summaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAgeFilesLines
* dev-util/intel-graphics-compiler-1.0.4111: mark stable on amd64Marek Szuba2020-11-031-1/+1
| | | | | Bug: https://bugs.gentoo.org/742374 Signed-off-by: Marek Szuba <marecki@gentoo.org>
* dev-util/intel-graphics-compiler: drop to m-nMarek Szuba2020-11-031-4/+1
| | | | Signed-off-by: Marek Szuba <marecki@gentoo.org>
* dev-util/intel-graphics-compiler-1.0.4111: only block llvm-10 in TCMarek Szuba2020-09-201-3/+9
| | | | | | | | | | | Turns out this version is not going anywhere any time soon, in fact there is a stablereq for it now - so let's handle this properly. Specifically, pkg_pretend() now dies if the current compiler is clang-10+ (I think it is safe to assume that if 10 does not work nor will the newer versions). Bug: https://bugs.gentoo.org/738934 Signed-off-by: Marek Szuba <marecki@gentoo.org>
* dev-util/intel-graphics-compiler: bump to 1.0.4944Marek Szuba2020-09-202-0/+73
| | | | | | | | | | | | The first version supporting VectorCompiler which actually builds (as well as lets intel-neo build afterwards; tested on 20.36 a moment ago) without patching with VC support disabled. Nb. I am still working on enabling VC support in Gentoo, dependencies have been taken care of but now src_compile fails due to IGC being unable to find functions and classes declared ELSEWHERE IN ITS OWN CODE. Signed-off-by: Marek Szuba <marecki@gentoo.org>
* Revert "dev-util/intel-graphics-compiler: bump to 1.0.4427"Marek Szuba2020-09-143-84/+0
| | | | | | | | | It might _build_ fine without VC but it still causes intel-neo to fail to compile. This reverts commit a8e3b1a013be805fff6aa076d62c32f3be86fa9b. Signed-off-by: Marek Szuba <marecki@gentoo.org>
* dev-util/intel-graphics-compiler: bump to 1.0.4427Marek Szuba2020-09-133-0/+84
| | | | | | | | This is the newest version that I can build successfully on my system with VC disabled, all the newer ones fail due to missing bits in IGC code itself which disabling VectorCompute seems to mask out. Signed-off-by: Marek Szuba <marecki@gentoo.org>
* dev-util/intel-graphics-compiler: bypass opencl-clang version checkMarek Szuba2020-09-072-0/+29
| | | | | | | | | | | Ebuilds already make sure we have a new-enough version of dev-libs/opencl-clang present, besides we have never even had versions older than 5.0.0 in the tree. No opencl-clang version check in CMake - no need to call readelf - no more issues on USE=-native-symlinks systems. Closes: https://bugs.gentoo.org/739138 Signed-off-by: Marek Szuba <marecki@gentoo.org>
* dev-util/intel-graphics-compiler: block llvm-10.0.1Marek Szuba2020-08-251-1/+3
| | | | | | | Causes compilation errors. Closes: https://bugs.gentoo.org/738934 Signed-off-by: Marek Szuba <marecki@gentoo.org>
* dev-util/intel-graphics-compiler: remove oldMarek Szuba2020-06-262-52/+0
| | | | Signed-off-by: Marek Szuba <marecki@gentoo.org>
* dev-util/intel-graphics-compiler: remove oldMarek Szuba2020-06-194-130/+0
| | | | Signed-off-by: Marek Szuba <marecki@gentoo.org>
* dev-util/intel-graphics-compiler: bump to 1.0.4111Marek Szuba2020-06-192-0/+52
| | | | | | Now with LLVM-10 support. Signed-off-by: Marek Szuba <marecki@gentoo.org>
* dev-util/intel-graphics-compiler: bump to 1.0.3826Marek Szuba2020-04-262-0/+52
| | | | | | | | | | Moreover, src_configure() now explicitly appends -DNDEBUG to CPPFLAGS - since commit 95577dd5076a8e9864e82879fd3af97cf63fcfe9 cmake.eclass doesn't do it any more and in IGC this definition does considerably more than just disable asserts. Bug: https://bugs.gentoo.org/718824 Signed-off-by: Marek Szuba <marecki@gentoo.org>
* dev-util/intel-graphics-compiler: remove oldMarek Szuba2020-03-123-74/+0
| | | | Signed-off-by: Marek Szuba <marecki@gentoo.org>
* dev-util/intel-graphics-compiler: Bump version to 1.0.3445Jan Vesely2020-03-123-0/+92
| | | | | | | | Add patch and enable opencl-clang:9 and llvm:9. Signed-off-by: Jan Vesely <jano.vesely@gmail.com> Closes: https://github.com/gentoo/gentoo/pull/14925 Signed-off-by: Marek Szuba <marecki@gentoo.org>
* dev-util/intel-graphics-compiler: switch to cmake.eclass in 1.0.3151Marek Szuba2020-01-271-1/+2
| | | | | Package-Manager: Portage-2.3.84, Repoman-2.3.20 Signed-off-by: Marek Szuba <marecki@gentoo.org>
* dev-util/intel-graphics-compiler: bump to 1.0.3151Marek Szuba2020-01-252-0/+37
| | | | | | | | It still depends on llvm-8 though, it gave me a lot of "undefined reference" errors when I tried to build it against 9.0.1. Package-Manager: Portage-2.3.84, Repoman-2.3.20 Signed-off-by: Marek Szuba <marecki@gentoo.org>
* dev-util/intel-graphics-compiler: remove oldMarek Szuba2019-12-102-37/+0
| | | | | Package-Manager: Portage-2.3.79, Repoman-2.3.16 Signed-off-by: Marek Szuba <marecki@gentoo.org>
* dev-util/intel-graphics-compiler: bump to 1.0.2990Marek Szuba2019-12-102-0/+37
| | | | | Package-Manager: Portage-2.3.79, Repoman-2.3.16 Signed-off-by: Marek Szuba <marecki@gentoo.org>
* dev-util/intel-graphics-compiler: bump to 1.0.2878Marek Szuba2019-11-282-0/+37
| | | | | Package-Manager: Portage-2.3.79, Repoman-2.3.16 Signed-off-by: Marek Szuba <marecki@gentoo.org>
* dev-util/intel-graphics-compiler: pass $LLVM_MAX_SLOT to get_llvm_prefix()Marek Szuba2019-10-072-0/+37
| | | | | | | | | | When invoked without max_slot, get_llvm_prefix() iterates over *all* LLVM slots known to llvm.eclass - including those exceeding LLVM_MAX_SLOT. As a consequence, an ebuild can e.g. end up getting installed into llvm:9 directories in spite of having been linked against llvm:8. Package-Manager: Portage-2.3.76, Repoman-2.3.16 Signed-off-by: Marek Szuba <marecki@gentoo.org>
* dev-util/intel-graphics-compiler: remove oldMarek Szuba2019-10-075-124/+0
| | | | | Package-Manager: Portage-2.3.76, Repoman-2.3.16 Signed-off-by: Marek Szuba <marecki@gentoo.org>
* dev-util/intel-graphics-compiler: bump to 1.0.11Marek Szuba2019-10-012-0/+37
| | | | | Package-Manager: Portage-2.3.76, Repoman-2.3.16 Signed-off-by: Marek Szuba <marecki@gentoo.org>
* dev-util/intel-graphics-compiler: fix opencl-clang filename mismatchMarek Szuba2019-08-052-0/+15
| | | | | | | | | | | | Upstream CMake files assumed that the shared-library file of opencl-clang is named libopencl-clang.so.x if LLVM version is x.0.0 but libopencl-clang.so.x.y.z otherwise. This is not correct, as demonstrated by opencl-clang-8.0.1 - the shared library is still called libopencl-clang.so.8. Make CMake always look for .so.x instead. Closes: https://bugs.gentoo.org/690520 Package-Manager: Portage-2.3.69, Repoman-2.3.16 Signed-off-by: Marek Szuba <marecki@gentoo.org>
* dev-util/intel-graphics-compiler: bump to 1.0.9Marek Szuba2019-07-033-0/+47
| | | | | Signed-off-by: Marek Szuba <marecki@gentoo.org> Package-Manager: Portage-2.3.66, Repoman-2.3.11
* dev-util/intel-graphics-compiler: set dev-libs/opencl-clang version capMarek Szuba2019-07-031-1/+1
| | | | | | | | | From 8.0.1 onwards opencl-clang uses the new library name in place of the legacy "common_clang". Make sure old IGC ebuilds do not try to use newer versions. Signed-off-by: Marek Szuba <marecki@gentoo.org> Package-Manager: Portage-2.3.66, Repoman-2.3.11
* dev-util/intel-graphics-compiler: remove oldMarek Szuba2019-07-033-82/+0
| | | | | Signed-off-by: Marek Szuba <marecki@gentoo.org> Package-Manager: Portage-2.3.66, Repoman-2.3.11
* dev-util/intel-graphics-compiler: bump to 1.0.8Marek Szuba2019-06-112-0/+35
| | | | | | | Now with proper gcc-9.1 support. Signed-off-by: Marek Szuba <marecki@gentoo.org> Package-Manager: Portage-2.3.66, Repoman-2.3.11
* dev-util/intel-graphics-compiler: bump to version 1.0.4 and EAPI-7Marek Szuba2019-05-242-0/+46
| | | | | | | | | Moreover, explicitly reject the use of gcc-9+ because they are known to fail. Bug: https://bugs.gentoo.org/685790 Signed-off-by: Marek Szuba <marecki@gentoo.org> Package-Manager: Portage-2.3.62, Repoman-2.3.11
* dev-util/intel-graphics-compiler: new packageMarek Szuba2019-05-013-0/+48
First-order dependency of Intel Graphics Compute Runtime. Signed-off-by: Marek Szuba <marecki@gentoo.org> Package-Manager: Portage-2.3.62, Repoman-2.3.11