#5951 closed defect (fixed)
OMC build time has increased
| Reported by: | Niklas Worschech | Owned by: | Martin Sjölund |
|---|---|---|---|
| Priority: | high | Milestone: | 1.16.0 |
| Component: | Build Environment | Version: | v1.16.0-dev |
| Keywords: | build time | Cc: | ruediger.franke@…, Andreas Heuermann |
Description
The omc build time has increased compared to the OpenModelica version 1.15
on linux the build time is for the 1.15 version
Complete build:
real 7m52.960s
user 68m50.538s
sys 8m19.201s
Rebuilt:
real 0m14.066s
user 0m23.606s
sys 0m5.898s
Wth the version e3d417519d
Complete
real 10m0.914s
user 80m55.100s
sys 9m16.211s
build
Rebuilt:
real 3m7.013s
user 22m0.388s
sys 1m7.458s
Attachments (1)
Change History (11)
by , 5 years ago
| Attachment: | omc_build_linux.log added |
|---|
comment:1 by , 5 years ago
comment:2 by , 5 years ago
| Cc: | added; removed |
|---|
comment:3 by , 5 years ago
The same output for the maintenance/v1.14 branch, i.e. rebuild without any change and grep 100% in the output of make:
[100%] Built target OMSIBase_static [100%] Built target OMSIBase [100%] Built target OMSIBase_static [100%] Built target OMSIBase [100%] Built target OMSIC_static [100%] Built target OMSIC [100%] Built target OMSIC_static [100%] Built target OMSIC
Still linear or already exponential infection rate towards 1.15 and master?
comment:4 by , 5 years ago
Started looking into this via:
https://github.com/OpenModelica/OpenModelica/pull/863
will continue with the OMSICpp.
comment:5 by , 5 years ago
Output for make -f Makefile.omdev.mingw runtimeCPPinstall -j3 via grep "100%"
[100%] Building C object CMakeFiles/umfpack_olib_DLONG_WSOLVE.dir/UMFPACK/Source/umfpack_solve.c.obj [100%] Built target umfpack_olib_DLONG_WSOLVE [100%] Linking C static library libbtf.a [100%] Built target btf [100%] Linking C static library libamd.a [100%] Built target amd [100%] Linking C static library libklu.a [100%] Built target klu [100%] Linking C static library libumfpack.a [100%] Built target umfpack [100%] Linking C static library libmetis.a [100%] Built target metis [100%] Linking CXX static library lib/libzmq.a [100%] Built target libzmq-static [100%] Linking C static library libcminpack.a [100%] Built target cminpack [100%] Linking C static library libexpat.a [100%] Built target expat [100%] Built target expat [100%] Building C object src/kinsol/CMakeFiles/sundials_kinsol_shared.dir/__/sundials/sundials_sptfqmr.c.obj [100%] Linking C shared library libsundials_kinsol.dll [100%] Built target sundials_kinsol_shared [100%] Building C object CMakeFiles/fmilib_shared.dir/src/Util/src/FMI2/fmi2_enums.c.obj [100%] Linking C shared library libfmilib_shared.dll [100%] Built target fmilib_shared [100%] Built target fmilib [100%] Linking C static library libOMSIBase_static.a [100%] Built target OMSIBase_static [100%] Linking C shared library libOMSIBase.dll [100%] Built target OMSIBase [100%] Linking C static library libOMSIC_static.a [100%] Built target OMSIC_static [100%] Linking C static library libOMSIC.a [100%] Built target OMSIC [100%] Linking CXX executable OMCZeroMQ.exe [100%] Built target OMCZeroMQ [100%] Linking CXX static library libOMCppSystemOMSI_static.a [100%] Built target OMCppSystemOMSI_static [100%] Linking CXX shared library libOMCppSystemOMSI.dll [100%] Built target OMCppSystemOMSI [100%] Built target OMCppSystemOMSI_static [100%] Built target OMCppSystemOMSI
from scratch:
real 66m22.514s user 14m1.201s sys 37m6.330s
Recompiling, no changes, via make -f Makefile.omdev.mingw runtimeCPPinstall -j3:
[100%] Built target OMSIBase_static [100%] Built target OMSIBase [100%] Built target OMSIC_static [100%] Built target OMSIC [100%] Built target OMCZeroMQ [100%] Linking CXX static library libOMCppSystemOMSI_static.a [100%] Built target OMCppSystemOMSI_static [100%] Linking CXX shared library libOMCppSystemOMSI.dll [100%] Built target OMCppSystemOMSI [100%] Built target OMCppSystemOMSI_static [100%] Built target OMCppSystemOMSI
takes:
real 9m38.649s user 0m49.528s sys 2m7.387s
I don't get yet why the OMSICpp is recompiled from scratch with no changes. Will debug more.
comment:6 by , 5 years ago
I think I found the problem of the always rebuild. In the CMakefile
\SimulationRuntime\OMSICpp\CMakeLists.txt there is an additional configuration file generation call at line 1169
GET_TARGET_PROPERTY(libOSU ${OSUName} LOCATION)
GET_FILENAME_COMPONENT(OSUName ${libOSU} NAME)
set (OSU_LIB ${OSUName})
configure_file (
"${PROJECT_SOURCE_DIR}/runtime/src/LibrariesConfig.h.in"
"${PROJECT_BINARY_DIR}/LibrariesConfig.h"
)
for the libraries that can be loaded during runtimen, but the omsu lib is not needed to load during runtime so it can be removed. If I comment out this command
I got this rebuild times:
real 0m15.370s
user 0m24.799s
sys 0m5.677s
I have to investigate the problem further, but maybe this solves the problem
comment:7 by , 5 years ago
This seems to have been fixed:
# first call, recompilation after git pull the PR changes adrpo33@ida-0030 MINGW64 /e/dev/OpenModelica $ time make -f Makefile.omdev.mingw -j3 runtimeCPPinstall > trace.txt 2>&1 real 8m59.247s user 0m50.438s sys 2m3.646s # second call adrpo33@ida-0030 MINGW64 /e/dev/OpenModelica $ time make -f Makefile.omdev.mingw -j3 runtimeCPPinstall > trace.txt 2>&1 real 1m55.422s user 0m31.664s sys 1m18.661s
comment:8 by , 5 years ago
$ fgrep "100%" trace.txt [100%] Built target OMSIBase_static [100%] Built target OMSIBase [100%] Built target OMSIC_static [100%] Built target OMSIC [100%] Built target OMCZeroMQ [100%] Built target OMCppSystemOMSI_static [100%] Built target OMCppSimController [100%] Built target OMCppSystemOMSI_static [100%] Built target OMCppSystemOMSI
comment:9 by , 5 years ago
| Resolution: | → fixed |
|---|---|
| Status: | new → closed |
Given @adrpo's report, I would tentatively close this ticket. Feel free to reopen it if there are still issues
comment:10 by , 5 years ago
| Milestone: | Future → 1.16.0 |
|---|
Here is the output of
make | grep 100%from a rebuild of the maintenance/v1.15 branch (skipping cmake errors and warnings). Already some duplicates for OMSI* and Kinsol, but without recompilation:The same output for the master branch: many more entries, even more duplicates and lot of recompilation even though nothing changed: