Compilation with lcms fast float plugin

[Michael Ezra]

I am getting Performing Test LCMS_FAST_FLOAT_SYSTEM - Failed in cmake output. should I add anything to the cmake command line?

‌

\$ cmake -G "MSYS Makefiles" -DLENSFUNDBDIR=share/lensfun -DCMAKE_BUILD_TYPE="release" -DPROC_TARGET_NUMBER="2" -DCACHE_NAME_SUFFIX="-dev" -DOPTION_OMP="ON" -DCMAKE_C_FLAGS="-m64 -mwin32 -msse2 -mfpmath=sse -mthreads -Wno-aggressive-loop-optimizations -Wno-parentheses -std=c++11 -mtune=generic -Werror=unused-label -fno-math-errno -Wall -Wuninitialized -Wno-deprecated-declarations -Wno-unused-result -fopenmp -Werror=unknown-pragmas -DNDEBUG -O3 -ftree-vectorize" -DCMAKE_EXE_LINKER_FLAGS="-m64 -mthreads -static-libgcc -mtune=generic -s -O3 -fno-use-linker-plugin" -S/c/code/repo-art
make installCMake Deprecation Warning at CMakeLists.txt:5 (cmake_minimum_required):
Compatibility with CMake < 2.8.12 will be removed from a future version of
CMake.

Update the VERSION argument <min> value or use a ...<max> suffix to tell
CMake that the project does not need compatibility with older versions.

-- CMAKE_BUILD_TYPE: release
-- searching for library exiv2 in C:/msys64/mingw64/lib
-- result: C:/msys64/mingw64/lib/libexiv2.dll.a
-- Performing Test LCMS_FAST_FLOAT_SYSTEM
-- Performing Test LCMS_FAST_FLOAT_SYSTEM - Failed
-- Performing Test LCMS_INTERNAL_VISIBLE
-- Performing Test LCMS_INTERNAL_VISIBLE - Success
-- Enabling lcms2 fast_float
-- searching for library lensfun in C:/msys64/mingw64/lib
-- result: C:/msys64/mingw64/lib/liblensfun.dll.a
-- Configuring done
-- Generating done
-- Build files have been written to: C:/code_builds/art

‌

‌

Searching for * lcms* in in /mingw64 I get this:

‌

[agriggio]

no worries. The test is to see if fast float is already built into lcms2 (only available in recent git versions) or whether we need to use the one shipped with art. From your output above, the latter is chosen. So it's all good

[Michael Ezra]

Great, thank you!

[Michael Ezra]

I didn’t yet notice performance differences through regular UI use. Where would the impact be most expected?

[agriggio]

it’s mostly on full-size output, where the gain is quite significant (half the time for a 24mpix file with neutral profile on my machine). If you have a fast machine the improvement in general ui use would be much less visible, and probably hardly noticeable

[Michael Ezra]

sounds good, thank you for adding this code in!