Changelog for beignet-1.3.2-6.27.x86_64.rpm :

* Thu Jan 24 2019 alarrosaAATTsuse.com- Add a _constraints file to specify this package needs at least 6500 MB of memory to build. Also use %limit_build to be sure parallel builds are taken into account.
* Sat Jan 19 2019 linnaeaAATTlavia.moe- Add Coffee Lake support, reusing Kaby Lake code.
* 0004-Enable-Coffee-Lake-support.patch
* Mon Jan 07 2019 linnaeaAATTlavia.moe- Fix build failure with llvm7 with patch from upstream Git:
* 0008-Add-preliminary-LLVM-7-support.patch (fate#325696)- Fix self-test failures on some systems:
* beignet-disable-NegAddOptimization.patch
* Wed Oct 31 2018 bwiedemannAATTsuse.com- drop indeterministic pch files to make package build reproducible
* Wed Aug 08 2018 alarrosaAATTsuse.com- Add patch beignet-llvm6.patch to fix build failure with llvm6- Fix installation of broken symlink for README.md- Use %license
* Wed Jan 31 2018 msrbAATTsuse.com- Remove dependency on clang-devel-static. (bnc#1065464)
* It was removed, clang-devel now again provides everything necessary as shared libraries.
* Thu Oct 26 2017 mpluskalAATTsuse.com- Update to version 1.3.2:
* Add some Gemini Lake devices support.
* Add llvm4.0 and llvm5.0 support.
* Add cl_intel_required_subgroup_size extension support.
* Fix some other bugs.- Drop no longer needed patches:
* 0001-Backend-Remove-old-llvm-support-code.patch
* 0002-Backend-Fix-an-include-file-error-problem.patch
* 0003-Backend-Refine-GEP-lowering-code.patch
* 0004-Backend-Refine-LLVM-version-check-macro.patch
* 0005-Backend-Refine-FCmp-one-and-une.patch
* 0006-utest-fix-image-qualifier-of-compiler_fill_gl_image-.patch
* 0007-Backend-Add-LLVM40-support.patch
* Thu May 04 2017 mardnhAATTgmx.de- Add patches for LLVM 4.0 support
* 0001-Backend-Remove-old-llvm-support-code.patch
* 0002-Backend-Fix-an-include-file-error-problem.patch
* 0003-Backend-Refine-GEP-lowering-code.patch
* 0004-Backend-Refine-LLVM-version-check-macro.patch
* 0005-Backend-Refine-FCmp-one-and-une.patch
* 0006-utest-fix-image-qualifier-of-compiler_fill_gl_image-.patch
* 0007-Backend-Add-LLVM40-support.patch
* Tue Mar 14 2017 mpluskalAATTsuse.com- Update to version 1.3.1:
* OpenCL 2.0 support.
* OpenCL event and enqueue module re-implement.
* Other OpenCL runtime driver refine.
* LLVM 3.9 support.
* Extension cl_khr_gl_sharing support.
* Extension intel_subgroups_short support.
* Large kernel compiling speed up.
* Register allocation improvement.
* Enable OpenCL 2.0 where device supported.
* Optimize int64 multiply int64.
* Add cl_khr_gl_sharing example and document.
* Fix some bugs.
* Sun Jan 01 2017 jengelhAATTinai.de- Trim filler words from description.
* Sat Nov 12 2016 mpluskalAATTsuse.com- Update to version 1.2.1
* Add some Apollo Lake devices support.
* Add llvm3.9 support.
* Fix some cl_intel_subgroups extension bugs.
* Avoid too many messages when the driver could not find good work group size.
* Fix some other bugs.
* Tue Aug 30 2016 mardnhAATTgmx.de- Update to 1.2.0
* 6th generation Intel Atom Processors (Broxton) support.
* 6th generation Intel Core Processors (Kabylake) support.
* LLVM 3.7 and 3.8 support.
* ASM debug infomation and profiling support.
* Experimental double data type support for processors after 6th generation(Broadwell) .
* Full cl_intel_subgroups extension.
* OpenCL 2.0 workgroup built-in functions.
* Local copy propagation optimization and other optimizations
* intel_accelerator extension and basic intel_motion_estimation extension.
* Android build.
* Refined printf implementation.
* Bug fixes since last release.- Drop no longer needed patches (fixed upstream)
* gcc6-support.patch
* llvm38-support.patch
* std-isnan-isinf.patch
* Mon Jul 04 2016 mardnhAATTgmx.de- Add patches to support build with GCC6.1
* gcc6-support.patch
* llvm38-support.patch
* std-isnan-isinf.patch
* Sun May 08 2016 mpluskalAATTsuse.com- Update to 1.1.2
* Add Broxton support.
* Add llvm3.7 support.
* Fix event bugs.
* Fix the annoying \"Failed to release userptr...\" error message.
* Fix printf bugs.- Drop no longer needed beignet-llvm37.diff
* Fri Dec 18 2015 mpluskalAATTsuse.com- Update to 1.1.1
* Dump asm to a file.
* Use addRemappedFile to avoid creating temporary cl source file.
* add collectInsnNum to collect block instruction number.
* Fix piglit clLinkProgram fail.
* Calculate appropriate timestamps for cl profile.- Use cmake macros
* Thu Oct 22 2015 normandAATTlinux.vnet.ibm.com- add ExclusiveArch %{ix86} x86_64 to avoid build error other archi
* Tue Oct 13 2015 mardnhAATTgmx.de- add patch: beignet-llvm37.diff to support llvm 3.7- add rpmlintrc
* Fri Jul 31 2015 mardnhAATTgmx.de- update to version 1.1.0 The highlighted improvements are as below:
* Added 6th generation Intel Core Processors (SKL) support.
* Added 5th generation Intel Atom Processors (BSW) support.
* Re-implement structurized control flow algorithm.
* Added half data type support after BDW.
* Implement a new BTI solution to support dynamic bti.
* Added extension cl_intel_subgroups support.
* Replace some built-ins with llvm intrinsic.
* Add Indirect structure argument read support.
* Enable cl_khr_spir extension to build and run from SPIR binary.
* Improvement some opencv test suite performance on all platforms.
* Several other bug fixes since last release.- removed patch beignet-icd-name-fix.diff (included upstream)
* Tue Mar 17 2015 mardnhAATTgmx.de- update to version 1.0.2 It\'s a bug fix release including some bug fixes, minor new features and some performance improvements. The highlighted improvements are as below:
* Added LLVM 3.6 support.
* Fixed a bug in popcount().
* Refactor the legalize pass and const expression handling.
* Fixed a bug when hardware is inaccessible.
* Enabled multiarch 32bit and 64bit coexisting.
* Implemented some new device side clCopyImage kernels.
* Optimize clEnqueueCopyImageToBuffer for 16 aligned buffer.
* Mon Jan 19 2015 mardnhAATTgmx.de- update to version 1.0.1
* Enable userptr support by default thus it\'s possible to achieve zero-copy when create a buffer object.
* Change to use TILING_Y mode on BDW platform which lead to a major performance improvement for some cases.
* Fix one liveness bug which is the last such type of known bug in the Gen Backend.
* Fix some builtin math function\'s accuracy bugs.
* Improve the strict mode sin/cos implementation, reduce the instruction count from 1700 to 400.
* Use Clang native sampler and image types, now image/sampler are fully comply with OpenCL spec.
* Fix bugs triggered by some popular applications such as darktable. Now darktable works fine on all the supported platforms.
* Add support for old system which doesn\'t have c++11 features.
* Sat Nov 15 2014 mardnhAATTgmx.de- update to version 1.0.0- removed patch fix_no-return-in-nonvoid-function-llvm_gen_backend.diff since it is now included upstream
* Sat Oct 18 2014 mardnhAATTgmx.de- added patch: fix_no-return-in-nonvoid-function-llvm_gen_backend.diff to address a build rpm-post build error - ----------------------------------------- E: beignet no-return-in-nonvoid-function backend/src/llvm/llvm_gen_backend.cpp:1108 - ------------------------------------------ add ncurses-devel to BuildRequires- added build flags CMAKE_BUILD_TYPE=Release