From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from lists.gentoo.org (pigeon.gentoo.org [208.92.234.80]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (4096 bits)) (No client certificate requested) by finch.gentoo.org (Postfix) with ESMTPS id 2627115812D for ; Sun, 05 Jan 2025 21:50:00 +0000 (UTC) Received: from pigeon.gentoo.org (localhost [127.0.0.1]) by pigeon.gentoo.org (Postfix) with SMTP id BC762E079C; Sun, 05 Jan 2025 21:49:57 +0000 (UTC) Received: from smtp.gentoo.org (smtp.gentoo.org [IPv6:2001:470:ea4a:1:5054:ff:fec7:86e4]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (4096 bits)) (No client certificate requested) by pigeon.gentoo.org (Postfix) with ESMTPS id 4580FE0769 for ; Sun, 05 Jan 2025 21:49:56 +0000 (UTC) Received: from oystercatcher.gentoo.org (oystercatcher.gentoo.org [148.251.78.52]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (4096 bits) server-digest SHA256) (No client certificate requested) by smtp.gentoo.org (Postfix) with ESMTPS id 3E291342F8E for ; Sun, 05 Jan 2025 21:49:55 +0000 (UTC) Received: from localhost.localdomain (localhost [IPv6:::1]) by oystercatcher.gentoo.org (Postfix) with ESMTP id 9883E19D2 for ; Sun, 05 Jan 2025 21:49:53 +0000 (UTC) From: "Sam James" To: gentoo-commits@lists.gentoo.org Content-Transfer-Encoding: 8bit Content-type: text/plain; charset=UTF-8 Reply-To: gentoo-dev@lists.gentoo.org, "Sam James" Message-ID: <1736113593.840739d7f65e0bd1827fcc4500e386f25a461cc1.sam@gentoo> Subject: [gentoo-commits] repo/gentoo:master commit in: sci-libs/rocPRIM/files/, sci-libs/rocPRIM/ X-VCS-Repository: repo/gentoo X-VCS-Files: sci-libs/rocPRIM/Manifest sci-libs/rocPRIM/files/rocPRIM-6.3.0-expand-isa-compatibility.patch sci-libs/rocPRIM/files/rocPRIM-6.3.0-fix-clang-template.patch sci-libs/rocPRIM/files/rocPRIM-6.3.0-fix-libcxx.patch sci-libs/rocPRIM/rocPRIM-6.3.0.ebuild X-VCS-Directories: sci-libs/rocPRIM/ sci-libs/rocPRIM/files/ X-VCS-Committer: sam X-VCS-Committer-Name: Sam James X-VCS-Revision: 840739d7f65e0bd1827fcc4500e386f25a461cc1 X-VCS-Branch: master Date: Sun, 05 Jan 2025 21:49:53 +0000 (UTC) Precedence: bulk List-Post: List-Help: List-Unsubscribe: List-Subscribe: List-Id: Gentoo Linux mail X-BeenThere: gentoo-commits@lists.gentoo.org X-Auto-Response-Suppress: DR, RN, NRN, OOF, AutoReply X-Archives-Salt: 20f5d213-1b63-4507-9858-31d23a388629 X-Archives-Hash: c95f2cb4bd6a9dc233bc4d9b705b12f5 commit: 840739d7f65e0bd1827fcc4500e386f25a461cc1 Author: Sv. Lockal gmail com> AuthorDate: Sun Dec 8 15:20:02 2024 +0000 Commit: Sam James gentoo org> CommitDate: Sun Jan 5 21:46:33 2025 +0000 URL: https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=840739d7 sci-libs/rocPRIM: add 6.3.0 Signed-off-by: Sv. Lockal gmail.com> Signed-off-by: Sam James gentoo.org> sci-libs/rocPRIM/Manifest | 1 + .../rocPRIM-6.3.0-expand-isa-compatibility.patch | 33 ++++++++++ .../files/rocPRIM-6.3.0-fix-clang-template.patch | 29 +++++++++ .../rocPRIM/files/rocPRIM-6.3.0-fix-libcxx.patch | 71 ++++++++++++++++++++++ sci-libs/rocPRIM/rocPRIM-6.3.0.ebuild | 70 +++++++++++++++++++++ 5 files changed, 204 insertions(+) diff --git a/sci-libs/rocPRIM/Manifest b/sci-libs/rocPRIM/Manifest index cf2bbede484f..1851042b0e4e 100644 --- a/sci-libs/rocPRIM/Manifest +++ b/sci-libs/rocPRIM/Manifest @@ -1,3 +1,4 @@ DIST rocPRIM-5.1.3.tar.gz 374510 BLAKE2B ab6960e7bd2159ac7a0d72ba6d594684f8d47028eca0940eabd35b346ac3d432a734ec575ee71341f339a9d0dead10dfafe81fb651a3b7da681aa900a5132d0e SHA512 54e14f2b40795153b862a64c9b3d336f35feccfda725daad75fae356b0147c577735884b65218c26c370ebec710af0f84aeef3103067a50db077cf9e24d2b64a DIST rocPRIM-5.7.1.tar.gz 535649 BLAKE2B 469bf59b5c92c37df96f58bac5a6559dc2cb879e5f82e539881d79bda57d23c25a9b114466042258e87862a88cd53d260155d7785c9a69a1800b0d05e931f56d SHA512 b2f0bc07dbc5cd111dd513f6c05dbc533908c52ea99caf77d6a4dd35db2b28f98c80006bd1f50a232873eab00b70e54231c38bfd1035c8845029d53690fd5e71 DIST rocPRIM-6.1.1.tar.gz 611372 BLAKE2B 984df214179a6d2c90b31bd30ec9472a0b985d9d488ea195968f25df4cac97ce48bfa4c322f2d578771ac63caeafc2c69c9dae911bc3a9d920ab9558ba7d6e15 SHA512 298eebc8a5d3fd7da71ab98a22fd39183436b3d9e14945e450f0fb35e38ea8fb99c5f2b342f1f77f21899a14d506ce96a965fbdb24f106ec9638dcd6afe6ae98 +DIST rocPRIM-6.3.0.tar.gz 745936 BLAKE2B 1cdcf886c823959bbb03f53cb6e17abdb009fa48f8e00e96f8d962dd28ef9fb8b6ed796be079b0a3948adc832de8e3449d6439711db821b8a26c5cdc7122ce61 SHA512 c38f3cebfb32c0c9d4f9c1ad261c1dbe793dec1ebc92cf024ef32606f99c91e178526d77a24ad18792099e6f3aafcea43f7bc9282bbc144672684cd5fe296a8b diff --git a/sci-libs/rocPRIM/files/rocPRIM-6.3.0-expand-isa-compatibility.patch b/sci-libs/rocPRIM/files/rocPRIM-6.3.0-expand-isa-compatibility.patch new file mode 100644 index 000000000000..8a13924c1056 --- /dev/null +++ b/sci-libs/rocPRIM/files/rocPRIM-6.3.0-expand-isa-compatibility.patch @@ -0,0 +1,33 @@ +diff --git a/rocprim/include/rocprim/device/config_types.hpp b/rocprim/include/rocprim/device/config_types.hpp +index 58729b1..7e356da 100644 +--- a/rocprim/include/rocprim/device/config_types.hpp ++++ b/rocprim/include/rocprim/device/config_types.hpp +@@ -204,16 +204,23 @@ constexpr bool prefix_equals(const char* lhs, const char* rhs, std::size_t n) + constexpr target_arch get_target_arch_from_name(const char* const arch_name, const std::size_t n) + { + constexpr const char* target_names[] +- = {"gfx803", "gfx900", "gfx906", "gfx908", "gfx90a", "gfx1030", "gfx1100", "gfx1102"}; ++ = {"gfx803", ++ "gfx900", "gfx902", "gfx909", "gfx912", ++ "gfx906", ++ "gfx908", ++ "gfx90a", ++ "gfx1030", "gfx1031", "gfx1032", "gfx1033", "gfx1034", "gfx1035", "gfx1036", ++ "gfx1100", "gfx1101", ++ "gfx1102", "gfx1103", "gfx1150", "gfx1151", "gfx1152", "gfx1153"}; + constexpr target_arch target_architectures[] = { + target_arch::gfx803, +- target_arch::gfx900, ++ target_arch::gfx900, target_arch::gfx900, target_arch::gfx900, target_arch::gfx900, + target_arch::gfx906, + target_arch::gfx908, + target_arch::gfx90a, +- target_arch::gfx1030, +- target_arch::gfx1100, +- target_arch::gfx1102, ++ target_arch::gfx1030, target_arch::gfx1030, target_arch::gfx1030, target_arch::gfx1030, target_arch::gfx1030, target_arch::gfx1030, target_arch::gfx1030, ++ target_arch::gfx1100, target_arch::gfx1100, ++ target_arch::gfx1102, target_arch::gfx1102, target_arch::gfx1102, target_arch::gfx1102, target_arch::gfx1102, target_arch::gfx1102, + }; + static_assert(sizeof(target_names) / sizeof(target_names[0]) + == sizeof(target_architectures) / sizeof(target_architectures[0]), diff --git a/sci-libs/rocPRIM/files/rocPRIM-6.3.0-fix-clang-template.patch b/sci-libs/rocPRIM/files/rocPRIM-6.3.0-fix-clang-template.patch new file mode 100644 index 000000000000..089ce6187979 --- /dev/null +++ b/sci-libs/rocPRIM/files/rocPRIM-6.3.0-fix-clang-template.patch @@ -0,0 +1,29 @@ +Fix compilation with clang-19 +Backports https://github.com/ROCm/rocPRIM/commit/c57b04baea872078bfb5b4af1165c7584fd3a95b +--- a/rocprim/include/rocprim/device/detail/device_radix_sort.hpp ++++ b/rocprim/include/rocprim/device/detail/device_radix_sort.hpp +@@ -526,15 +526,15 @@ ROCPRIM_DEVICE ROCPRIM_FORCE_INLINE void sort_single(KeysInputIterator keys_i + + ROCPRIM_SHARED_MEMORY typename sort_single_helper::storage_type storage; + +- sort_single_helper().template sort_single(keys_input, +- keys_output, +- values_input, +- values_output, +- size, +- decomposer, +- bit, +- current_radix_bits, +- storage); ++ sort_single_helper().template sort_single<>(keys_input, ++ keys_output, ++ values_input, ++ values_output, ++ size, ++ decomposer, ++ bit, ++ current_radix_bits, ++ storage); + } + + template diff --git a/sci-libs/rocPRIM/files/rocPRIM-6.3.0-fix-libcxx.patch b/sci-libs/rocPRIM/files/rocPRIM-6.3.0-fix-libcxx.patch new file mode 100644 index 000000000000..3827829483fb --- /dev/null +++ b/sci-libs/rocPRIM/files/rocPRIM-6.3.0-fix-libcxx.patch @@ -0,0 +1,71 @@ +Fix ambiguous declarations for libc++, which treats __int128_t as integral. +The issue was fixed in develop branch, but not in release. +Upstream commit: https://github.com/ROCm/rocPRIM/commit/81f6073790dee8d310e5b68cf46aae4b797cc059 +--- a/rocprim/include/rocprim/thread/radix_key_codec.hpp ++++ b/rocprim/include/rocprim/thread/radix_key_codec.hpp +@@ -78,6 +78,7 @@ struct radix_key_codec_integral + struct radix_key_codec_integral< + Key, +@@ -104,6 +105,7 @@ struct radix_key_codec_integral< + return static_cast(bit_key >> start) & mask; + } + }; ++#endif + + template + struct radix_key_codec_integral + struct radix_key_codec_integral(bit_key >> start) & mask; + } + }; ++#endif + + template + struct radix_key_codec_floating +@@ -225,6 +229,7 @@ struct radix_key_codec_base::type> + {}; + ++#ifndef _LIBCPP_VERSION + template + struct radix_key_codec_base::value>::type> +@@ -236,6 +241,7 @@ struct radix_key_codec_base::value>::type> + : radix_key_codec_integral + {}; ++#endif + + template<> + struct radix_key_codec_base +--- a/test/rocprim/test_utils_data_generation.hpp ++++ b/test/rocprim/test_utils_data_generation.hpp +@@ -268,6 +268,7 @@ inline OutputIter segmented_generate_n(OutputIter it, size_t size, Generator&& g + return it + size; + } + ++#ifndef _LIBCPP_VERSION + template + inline auto generate_random_data_n(OutputIter it, size_t size, U min, V max, Generator&& gen) + -> std::enable_if_t, __int128_t>::value, OutputIter> +@@ -299,6 +300,7 @@ inline auto generate_random_data_n(OutputIter it, size_t size, U min, V max, Gen + + return segmented_generate_n(it, size, [&]() { return static_cast(distribution(gen)); }); + } ++#endif + + template + inline auto generate_random_data_n(OutputIter it, size_t size, U min, V max, Generator&& gen) diff --git a/sci-libs/rocPRIM/rocPRIM-6.3.0.ebuild b/sci-libs/rocPRIM/rocPRIM-6.3.0.ebuild new file mode 100644 index 000000000000..cd561fb6b22d --- /dev/null +++ b/sci-libs/rocPRIM/rocPRIM-6.3.0.ebuild @@ -0,0 +1,70 @@ +# Copyright 1999-2024 Gentoo Authors +# Distributed under the terms of the GNU General Public License v2 + +EAPI=8 + +ROCM_VERSION=${PV} +inherit cmake flag-o-matic rocm + +DESCRIPTION="HIP parallel primitives for developing performant GPU-accelerated code on ROCm" +HOMEPAGE="https://github.com/ROCm/rocPRIM" +SRC_URI="https://github.com/ROCm/rocPRIM/archive/rocm-${PV}.tar.gz -> rocPRIM-${PV}.tar.gz" +S="${WORKDIR}/rocPRIM-rocm-${PV}" + +LICENSE="MIT" +SLOT="0/$(ver_cut 1-2)" +KEYWORDS="~amd64" +IUSE="benchmark test" +REQUIRED_USE=" + benchmark? ( ${ROCM_REQUIRED_USE} ) + test? ( ${ROCM_REQUIRED_USE} ) +" + +RDEPEND="dev-util/hip + benchmark? ( dev-cpp/benchmark ) + test? ( dev-cpp/gtest )" +BDEPEND="dev-build/rocm-cmake + >=dev-build/cmake-3.22" +DEPEND="${RDEPEND}" + +RESTRICT="!test? ( test )" + +PATCHES=( + "${FILESDIR}"/${PN}-6.3.0-expand-isa-compatibility.patch + "${FILESDIR}"/${PN}-6.3.0-fix-clang-template.patch + "${FILESDIR}"/${PN}-6.3.0-fix-libcxx.patch +) + +src_prepare() { + # install benchmark files + if use benchmark; then + sed -e "/get_filename_component/s,\${BENCHMARK_SOURCE},${PN}_\${BENCHMARK_SOURCE}," \ + -e "/add_executable/a\ install(TARGETS \${BENCHMARK_TARGET})" -i benchmark/CMakeLists.txt || die + fi + + cmake_src_prepare +} + +src_configure() { + rocm_use_hipcc + + # too many warnings in tests + append-cxxflags -Wno-explicit-specialization-storage-class -Wno-deprecated-declarations + + local mycmakeargs=( + -DCMAKE_SKIP_RPATH=ON + -DAMDGPU_TARGETS="$(get_amdgpu_flags)" + -DBUILD_TEST=$(usex test ON OFF) + -DBUILD_BENCHMARK=$(usex benchmark ON OFF) + -DBUILD_FILE_REORG_BACKWARD_COMPATIBILITY=OFF + -DROCM_SYMLINK_LIBS=OFF + ) + + cmake_src_configure +} + +src_test() { + check_amdgpu + # uses HMM to fit tests to default <512M iGPU VRAM + ROCPRIM_USE_HMM="1" cmake_src_test -j1 +}