb_debug_outer_)&
& write(debug_unit,*) me,' ',trim(name),': Done initial alltoall',nsnds,nrcvs
idxs = 0
idxr = 0
@@ -224,10 +234,10 @@ Subroutine psb_dsphalo(a,desc_a,blk,info,rowcnv,colcnv,&
n_el_send = ipdxv(counter+psb_n_elem_send_)
bsdindx(proc+1) = idxs
idxs = idxs + sdsz(proc+1)
idxs = idxs + sdsz(proc+1)
brvindx(proc+1) = idxr
idxr = idxr + rvsz(proc+1)
counter = counter+n_el_send+3
idxr = idxr + rvsz(proc+1)
counter = counter+n_el_send+3
Enddo
iszr=sum(rvsz)
@@ -237,7 +247,7 @@ Subroutine psb_dsphalo(a,desc_a,blk,info,rowcnv,colcnv,&
& ' Send:',sdsz(:),' Receive:',rvsz(:)
mat_recv = iszr
iszs=sum(sdsz)
if (info == psb_success_) call psb_ensure_size(max(iszs,1),iasnd,info)
call psb_ensure_size(max(iszs,1),iasnd,info)
if (info == psb_success_) call psb_ensure_size(max(iszs,1),jasnd,info)
if (info == psb_success_) call psb_ensure_size(max(iszs,1),valsnd,info)
#if defined(IPK4) && defined(LPK8)
@@ -337,29 +347,33 @@ Subroutine psb_dsphalo(a,desc_a,blk,info,rowcnv,colcnv,&
ipx = 1
counter=1
idx = 0
tot_elem=0
!
! Make sure to get all columns in csget.
! This is necessary when sphalo is used to compute a transpose,
! as opposed to just gathering halo for spspmm purposes.
!
ncg = huge(ncg)
tot_elem = 0
Do
proc=ipdxv(counter)
proc = ipdxv(counter)
if (proc == -1) exit
n_el_recv=ipdxv(counter+psb_n_elem_recv_)
counter=counter+n_el_recv
n_el_send=ipdxv(counter+psb_n_elem_send_)
n_el_recv = ipdxv(counter+psb_n_elem_recv_)
counter = counter+n_el_recv
n_el_send = ipdxv(counter+psb_n_elem_send_)
Do j=0,n_el_send-1
idx = ipdxv(counter+psb_elem_send_+j)
n_elem = a%get_nz_row(idx)
call a%csget(idx,idx,ngtz,iasnd,jasnd,valsnd,info,&
& append=.true.,nzin=tot_elem)
& append=.true.,nzin=tot_elem,jmax=ncg)
if (info /= psb_success_) then
info=psb_err_from_subroutine_
ch_err='psb_sp_getrow'
call psb_errpush(info,name,a_err=ch_err)
call psb_errpush(info,name,a_err='psb_sp_getrow')
goto 9999
end if
tot_elem=tot_elem+n_elem
tot_elem = tot_elem+ngtz
Enddo
ipx = ipx + 1
ipx = ipx + 1
counter = counter+n_el_send+3
Enddo
nz = tot_elem
@@ -368,8 +382,7 @@ Subroutine psb_dsphalo(a,desc_a,blk,info,rowcnv,colcnv,&
if (colcnv_) call psb_loc_to_glob(jasnd(1:nz),desc_a,info,iact='I')
if (info /= psb_success_) then
info=psb_err_from_subroutine_
ch_err='psb_loc_to_glob'
call psb_errpush(info,name,a_err=ch_err)
call psb_errpush(info,name,a_err='psb_loc_to_glob')
goto 9999
end if
base/tools/psb_ssphalo.F90
+34 -21
View File
@@ -80,7 +80,7 @@ Subroutine psb_ssphalo(a,desc_a,blk,info,rowcnv,colcnv,&
& n_el_send,k,n_el_recv,idx, r, tot_elem,&
& n_elem, j, ipx,mat_recv, iszs, iszr,idxs,idxr,nz,&
& irmin,icmin,irmax,icmax,data_,ngtz,totxch,nxs, nxr,&
& l1, err_act
& l1, err_act, nsnds, nrcvs, nrg, ncg
integer(psb_mpk_) :: icomm, minfo
integer(psb_mpk_), allocatable :: brvindx(:), &
& rvsz(:), bsdindx(:),sdsz(:)
@@ -187,7 +187,7 @@ Subroutine psb_ssphalo(a,desc_a,blk,info,rowcnv,colcnv,&
call desc_a%get_list(data_,pdxv,totxch,nxr,nxs,info)
ipdxv = pdxv%get_vect()
! For all rows in the halo descriptor, extract and send/receive.
! For all rows in the halo descriptor, extract the row size
Do
proc=ipdxv(counter)
if (proc == -1) exit
@@ -196,8 +196,8 @@ Subroutine psb_ssphalo(a,desc_a,blk,info,rowcnv,colcnv,&
n_el_send = ipdxv(counter+psb_n_elem_send_)
tot_elem = 0
Do j=0,n_el_send-1
idx = ipdxv(counter+psb_elem_send_+j)
n_elem = a%get_nz_row(idx)
idx = ipdxv(counter+psb_elem_send_+j)
n_elem = a%get_nz_row(idx)
tot_elem = tot_elem+n_elem
Enddo
sdsz(proc+1) = tot_elem
@@ -205,6 +205,12 @@ Subroutine psb_ssphalo(a,desc_a,blk,info,rowcnv,colcnv,&
counter = counter+n_el_send+3
Enddo
!
! Exchange row sizes, so as to know sends/receives.
! This is different from the halo exchange because the
! size of the rows may vary, as opposed to fixed
! (multi) vector row size.
!
call mpi_alltoall(sdsz,1,psb_mpi_mpk_,&
& rvsz,1,psb_mpi_mpk_,icomm,minfo)
if (info /= psb_success_) then
@@ -212,6 +218,10 @@ Subroutine psb_ssphalo(a,desc_a,blk,info,rowcnv,colcnv,&
call psb_errpush(info,name,a_err='mpi_alltoall')
goto 9999
end if
nsnds = count(sdsz /= 0)
nrcvs = count(rvsz /= 0)
if (debug_level >= psb_debug_outer_)&
& write(debug_unit,*) me,' ',trim(name),': Done initial alltoall',nsnds,nrcvs
idxs = 0
idxr = 0
@@ -224,10 +234,10 @@ Subroutine psb_ssphalo(a,desc_a,blk,info,rowcnv,colcnv,&
n_el_send = ipdxv(counter+psb_n_elem_send_)
bsdindx(proc+1) = idxs
idxs = idxs + sdsz(proc+1)
idxs = idxs + sdsz(proc+1)
brvindx(proc+1) = idxr
idxr = idxr + rvsz(proc+1)
counter = counter+n_el_send+3
idxr = idxr + rvsz(proc+1)
counter = counter+n_el_send+3
Enddo
iszr=sum(rvsz)
@@ -237,7 +247,7 @@ Subroutine psb_ssphalo(a,desc_a,blk,info,rowcnv,colcnv,&
& ' Send:',sdsz(:),' Receive:',rvsz(:)
mat_recv = iszr
iszs=sum(sdsz)
if (info == psb_success_) call psb_ensure_size(max(iszs,1),iasnd,info)
call psb_ensure_size(max(iszs,1),iasnd,info)
if (info == psb_success_) call psb_ensure_size(max(iszs,1),jasnd,info)
if (info == psb_success_) call psb_ensure_size(max(iszs,1),valsnd,info)
#if defined(IPK4) && defined(LPK8)
@@ -337,29 +347,33 @@ Subroutine psb_ssphalo(a,desc_a,blk,info,rowcnv,colcnv,&
ipx = 1
counter=1
idx = 0
tot_elem=0
!
! Make sure to get all columns in csget.
! This is necessary when sphalo is used to compute a transpose,
! as opposed to just gathering halo for spspmm purposes.
!
ncg = huge(ncg)
tot_elem = 0
Do
proc=ipdxv(counter)
proc = ipdxv(counter)
if (proc == -1) exit
n_el_recv=ipdxv(counter+psb_n_elem_recv_)
counter=counter+n_el_recv
n_el_send=ipdxv(counter+psb_n_elem_send_)
n_el_recv = ipdxv(counter+psb_n_elem_recv_)
counter = counter+n_el_recv
n_el_send = ipdxv(counter+psb_n_elem_send_)
Do j=0,n_el_send-1
idx = ipdxv(counter+psb_elem_send_+j)
n_elem = a%get_nz_row(idx)
call a%csget(idx,idx,ngtz,iasnd,jasnd,valsnd,info,&
& append=.true.,nzin=tot_elem)
& append=.true.,nzin=tot_elem,jmax=ncg)
if (info /= psb_success_) then
info=psb_err_from_subroutine_
ch_err='psb_sp_getrow'
call psb_errpush(info,name,a_err=ch_err)
call psb_errpush(info,name,a_err='psb_sp_getrow')
goto 9999
end if
tot_elem=tot_elem+n_elem
tot_elem = tot_elem+ngtz
Enddo
ipx = ipx + 1
ipx = ipx + 1
counter = counter+n_el_send+3
Enddo
nz = tot_elem
@@ -368,8 +382,7 @@ Subroutine psb_ssphalo(a,desc_a,blk,info,rowcnv,colcnv,&
if (colcnv_) call psb_loc_to_glob(jasnd(1:nz),desc_a,info,iact='I')
if (info /= psb_success_) then
info=psb_err_from_subroutine_
ch_err='psb_loc_to_glob'
call psb_errpush(info,name,a_err=ch_err)
call psb_errpush(info,name,a_err='psb_loc_to_glob')
goto 9999
end if
base/tools/psb_zsphalo.F90
+34 -21
View File
@@ -80,7 +80,7 @@ Subroutine psb_zsphalo(a,desc_a,blk,info,rowcnv,colcnv,&
& n_el_send,k,n_el_recv,idx, r, tot_elem,&
& n_elem, j, ipx,mat_recv, iszs, iszr,idxs,idxr,nz,&
& irmin,icmin,irmax,icmax,data_,ngtz,totxch,nxs, nxr,&
& l1, err_act
& l1, err_act, nsnds, nrcvs, nrg, ncg
integer(psb_mpk_) :: icomm, minfo
integer(psb_mpk_), allocatable :: brvindx(:), &
& rvsz(:), bsdindx(:),sdsz(:)
@@ -187,7 +187,7 @@ Subroutine psb_zsphalo(a,desc_a,blk,info,rowcnv,colcnv,&
call desc_a%get_list(data_,pdxv,totxch,nxr,nxs,info)
ipdxv = pdxv%get_vect()
! For all rows in the halo descriptor, extract and send/receive.
! For all rows in the halo descriptor, extract the row size
Do
proc=ipdxv(counter)
if (proc == -1) exit
@@ -196,8 +196,8 @@ Subroutine psb_zsphalo(a,desc_a,blk,info,rowcnv,colcnv,&
n_el_send = ipdxv(counter+psb_n_elem_send_)
tot_elem = 0
Do j=0,n_el_send-1
idx = ipdxv(counter+psb_elem_send_+j)
n_elem = a%get_nz_row(idx)
idx = ipdxv(counter+psb_elem_send_+j)
n_elem = a%get_nz_row(idx)
tot_elem = tot_elem+n_elem
Enddo
sdsz(proc+1) = tot_elem
@@ -205,6 +205,12 @@ Subroutine psb_zsphalo(a,desc_a,blk,info,rowcnv,colcnv,&
counter = counter+n_el_send+3
Enddo
!
! Exchange row sizes, so as to know sends/receives.
! This is different from the halo exchange because the
! size of the rows may vary, as opposed to fixed
! (multi) vector row size.
!
call mpi_alltoall(sdsz,1,psb_mpi_mpk_,&
& rvsz,1,psb_mpi_mpk_,icomm,minfo)
if (info /= psb_success_) then
@@ -212,6 +218,10 @@ Subroutine psb_zsphalo(a,desc_a,blk,info,rowcnv,colcnv,&
call psb_errpush(info,name,a_err='mpi_alltoall')
goto 9999
end if
nsnds = count(sdsz /= 0)
nrcvs = count(rvsz /= 0)
if (debug_level >= psb_debug_outer_)&
& write(debug_unit,*) me,' ',trim(name),': Done initial alltoall',nsnds,nrcvs
idxs = 0
idxr = 0
@@ -224,10 +234,10 @@ Subroutine psb_zsphalo(a,desc_a,blk,info,rowcnv,colcnv,&
n_el_send = ipdxv(counter+psb_n_elem_send_)
bsdindx(proc+1) = idxs
idxs = idxs + sdsz(proc+1)
idxs = idxs + sdsz(proc+1)
brvindx(proc+1) = idxr
idxr = idxr + rvsz(proc+1)
counter = counter+n_el_send+3
idxr = idxr + rvsz(proc+1)
counter = counter+n_el_send+3
Enddo
iszr=sum(rvsz)
@@ -237,7 +247,7 @@ Subroutine psb_zsphalo(a,desc_a,blk,info,rowcnv,colcnv,&
& ' Send:',sdsz(:),' Receive:',rvsz(:)
mat_recv = iszr
iszs=sum(sdsz)
if (info == psb_success_) call psb_ensure_size(max(iszs,1),iasnd,info)
call psb_ensure_size(max(iszs,1),iasnd,info)
if (info == psb_success_) call psb_ensure_size(max(iszs,1),jasnd,info)
if (info == psb_success_) call psb_ensure_size(max(iszs,1),valsnd,info)
#if defined(IPK4) && defined(LPK8)
@@ -337,29 +347,33 @@ Subroutine psb_zsphalo(a,desc_a,blk,info,rowcnv,colcnv,&
ipx = 1
counter=1
idx = 0
tot_elem=0
!
! Make sure to get all columns in csget.
! This is necessary when sphalo is used to compute a transpose,
! as opposed to just gathering halo for spspmm purposes.
!
ncg = huge(ncg)
tot_elem = 0
Do
proc=ipdxv(counter)
proc = ipdxv(counter)
if (proc == -1) exit
n_el_recv=ipdxv(counter+psb_n_elem_recv_)
counter=counter+n_el_recv
n_el_send=ipdxv(counter+psb_n_elem_send_)
n_el_recv = ipdxv(counter+psb_n_elem_recv_)
counter = counter+n_el_recv
n_el_send = ipdxv(counter+psb_n_elem_send_)
Do j=0,n_el_send-1
idx = ipdxv(counter+psb_elem_send_+j)
n_elem = a%get_nz_row(idx)
call a%csget(idx,idx,ngtz,iasnd,jasnd,valsnd,info,&
& append=.true.,nzin=tot_elem)
& append=.true.,nzin=tot_elem,jmax=ncg)
if (info /= psb_success_) then
info=psb_err_from_subroutine_
ch_err='psb_sp_getrow'
call psb_errpush(info,name,a_err=ch_err)
call psb_errpush(info,name,a_err='psb_sp_getrow')
goto 9999
end if
tot_elem=tot_elem+n_elem
tot_elem = tot_elem+ngtz
Enddo
ipx = ipx + 1
ipx = ipx + 1
counter = counter+n_el_send+3
Enddo
nz = tot_elem
@@ -368,8 +382,7 @@ Subroutine psb_zsphalo(a,desc_a,blk,info,rowcnv,colcnv,&
if (colcnv_) call psb_loc_to_glob(jasnd(1:nz),desc_a,info,iact='I')
if (info /= psb_success_) then
info=psb_err_from_subroutine_
ch_err='psb_loc_to_glob'
call psb_errpush(info,name,a_err=ch_err)
call psb_errpush(info,name,a_err='psb_loc_to_glob')
goto 9999
end if