base/modules/psb_error_mod.F90
base/modules/psb_psblas_mod.f90
base/newserial/psbn_base_mat_mod.f03
base/newserial/psbn_d_base_mat_mod.f03
base/newserial/psbn_d_coo_impl.f03
base/newserial/psbn_d_csr_impl.f03
base/newserial/psbn_d_csr_mat_mod.f03
base/newserial/psbn_mat_mod.f03
base/psblas/psb_dnrmi.f90
base/psblas/psb_dspmm.f90
base/psblas/psb_dspsm.f90
base/tools/psb_dspalloc.f90
prec/psb_dbjac_aply.f90
prec/psb_dbjac_bld.f90
prec/psb_dilu_fct.f90
prec/psb_dprecbld.f90
prec/psb_dprecinit.f90
prec/psb_prec_mod.f90
prec/psb_prec_type.f90
test/pargen/ppde.f90
test/pargen/runs/ppde.inp
Now both BJAC_BLD and CSSV work. Really!
And initial performance is not too bad.
Lots and lots of details to be fixed yet...........
base/newserial/psbn_d_base_mat_mod.f03
base/newserial/psbn_d_csr_mat_mod.f03
prec/psb_ddiagsc_bld.f90
prec/psb_dprecbld.f90
prec/psb_dprecinit.f90
prec/psb_prec_mod.f90
test/pargen/runs/ppde.inp
test/serial/psbn_d_cxx_impl.f03
test/serial/psbn_d_cxx_mat_mod.f03
util/psb_mat_dist_mod.f90
Moved csgetblk and csclip to the base level, the only specific
implementation is that of csgetrow.
base/Makefile
base/modules/Makefile
base/modules/psb_psblas_mod.f90
base/modules/psb_tools_mod.f90
base/newserial/Makefile
base/newserial/README
base/newserial/psbn_base_mat_mod.f03
base/newserial/psbn_d_base_mat_mod.f03
base/newserial/psbn_d_coo_impl.f03
base/newserial/psbn_d_csr_impl.f03
base/newserial/psbn_d_csr_mat_mod.f03
base/newserial/psbn_mat_mod.f03
base/psblas/psb_dnrmi.f90
base/psblas/psb_dspmm.f90
base/tools/psb_dspalloc.f90
base/tools/psb_dspasb.f90
base/tools/psb_dspfree.f90
base/tools/psb_dspins.f90
krylov/psb_dbicg.f90
krylov/psb_dcg.F90
krylov/psb_dcgs.f90
krylov/psb_dcgstab.F90
krylov/psb_dcgstabl.f90
krylov/psb_drgmres.f90
krylov/psb_krylov_mod.f90
prec/psb_dprecbld.f90
prec/psb_dprecinit.f90
prec/psb_dprecset.f90
prec/psb_prec_mod.f90
test/pargen/ppde.f90
test/pargen/runs/ppde.inp
test/serial/d_coo_matgen.f03
test/serial/d_matgen.f03
test/serial/psbn_d_cxx_mat_mod.f03
Minimal set of changes to run BICGSTAB in serial mode with NONE as a
preconditioner. It does now!
base/newserial/psbn_base_mat_mod.f03
base/newserial/psbn_d_base_mat_mod.f03
base/newserial/psbn_d_coo_impl.f03
base/newserial/psbn_d_csr_impl.f03
base/newserial/psbn_d_csr_mat_mod.f03
base/newserial/psbn_mat_mod.f03
test/serial/d_matgen.f03
Added SCAL.
Defined CSGET/CSCLIP: they work on the same inner kernel
implementation!
test/serial/Makefile
test/serial/d_matgen.f03
test/serial/psbn_d_cxx_impl.f03
test/serial/psbn_d_cxx_mat_mod.f03
First successful test of user-defined storage formats.
base/modules/psb_error_mod.F90
base/newserial/psbn_base_mat_mod.f03
base/newserial/psbn_d_base_mat_mod.f03
base/newserial/psbn_d_csr_impl.f03
base/newserial/psbn_d_csr_mat_mod.f03
base/newserial/psbn_mat_impl.f03
base/newserial/psbn_mat_mod.f03
test/serial/Makefile
test/serial/d_matgen.f03
First round of encapsulation of sparse matrix data types.
Makefile
psbn_base_mat_mod.f03
psbn_d_base_mat_mod.f03
psbn_mat_mod.f03
Implementd many basic methods.
Reviewed visibility, make names of subroutines/functions always
private (thus only accessible via the object).
Changelog
base/modules/Makefile
base/modules/psb_ip_reord_mod.f90
base/newserial/Makefile
base/newserial/psbn_base_mat_mod.f03
base/newserial/psbn_coo_mat.f03
base/newserial/psbn_d_base_mat_mod.f03
base/newserial/psbn_d_coo_impl.f03
base/newserial/psbn_mat_mod.f03
base/serial/Makefile
base/serial/psb_ip_reord_mod.f90
config/pac.m4
Reworked base_mat to include COO (everybody must know about COO to
define A%TO_COO() and A%FROM_COO)
base/Makefile
base/comm/psb_chalo.f90
base/comm/psb_dhalo.f90
base/comm/psb_shalo.f90
base/comm/psb_zhalo.f90
base/modules/Makefile
base/newserial/psbn_coo_mat.f03
base/newserial/psbn_d_base_mat_mod.f03
base/psblas/psb_casum.f90
base/psblas/psb_caxpby.f90
base/psblas/psb_cdot.f90
base/psblas/psb_cnrm2.f90
base/psblas/psb_dasum.f90
base/psblas/psb_daxpby.f90
base/psblas/psb_ddot.f90
base/psblas/psb_dnrm2.f90
base/psblas/psb_sasum.f90
base/psblas/psb_saxpby.f90
base/psblas/psb_sdot.f90
base/psblas/psb_snrm2.f90
base/psblas/psb_zasum.f90
base/psblas/psb_zaxpby.f90
base/psblas/psb_zdot.f90
base/psblas/psb_znrm2.f90
base/serial/dp/Makefile
base/serial/f77/ccsmm.f
base/serial/f77/scsmm.f
configure.ac
configure
Fixes for the handling of interface mismatch in the NAG compiler: use
-wmismatch=mpi_gather.... and save mismatch_all just for psb_penv and
psb_blacs.
This enabled discovery of a number of minor (and not so minor)
errors..
Still waiting to hear about the bugs in F2003 mode.
base/serial/Makefile
base/serial/aux/camsr.f90
base/serial/aux/camsrx.f90
base/serial/aux/dmsr.f90
base/serial/aux/dmsrx.f90
base/serial/aux/imsr.f90
base/serial/aux/imsrx.f90
base/serial/aux/smsr.f90
base/serial/aux/smsrx.f90
base/serial/aux/zamsr.f90
base/serial/aux/zamsrx.f90
base/serial/dp/Makefile
base/serial/dp/ccoco.f
base/serial/dp/ccocr.f
base/serial/dp/dcoco.f
base/serial/dp/dcocr.f
base/serial/dp/djdcox.f
base/serial/dp/reordvn.f
base/serial/dp/scoco.f
base/serial/dp/scocr.f
base/serial/dp/zcoco.f
base/serial/dp/zcocr.f
base/serial/psb_cfixcoo.f90
base/serial/psb_dfixcoo.f90
base/serial/psb_ip_reord_mod.f90
base/serial/psb_sfixcoo.f90
base/serial/psb_zfixcoo.f90
Defined a new module ip_reord to handle reordering based on a
list-sort output; this will avoid future type mismatch, such as the
one causing trouble in the final mld2p4 testing for complex single
precision.
Salvatore Filippone