summaryrefslogtreecommitdiffstats
path: root/build
Commit message (Collapse)AuthorAgeFilesLines
* Let autogen.sh use either glibtoolize or libtoolize on macOSWillem Jan Palenstijn2017-03-201-2/+5
| | | | This is necessary since conda's libtool installs libtoolize on macOS.
* Start work on CFloat32Data3DGPU to allow persistent/external GPU memoryWillem Jan Palenstijn2017-02-082-2/+10
|
* Clean up symlink after failed python buildWillem Jan Palenstijn2017-02-081-1/+1
|
* Add small portability comment about CUDA 5.5Willem Jan Palenstijn2017-01-191-0/+1
|
* Avoid 'echo -e' for improved sh portabilityWillem Jan Palenstijn2017-01-191-1/+1
| | | | This fixes the CUDA .d file fixup for macOS.
* Fix build_release.batWillem Jan Palenstijn2016-12-092-12/+8
|
* Update version to 1.8Willem Jan Palenstijn2016-12-051-1/+1
|
* Update headers (website+2016)Willem Jan Palenstijn2016-11-281-4/+3
|
* Separate C++ and python builds & make conda work nicelyHolger Kohr2016-11-251-2/+2
| | | | | | | | | | | | - make builder (= advanced user or us ourselves) choose compilers and CUDA - add a check for the C++11 flag for nvcc to work around the infamous boost bug if necessary - use conda boost to build the C++ library - simplify python bindings conda recipe to only build the bindings; the C++ library is now a build and runtime dependency - add runtime dependencies to python bindings recipe - some small adjustments to builder.py
* Add automated Windows build scriptsWillem Jan Palenstijn2016-11-247-0/+279
|
* Update gen.py for VS2015Willem Jan Palenstijn2016-11-241-39/+78
|
* Update CUDA archsWillem Jan Palenstijn2016-11-241-1/+1
|
* Switch to CUDA 8.0Willem Jan Palenstijn2016-11-241-2/+2
|
* Remove unused Cuda.rules fileWillem Jan Palenstijn2016-11-241-358/+0
|
* Make shared objects 755Willem Jan Palenstijn2016-11-231-5/+5
|
* Remove resolved TODOWillem Jan Palenstijn2016-11-221-1/+0
|
* Rename python-root-install to install-python-site-packagesWillem Jan Palenstijn2016-11-221-5/+5
|
* Avoid six for PYINCDIR in Makefile.inWillem Jan Palenstijn2016-11-221-2/+1
|
* Show installation paths at end of configureWillem Jan Palenstijn2016-11-221-7/+42
|
* Clean up linking against libpython for oct/mexWillem Jan Palenstijn2016-11-221-13/+4
|
* Disable octave interface by defaultWillem Jan Palenstijn2016-11-221-2/+2
|
* Build astra_mex_plugin for octaveWillem Jan Palenstijn2016-11-221-5/+18
|
* Overhaul package installationWillem Jan Palenstijn2016-11-212-70/+125
| | | | | | | | | | | | | | | | | | | | | | | | | There are now three ways of installing from configure/make: ./configure --with-install-type=prefix (default) libraries go into @libdir@ matlab tools/mex go into @datadir@/astra/matlab octave tools/mex go into @datadir@/astra/octave python module goes into site-packages ./configure --with-install-type=dir libraries go into @prefix@/lib matlab tools/mex go into @prefix@/matlab octave tools/mex go into @prefix@/octave python module goes into @prefix@/python ./configure --with-install-type=module matlab tools/mex go into @prefix@/matlab octave tools/mex go into @prefix@/octave python module goes into site-packages library is installed along with the matlab/octave/python module(s), with rpath
* Update tested cuda compute capabilitiesWillem Jan Palenstijn2016-11-211-2/+2
|
* Separate python build and install stepsWillem Jan Palenstijn2016-11-211-6/+7
|
* Show NVCCFLAGS helpWillem Jan Palenstijn2016-11-211-0/+2
|
* Add install-octave to .PHONYWillem Jan Palenstijn2016-11-211-1/+1
|
* Add experimental support for OctaveWillem Jan Palenstijn2016-11-182-1/+73
| | | | Based on initial patch by @kalvdans.
* Add SIRT pluginWillem Jan Palenstijn2016-11-181-0/+2
|
* Add comment about setting CC to CXXDaniel M. Pelt2016-04-191-0/+2
|
* Use CXX for Python compilation as wellDaniel M. Pelt2016-04-191-2/+2
|
* Merge pull request #32 from dmpelt/split-condaWillem Jan Palenstijn2016-03-031-1/+1
|\ | | | | Split conda package into c++ lib and python parts
| * Include PluginAlgorithm in non-Python builds as wellDaniel M. Pelt2016-03-031-1/+1
| |
* | Revert "Use nvcc's -MT option"Willem Jan Palenstijn2016-03-031-2/+4
|/ | | | | | This reverts commit 29c5c86e56697286a92031aaa9fdd903b3bcc426. This fixes building with CUDA 5.5.
* Merge pull request #111 from wjp/pluginbuildWillem Jan Palenstijn2016-02-221-3/+11
|\ | | | | Remove dependency of libastra on libpython
| * Add missing .PHONY targetsWillem Jan Palenstijn2016-01-211-1/+1
| |
| * Initialize Python plugins when available in MatlabWillem Jan Palenstijn2016-01-211-0/+10
| |
| * Remove dependency of libastra on libpython by refactoring PluginAlgorithmWillem Jan Palenstijn2016-01-211-2/+0
| |
* | Update gen.py to work on more python versionsWillem Jan Palenstijn2016-02-101-1/+6
|/
* Fix VPATH problem with older autoconf versionsWillem Jan Palenstijn2016-01-212-1/+7
|
* Don't build cuda non-PIC object if static libs are disabledWillem Jan Palenstijn2016-01-201-2/+10
| | | | | This significantly speeds up builds since we usually pass disable-static to libtool.
* Try to improve nvcc dependency file generationWillem Jan Palenstijn2016-01-201-0/+3
| | | | | | We now append empty targets for all dependencies listed in the .d file generated by nvcc. This mimics the behaviour of gcc's -MP option, and prevents deleted header files from breaking the build.
* Use nvcc's -MT optionWillem Jan Palenstijn2016-01-201-4/+2
|
* Avoid incorrect include dirs when building python modulesWillem Jan Palenstijn2016-01-201-2/+5
|
* Add check for scipyWillem Jan Palenstijn2016-01-201-0/+7
| | | | The module matrix_c.pyx uses it.
* Fail when python six module is not foundWillem Jan Palenstijn2016-01-201-1/+1
|
* Place Python build/temporary files in build directoryWillem Jan Palenstijn2016-01-201-9/+12
| | | | This also allows out-of-tree builds for Python.
* Allow out-of-tree buildsWillem Jan Palenstijn2016-01-201-30/+37
|
* Merge pull request #105 from wjp/stringsWillem Jan Palenstijn2016-01-181-1/+0
|\ | | | | Replace boost::lexical_cast by stringstreams
| * Replace boost::lexical_cast by stringstreamsWillem Jan Palenstijn2016-01-051-1/+0
| | | | | | | | | | | | This is to avoid the dependence of lexical_cast on the current locale. The stringstreams used for the new string parsing/output functions are explicitly imbued with the C/classic locale.