! Parallel Sparse BLAS GPU plugin ! (C) Copyright 2013 ! ! Salvatore Filippone ! Alessandro Fanfarillo ! ! Redistribution and use in source and binary forms, with or without ! modification, are permitted provided that the following conditions ! are met: ! 1. Redistributions of source code must retain the above copyright ! notice, this list of conditions and the following disclaimer. ! 2. Redistributions in binary form must reproduce the above copyright ! notice, this list of conditions, and the following disclaimer in the ! documentation and/or other materials provided with the distribution. ! 3. The name of the PSBLAS group or the names of its contributors may ! not be used to endorse or promote products derived from this ! software without specific written permission. ! ! THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS ! ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED ! TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR ! PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE PSBLAS GROUP OR ITS CONTRIBUTORS ! BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR ! CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF ! SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS ! INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN ! CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ! ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE ! POSSIBILITY OF SUCH DAMAGE. ! subroutine psb_s_cuda_cp_elg_from_coo(a,b,info) use psb_base_mod #ifdef HAVE_SPGPU use elldev_mod use psb_vectordev_mod use psb_s_cuda_elg_mat_mod, psb_protect_name => psb_s_cuda_cp_elg_from_coo use psi_ext_util_mod use psb_cuda_env_mod #else use psb_s_cuda_elg_mat_mod #endif implicit none class(psb_s_cuda_elg_sparse_mat), intent(inout) :: a class(psb_s_coo_sparse_mat), intent(in) :: b integer(psb_ipk_), intent(out) :: info !locals Integer(Psb_ipk_) :: nza, nr, i,j,k, idl,err_act, nc, nzm, & & ir, ic, ld, ldv, hacksize integer(psb_ipk_) :: debug_level, debug_unit character(len=20) :: name type(psb_s_coo_sparse_mat) :: tmp integer(psb_ipk_), allocatable :: idisp(:) info = psb_success_ #ifdef HAVE_SPGPU hacksize = max(1,psb_cuda_WarpSize()) #else hacksize = 1 #endif if (b%is_dev()) call b%sync() if (b%is_by_rows()) then #ifdef HAVE_SPGPU call psi_s_count_ell_from_coo(a,b,idisp,ldv,nzm,info,hacksize=hacksize) if (c_associated(a%deviceMat)) then call freeEllDevice(a%deviceMat) endif nr = b%get_nrows() nc = b%get_ncols() nza = b%get_nzeros() info = FallocEllDevice(a%deviceMat,nr,nzm,nza,nc,spgpu_type_double,1) if (info == 0) info = psi_CopyCooToElg(nr,nc,nza, hacksize,ldv,nzm, & & a%irn,idisp,b%ja,b%val, a%deviceMat) call a%set_dev() #else call psi_s_convert_ell_from_coo(a,b,info,hacksize=hacksize) call a%set_host() #endif else call b%cp_to_coo(tmp,info) #ifdef HAVE_SPGPU call psi_s_count_ell_from_coo(a,tmp,idisp,ldv,nzm,info,hacksize=hacksize) if (c_associated(a%deviceMat)) then call freeEllDevice(a%deviceMat) endif nr = b%get_nrows() nc = b%get_ncols() nza = b%get_nzeros() info = FallocEllDevice(a%deviceMat,nr,nzm,nza,nc,spgpu_type_double,1) if (info == 0) info = psi_CopyCooToElg(nr,nc,nza, hacksize,ldv,nzm, & & a%irn,idisp,tmp%ja,tmp%val, a%deviceMat) call a%set_dev() #else call psi_s_convert_ell_from_coo(a,tmp,info,hacksize=hacksize) call a%set_host() #endif end if if (info /= psb_success_) goto 9999 return 9999 continue info = psb_err_alloc_dealloc_ return contains subroutine psi_s_count_ell_from_coo(a,b,idisp,ldv,nzm,info,hacksize) use psb_base_mod use psi_ext_util_mod implicit none class(psb_s_ell_sparse_mat), intent(inout) :: a class(psb_s_coo_sparse_mat), intent(in) :: b integer(psb_ipk_), allocatable, intent(out) :: idisp(:) integer(psb_ipk_), intent(out) :: info, nzm, ldv integer(psb_ipk_), intent(in), optional :: hacksize !locals Integer(Psb_ipk_) :: nza, nr, i,j,k, idl,err_act, nc, & & ir, ic, hsz_ real(psb_dpk_) :: t0,t1 logical, parameter :: timing=.true. info = psb_success_ nr = b%get_nrows() nc = b%get_ncols() nza = b%get_nzeros() hsz_ = 1 if (present(hacksize)) then if (hacksize> 1) hsz_ = hacksize end if ! Make ldv a multiple of hacksize ldv = ((nr+hsz_-1)/hsz_)*hsz_ ! If it is sorted then we can lessen memory impact a%psb_s_base_sparse_mat = b%psb_s_base_sparse_mat ! First compute the number of nonzeros in each row. call psb_realloc(nr,a%irn,info) if (info == psb_success_) call psb_realloc(nr+1,idisp,info) if (info /= psb_success_) return if (timing) t0=psb_wtime() a%irn = 0 do i=1, nza ir = b%ia(i) a%irn(ir) = a%irn(ir) + 1 end do nzm = 0 a%nzt = 0 idisp(1) = 0 do i=1,nr nzm = max(nzm,a%irn(i)) a%nzt = a%nzt + a%irn(i) idisp(i+1) = a%nzt end do end subroutine psi_s_count_ell_from_coo end subroutine psb_s_cuda_cp_elg_from_coo