freebsd-ports/lang/intel-compute-runtime/Makefile

# $FreeBSD$

PORTNAME=	compute-runtime
DISTVERSION=	20.13.16352
CATEGORIES=	lang
PKGNAMEPREFIX=	intel-

MAINTAINER=	jbeich@FreeBSD.org
COMMENT=	OpenCL implementation for Intel HD 5000 (Gen8) or newer

LICENSE=	MIT
LICENSE_FILE=	${WRKSRC}/LICENSE

ONLY_FOR_ARCHS=	amd64 i386
ONLY_FOR_ARCHS_REASON=	Only Intel GPUs on x86 are supported

LIB_DEPENDS=	libigc.so:devel/intel-graphics-compiler \
		libigdgmm.so:multimedia/gmmlib

USES=		cmake compiler:c++14-lang localbase:ldflags pkgconfig
USE_GITHUB=	yes
USE_LDCONFIG=	yes
GH_ACCOUNT=	intel
CMAKE_ON=	SKIP_UNIT_TESTS
CMAKE_ARGS=	-DNEO_DRIVER_VERSION:STRING="${DISTVERSIONFULL}"
PLIST_FILES=	etc/OpenCL/vendors/intel.icd \
		bin/ocloc \
		lib/intel-opencl/libigdrcl.so

post-patch:
	@${REINPLACE_CMD} -e '/-Werror/d' \
		-e '/FORCE_RESPONSE_FILE/d' \
		${WRKSRC}/CMakeLists.txt
	@${REINPLACE_CMD} -e '/OCL_ICD_VENDORDIR/s,/etc,${PREFIX}&,' \
		${WRKSRC}/package.cmake

post-install:
# install/strip has no effect on install(FILES), so strip manually
	${STRIP_CMD} ${STAGEDIR}${PREFIX}/bin/*
	${STRIP_CMD} ${STAGEDIR}${PREFIX}/lib/intel-opencl/*.so

.include <bsd.port.mk>
devel/compute-runtime: add new port Tested by: clpeak, waifu2x-converter-cpp via drm-v4.9 on Skylake The Intel(R) Graphics Compute Runtime for OpenCL(TM) is an open source project to converge Intel's development efforts on OpenCL(TM) compute stacks supporting the GEN graphics hardware architecture. https://01.org/compute-runtime 2019-12-07 18:31:44 +01:00			`# $FreeBSD$`

			`PORTNAME= compute-runtime`
lang/intel-compute-runtime: update to 20.13.16352 Changes: https://github.com/intel/compute-runtime/compare/20.12.16259...20.13.16352 Reported by: GitHub (watch releases) 2020-04-03 20:08:19 +02:00			`DISTVERSION= 20.13.16352`
devel/compute-runtime: add new port Tested by: clpeak, waifu2x-converter-cpp via drm-v4.9 on Skylake The Intel(R) Graphics Compute Runtime for OpenCL(TM) is an open source project to converge Intel's development efforts on OpenCL(TM) compute stacks supporting the GEN graphics hardware architecture. https://01.org/compute-runtime 2019-12-07 18:31:44 +01:00			`CATEGORIES= lang`
lang/compute-runtime: prefix package like other distributions Compute runtime unlike Beignet maybe used by other CPU vendors for a common runtime used by GPGPU. 2019-12-13 06:34:59 +01:00			`PKGNAMEPREFIX= intel-`
devel/compute-runtime: add new port Tested by: clpeak, waifu2x-converter-cpp via drm-v4.9 on Skylake The Intel(R) Graphics Compute Runtime for OpenCL(TM) is an open source project to converge Intel's development efforts on OpenCL(TM) compute stacks supporting the GEN graphics hardware architecture. https://01.org/compute-runtime 2019-12-07 18:31:44 +01:00
			`MAINTAINER= jbeich@FreeBSD.org`
lang/{clover,beignet}: reword to avoid confusion with ocl-icd Currently only one port provides libOpenCL.so, the rest implement OpenCL (as drivers) for different devices. 2019-12-24 09:02:06 +01:00			`COMMENT= OpenCL implementation for Intel HD 5000 (Gen8) or newer`
devel/compute-runtime: add new port Tested by: clpeak, waifu2x-converter-cpp via drm-v4.9 on Skylake The Intel(R) Graphics Compute Runtime for OpenCL(TM) is an open source project to converge Intel's development efforts on OpenCL(TM) compute stacks supporting the GEN graphics hardware architecture. https://01.org/compute-runtime 2019-12-07 18:31:44 +01:00
			`LICENSE= MIT`
			`LICENSE_FILE= ${WRKSRC}/LICENSE`

			`ONLY_FOR_ARCHS= amd64 i386`
			`ONLY_FOR_ARCHS_REASON= Only Intel GPUs on x86 are supported`

			`LIB_DEPENDS= libigc.so:devel/intel-graphics-compiler \`
			`libigdgmm.so:multimedia/gmmlib`

lang/intel-compute-runtime: update to 20.03.15346 Changes: https://github.com/intel/compute-runtime/compare/20.02.15268...20.03.15346 Reported by: GitHub (watch releases) 2020-01-24 17:06:04 +01:00			`USES= cmake compiler:c++14-lang localbase:ldflags pkgconfig`
devel/compute-runtime: add new port Tested by: clpeak, waifu2x-converter-cpp via drm-v4.9 on Skylake The Intel(R) Graphics Compute Runtime for OpenCL(TM) is an open source project to converge Intel's development efforts on OpenCL(TM) compute stacks supporting the GEN graphics hardware architecture. https://01.org/compute-runtime 2019-12-07 18:31:44 +01:00			`USE_GITHUB= yes`
			`USE_LDCONFIG= yes`
			`GH_ACCOUNT= intel`
lang/intel-compute-runtime: update to 20.01.15264 Changes: https://github.com/intel/compute-runtime/compare/19.52.15209...20.01.15264 Reported by: GitHub (watch releases) 2020-01-10 15:07:50 +01:00			`CMAKE_ON= SKIP_UNIT_TESTS`
devel/compute-runtime: add new port Tested by: clpeak, waifu2x-converter-cpp via drm-v4.9 on Skylake The Intel(R) Graphics Compute Runtime for OpenCL(TM) is an open source project to converge Intel's development efforts on OpenCL(TM) compute stacks supporting the GEN graphics hardware architecture. https://01.org/compute-runtime 2019-12-07 18:31:44 +01:00			`CMAKE_ARGS= -DNEO_DRIVER_VERSION:STRING="${DISTVERSIONFULL}"`
			`PLIST_FILES= etc/OpenCL/vendors/intel.icd \`
			`bin/ocloc \`
lang/intel-compute-runtime: update to 20.11.16158 Changes: https://github.com/intel/compute-runtime/compare/20.10.16087...20.11.16158 Reported by: GitHub (watch releases) 2020-03-21 00:03:36 +01:00			`lib/intel-opencl/libigdrcl.so`
devel/compute-runtime: add new port Tested by: clpeak, waifu2x-converter-cpp via drm-v4.9 on Skylake The Intel(R) Graphics Compute Runtime for OpenCL(TM) is an open source project to converge Intel's development efforts on OpenCL(TM) compute stacks supporting the GEN graphics hardware architecture. https://01.org/compute-runtime 2019-12-07 18:31:44 +01:00
			`post-patch:`
			`@${REINPLACE_CMD} -e '/-Werror/d' \`
lang/compute-runtime: unbreak build with ninja BSD ar(1) from base doesn't understand @file which CMake projects can use to work around maximum number of arguments on command line. $ echo /dev/null >foo.rsp $ /usr/bin/ar qc libfoo.a @foo.rsp ar: warning: can't open file: @foo.rsp: No such file or directory leading to FAILED: bin/ocloc : && /usr/bin/c++ -O2 -pipe -fstack-protector-strong -fno-strict-aliasing -mretpoline -ftemplate-depth=1024 -Wall -Wempty-body -Wignored-qualifiers -Wtype-limits -Wuninitialized -Wextra -Wno-unused-parameter -Wno-missing-field-initializers -Winvalid-pch -Wshorten-64-to-32 -Wno-unused-local-typedefs -DSANITIZER_BUILD -Wno-deprecated-register -fstack-protector-strong -O2 -D_FORTIFY_SOURCE=2 -Wformat -Wformat-security -O2 -pipe -fstack-protector-strong -fno-strict-aliasing -fstack-protector-strong @CMakeFiles/ocloc.rsp -o bin/ocloc && : ld: error: undefined symbol: CLElfLib::CElfReader::CElfReader(std::__1::vector<char, std::__1::allocator<char> >&) >>> referenced by binary_decoder.cpp >>> offline_compiler/CMakeFiles/ocloc.dir/decoder/binary_decoder.cpp.o:(BinaryDecoder::getDevBinary()) ld: error: undefined symbol: CLElfLib::CElfReader::getSectionData(unsigned long) >>> referenced by binary_decoder.cpp >>> offline_compiler/CMakeFiles/ocloc.dir/decoder/binary_decoder.cpp.o:(BinaryDecoder::getDevBinary()) >>> referenced by binary_decoder.cpp >>> offline_compiler/CMakeFiles/ocloc.dir/decoder/binary_decoder.cpp.o:(BinaryDecoder::getDevBinary()) >>> referenced by binary_decoder.cpp >>> offline_compiler/CMakeFiles/ocloc.dir/decoder/binary_decoder.cpp.o:(BinaryDecoder::getDevBinary()) >>> referenced by binary_decoder.cpp >>> offline_compiler/CMakeFiles/ocloc.dir/decoder/binary_decoder.cpp.o:(BinaryDecoder::getDevBinary()) ld: error: undefined symbol: CLElfLib::CElfWriter::resolveBinary(std::__1::vector<char, std::__1::allocator<char> >&) >>> referenced by binary_encoder.cpp >>> offline_compiler/CMakeFiles/ocloc.dir/decoder/binary_encoder.cpp.o:(BinaryEncoder::createElf()) >>> referenced by offline_compiler.cpp >>> offline_compiler/CMakeFiles/ocloc.dir/offline_compiler.cpp.o:(NEO::OfflineCompiler::generateElfBinary()) c++: error: linker command failed with exit code 1 (use -v to see invocation) ninja: build stopped: subcommand failed. 2019-12-08 18:52:36 +01:00			`-e '/FORCE_RESPONSE_FILE/d' \`
devel/compute-runtime: add new port Tested by: clpeak, waifu2x-converter-cpp via drm-v4.9 on Skylake The Intel(R) Graphics Compute Runtime for OpenCL(TM) is an open source project to converge Intel's development efforts on OpenCL(TM) compute stacks supporting the GEN graphics hardware architecture. https://01.org/compute-runtime 2019-12-07 18:31:44 +01:00			`${WRKSRC}/CMakeLists.txt`
lang/intel-compute-runtime: update to 20.12.16259 Changes: https://github.com/intel/compute-runtime/compare/20.11.16158...20.12.16259 Reported by: GitHub (watch releases) 2020-03-27 15:34:19 +01:00			`@${REINPLACE_CMD} -e '/OCL_ICD_VENDORDIR/s,/etc,${PREFIX}&,' \`
devel/compute-runtime: add new port Tested by: clpeak, waifu2x-converter-cpp via drm-v4.9 on Skylake The Intel(R) Graphics Compute Runtime for OpenCL(TM) is an open source project to converge Intel's development efforts on OpenCL(TM) compute stacks supporting the GEN graphics hardware architecture. https://01.org/compute-runtime 2019-12-07 18:31:44 +01:00			`${WRKSRC}/package.cmake`

			`post-install:`
			`# install/strip has no effect on install(FILES), so strip manually`
			`${STRIP_CMD} ${STAGEDIR}${PREFIX}/bin/*`
			`${STRIP_CMD} ${STAGEDIR}${PREFIX}/lib/intel-opencl/*.so`

			`.include <bsd.port.mk>`