rpm/mumps
SHA256
1
0
Fork 0
forked from pool/mumps
Code Pull Requests Activity

Accepting request 875060 from home:mslacken:branches:science

Browse Source 
- updated to version 5.3.5, changes are
  * Fixed 2x2 pivots bug from 5.3.4 release in MPI LDLT factorization
  * Fixed ICNTL(8)=-2 option during analysis (code and documentation)
  * Fixed a rare bug (segfault) related to dynamic storage management on numerically difficult matrices
  * Fixed a rare deadlock in BLR for symmetric matrices
  * Fixed an uninitialized variable (which could lead to incorrect -19 error)
  * Minor fix in userguide (CNTL(1) vs. ICNTL(1) in ICNTL(36) description)
  * Fixed a possible runtime issue during solve, related to "TO_PROCESS" array 
  * Assume ilp64 MPI interface only applies to Fortran in c_example.c
  * Note on gfortran-10 compilation added
  * Avoid intent on pointers (F2003-only)
  * More robust multithreading for matrix reformatting (arrowheads)
  * Fixed ICNTL(31) interpretation in case of repeated analysis
  * Fixed multiple mpif.h inclusion (distributed rhs, ifort+openmpi)
  * Fixed computation of effectively used memory statistics
  * Improved multithreaded performance of BLR backward solve
  * Fixed return code in build_mumps_int_def.c + openmp compilation (pgi)
  * Forbid a loop vectorization in [sdcz]sol_c.F (segfault with ifort) 
  * New feature: distributed right-hand sides
  * Improved time for arrowheads construction (single MPI case, mainly)
  * C interface: ability to know if MUMPS_INT is 64-bit from include file
  * Improved BLR performance when CNTL(1)=0.0 and ICNTL(36)=1
  * Fixed INFO(34),INFO(35),INFO(37),INFO(38) on processes with rank > 0
  * More portable MPI_IS_IN_PLACE feature in libseq
  * Fixed determinant computation when Cholesky ScaLapack is used
  * Information on advancement (flops done) on each MPI process
  * Allow rhs_sparse and irhs_sparse to be unassociated if nz_rhs=0
  * Fixed INFO(30) and INFO(31) computation on MPI processes with rank > 0
  * OMP collapsed loops: avoid FIRSTPRIVATE on internal loop bound (for pgi)
  * Fix for compilers not freeing local allocatable arrays (64-bit metis)

OBS-URL: https://build.opensuse.org/request/show/875060
OBS-URL: https://build.opensuse.org/package/show/science/mumps?expand=0&rev=55
This commit is contained in:
Egbert Eich 2021-02-26 14:18:21 +00:00 committed by Git OBS Bridge
parent f0ac7b5c85
commit f426b846d4
5 changed files with 55 additions and 106 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

92
Makefiles-Serialize-libseq-libplat-mommond_mod-for-parallel-builds.patch
View File

@ -1,92 +0,0 @@
From: Egbert Eich <eich@suse.com>
Date: Sun Feb 10 17:16:51 2019 +0100
Subject: Makefiles: Serialize libseq, libplat, mommond_mod for parallel builds
Patch-mainline: Not yet
Git-commit: d8e18ae193f2fd67761c8226a3ad9bcf80d4230c
References:
Signed-off-by: Egbert Eich <eich@suse.com>
---
Makefile | 17 ++++++++++-------
src/Makefile | 10 ++++++----
2 files changed, 16 insertions(+), 11 deletions(-)
diff --git a/Makefile b/Makefile
index 4636e85..489753b 100644
--- a/Makefile
+++ b/Makefile
@@ -9,18 +9,18 @@ default: dexamples
.PHONY: default alllib all c z s d \
sexamples dexamples cexamples zexamples multi_example \
- mumps_lib requiredobj libseqneeded clean
+ mumps_lib common requiredobj libseqneeded clean
alllib: c z s d
all: cexamples zexamples sexamples dexamples multi_example
-c:
+c: requiredobj
$(MAKE) ARITH=c mumps_lib
-z:
+z: requiredobj
$(MAKE) ARITH=z mumps_lib
-s:
+s: requiredobj
$(MAKE) ARITH=s mumps_lib
-d:
+d: requiredobj
$(MAKE) ARITH=d mumps_lib
@@ -36,7 +36,10 @@ Makefile.inc:
include Makefile.inc
-mumps_lib: requiredobj
+common:
+ (cd src ; $(MAKE) common)
+
+mumps_lib:
(cd src ; $(MAKE) $(ARITH))
cexamples: c
@@ -54,7 +57,7 @@ dexamples: d
multi_example: s d c z
(cd examples ; $(MAKE) multi)
-requiredobj: Makefile.inc $(LIBSEQNEEDED) $(libdir)/libpord$(PLAT)$(LIBEXT)
+requiredobj: Makefile.inc $(LIBSEQNEEDED) $(libdir)/libpord$(PLAT)$(LIBEXT) common
# dummy MPI library (sequential version)
diff --git a/src/Makefile b/src/Makefile
index 88fc032..5a3c6e2 100644
--- a/src/Makefile
+++ b/src/Makefile
@@ -12,13 +12,13 @@ default: d
.PHONY: default s d c z mumps_lib clean
-s:
+s: common
$(MAKE) ARITH=s mumps_lib
-d:
+d: common
$(MAKE) ARITH=d mumps_lib
-c:
+c: common
$(MAKE) ARITH=c mumps_lib
-z:
+z: common
$(MAKE) ARITH=z mumps_lib
include $(topdir)/Makefile.inc
@@ -78,6 +78,8 @@
tools_common.o \
sol_common.o
+common: $(OBJS_COMMON_MOD) $(OBJS_COMMON_OTHER)
+
OBJS_MOD = \
$(ARITH)ana_aux.o \
$(ARITH)ana_aux_par.o \

3
mumps-5.2.1.tar.gz
View File

@ -1,3 +0,0 @@
version https://git-lfs.github.com/spec/v1
oid sha256:d988fc34dfc8f5eee0533e361052a972aa69cc39ab193e7f987178d24981744a
size 3831942

3
mumps-5.3.5.tar.gz Normal file
View File

@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:e5d665fdb7043043f0799ae3dbe3b37e5b200d1ab7a6f7b2a4e463fd89507fa4
size 3940501

42
mumps.changes
View File

@ -1,3 +1,45 @@
-------------------------------------------------------------------
Thu Feb 25 09:48:50 UTC 2021 - Christian Goll <cgoll@suse.com>
- updated to version 5.3.5, changes are
* Fixed 2x2 pivots bug from 5.3.4 release in MPI LDLT factorization
* Fixed ICNTL(8)=-2 option during analysis (code and documentation)
* Fixed a rare bug (segfault) related to dynamic storage management on numerically difficult matrices
* Fixed a rare deadlock in BLR for symmetric matrices
* Fixed an uninitialized variable (which could lead to incorrect -19 error)
* Minor fix in userguide (CNTL(1) vs. ICNTL(1) in ICNTL(36) description)
* Fixed a possible runtime issue during solve, related to "TO_PROCESS" array
* Assume ilp64 MPI interface only applies to Fortran in c_example.c
* Note on gfortran-10 compilation added
* Avoid intent on pointers (F2003-only)
* More robust multithreading for matrix reformatting (arrowheads)
* Fixed ICNTL(31) interpretation in case of repeated analysis
* Fixed multiple mpif.h inclusion (distributed rhs, ifort+openmpi)
* Fixed computation of effectively used memory statistics
* Improved multithreaded performance of BLR backward solve
* Fixed return code in build_mumps_int_def.c + openmp compilation (pgi)
* Forbid a loop vectorization in [sdcz]sol_c.F (segfault with ifort)
* New feature: distributed right-hand sides
* Improved time for arrowheads construction (single MPI case, mainly)
* C interface: ability to know if MUMPS_INT is 64-bit from include file
* Improved BLR performance when CNTL(1)=0.0 and ICNTL(36)=1
* Fixed INFO(34),INFO(35),INFO(37),INFO(38) on processes with rank > 0
* More portable MPI_IS_IN_PLACE feature in libseq
* Fixed determinant computation when Cholesky ScaLapack is used
* Information on advancement (flops done) on each MPI process
* Allow rhs_sparse and irhs_sparse to be unassociated if nz_rhs=0
* Fixed INFO(30) and INFO(31) computation on MPI processes with rank > 0
* OMP collapsed loops: avoid FIRSTPRIVATE on internal loop bound (for pgi)
* Fix for compilers not freeing local allocatable arrays (64-bit metis)
* Fixed RINFO(5-6) and RINFOG(15-16) metrics (entries=>bytes)
* C interface: A_ELT/SCHUR/RHS/REDRHS/RHS_loc/SOL_loc may exceed 2^31 entries
* Local Schur (ICNTL(19)=2 or 3) may now exceed 2^31 entries
* Fixed internal dynamic storage of blocks with more than 2^31 entries
* Fixed a bug in the parallel analysis that limited scalability
- removed Makefiles-Serialize-libseq-libplat-mommond_mod-for-parallel-builds.patch
* functionaltiy is included upstream
-------------------------------------------------------------------
Fri Feb 5 20:31:13 UTC 2021 - Egbert Eich <eich@suse.com>

21
mumps.spec
View File

@ -19,7 +19,7 @@
%global flavor @BUILD_FLAVOR@%{nil}
%define pname mumps
%define ver 5.2.1
%define ver 5.3.5
%define so_ver 5
%define openblas_vers 0.3.6
%global _lto_cflags %{_lto_cflags} -ffat-lto-objects
@ -181,7 +181,7 @@ ExclusiveArch: do_not_build
# macro mpi is used by macros for master package
%global mpi_family mvapich2
%define mumps_f77_mpilibs -lfmpich -lmpich
%undefine mpi_ver
%undefine mpi_ver
%bcond_without hpc
%endif
@ -236,7 +236,7 @@ ExclusiveArch: do_not_build
# macro mpi is used by macros for master package
%global mpi_family mvapich2
%define mumps_f77_mpilibs -lfmpich -lmpich
%undefine mpi_ver
%undefine mpi_ver
%bcond_without hpc
%endif
@ -291,7 +291,7 @@ ExclusiveArch: do_not_build
# macro mpi is used by macros for master package
%global mpi_family mvapich2
%define mumps_f77_mpilibs -lfmpich -lmpich
%undefine mpi_ver
%undefine mpi_ver
%bcond_without hpc
%endif
@ -346,7 +346,7 @@ ExclusiveArch: do_not_build
# macro mpi is used by macros for master package
%global mpi_family mvapich2
%define mumps_f77_mpilibs -lfmpich -lmpich
%undefine mpi_ver
%undefine mpi_ver
%bcond_without hpc
%endif
@ -401,7 +401,7 @@ ExclusiveArch: do_not_build
# macro mpi is used by macros for master package
%global mpi_family mvapich2
%define mumps_f77_mpilibs -lfmpich -lmpich
%undefine mpi_ver
%undefine mpi_ver
%bcond_without hpc
%endif
@ -421,7 +421,7 @@ ExclusiveArch: do_not_build
%{?mpi_family:%{bcond_without mpi}}%{!?mpi_family:%{bcond_with mpi}}
# openmpi 1 was called just "openmpi" in Leap 15.x/SLE15
# openmpi 1 was called just "openmpi" in Leap 15.x/SLE15
%if 0%{?suse_version} >= 1550 || "%{mpi_family}" != "openmpi" || "%{mpi_ver}" != "1"
%define mpi_ext %{?mpi_ver}
%endif
@ -474,7 +474,6 @@ Release: 0
URL: http://mumps.enseeiht.fr/
Source0: http://mumps.enseeiht.fr/MUMPS_%{version}.tar.gz#/%{pname}-%{version}.tar.gz
Source1: Makefile.inc
Patch1: Makefiles-Serialize-libseq-libplat-mommond_mod-for-parallel-builds.patch
%if %{without hpc}
BuildRequires: gcc-fortran
%{?with_scotch:BuildRequires: %{scotch}%{?with_mpi:-%{mpi_family}%{?mpi_ext}}-devel}
@ -500,7 +499,6 @@ particular focus on symmetric positive definite matrices. It can
operate on distributed matrices e.g. over a cluster. It has Fortran and
C interfaces, and can interface with ordering tools such as Scotch.
%package -n %{libname}
Summary: A MUltifrontal Massively Parallel Sparse direct Solver
Group: System/Libraries
@ -657,7 +655,7 @@ matlab and scilab extensions are provided in /usr/share/doc/packages/mumps.
%prep
%setup -q -n %{PNAME}_%{version}
%patch1 -p1
#%%patch1 -p1
%build
@ -739,7 +737,7 @@ make %{?_smp_mflags} alllib
%{!?with_mpi:cp -P libseq/libmpiseq*.a lib/}
mkdir lib/tmp; cd lib/tmp;
mkdir lib/tmp; cd lib/tmp;
%define LORDERINGS -lpord%{?PLAT} %{?scotch:-l%{scotch}}
%if %{without mpi}
rm -f *.o; ar -x ../libmpiseq%{?PLAT}.a
@ -850,6 +848,7 @@ rm -rf examples/*.o examples/*simpletest examples/*_save_restore examples/c_exam
%post -n %{libname} -p /sbin/ldconfig
%postun -n %{libname} -p /sbin/ldconfig
%else
%post -n %{libname}
/sbin/ldconfig -N %{my_libdir}