dev-util/Tensile: add 5.4.2

Signed-off-by: Yiyang Wu <xgreenlandforwyy@gmail.com>
Signed-off-by: Benda Xu <heroxbd@gentoo.org>

4 files changed, 178 insertions, 0 deletions

diff --git a/dev-util/Tensile/Manifest b/dev-util/Tensile/Manifest
index c3f43afdabe4..dd1df20ec814 100644
--- a/dev-util/Tensile/Manifest
+++ b/dev-util/Tensile/Manifest
@@ -1,3 +1,4 @@
 DIST Tensile-5.0.2-PR1419.patch.gz 43754 BLAKE2B 3b1d7333ed0115f77816806e7735e25d897dbae3f7a40ece61d655c8f016d9d1c804b586b4e0e75661e20c517f5937a61d5ead25a545f0ce2e4a178bcbb3e7be SHA512 0154d8e14d335f4ecd86e5db335289f8662cede7e5d4b93bf6c56d8e48c1d7d0457271c078c9abbbca1f4aa91f028f6ce6a89efb783ba0207d0f39fd5de8a42c
 DIST rocm-Tensile-5.0.2.tar.gz 14362090 BLAKE2B d44c3135496db41c4eb2545f6054c84493e89ef6357da881dae6859763338417e7a9e142cdce7e2de0536684bb641049ad0440964adbe68d03ca329f01efabc9 SHA512 f5ab7d7cbc5ba25498a02a6b7c07d5a9a548edcd44b439b92c8718072382adc18ae78274db1a7174e06436ac1396f3b4edafc7f491a2c87086c368e7270e64e3
 DIST rocm-Tensile-5.1.3.tar.gz 14357517 BLAKE2B 7c5881cc2b1d25db39f54a99f9f0a08395a5f7a58deebe74d9868e8f7044db5ae2f3907196adf16dc6abeba0f1f47f5cce5c8900891276daff8e4ad614cf3b94 SHA512 4405cb6de65c1da1fd93910fc901402a24e996ff04eccadf483c25d8eb518e03ea9bbf00e372dd5fe377696e71c39355603095f3c5fc1e26d311ed15d8e3b734
+DIST rocm-Tensile-5.4.2.tar.gz 14413805 BLAKE2B ca1c13785164439f6fe83f16b36b7a0b06a63b70b9f0741d77b0fff98f38d1e1c7e9a54ef61897e6eb0a4eaf52ec9e5f14f42a7e6fd657f26306afa3de0c74a2 SHA512 d99736a202c7353979e732941199b4b20e53ff2b6c19a7175609440a840aa5624bc53a8013c435f31ac37933527771521ff02137c1c8fd53c5317eca8533be80
diff --git a/dev-util/Tensile/Tensile-5.4.2.ebuild b/dev-util/Tensile/Tensile-5.4.2.ebuild
new file mode 100644
index 000000000000..09bccd0ab2dd
--- /dev/null
+++ b/dev-util/Tensile/Tensile-5.4.2.ebuild
@@ -0,0 +1,89 @@
+# Copyright 1999-2023 Gentoo Authors
+# Distributed under the terms of the GNU General Public License v2
+
+EAPI=8
+
+PYTHON_COMPAT=( python3_{10..11} )
+DISTUTILS_USE_PEP517=setuptools
+inherit distutils-r1 llvm prefix
+
+LLVM_MAX_SLOT=15
+
+DESCRIPTION="Stretching GPU performance for GEMMs and tensor contractions"
+HOMEPAGE="https://github.com/ROCmSoftwarePlatform/Tensile"
+SRC_URI="https://github.com/ROCmSoftwarePlatform/Tensile/archive/rocm-${PV}.tar.gz -> rocm-Tensile-${PV}.tar.gz"
+S="${WORKDIR}/${PN}-rocm-${PV}"
+
+LICENSE="MIT"
+KEYWORDS="~amd64"
+SLOT="0/$(ver_cut 1-2)"
+
+# Not compatible with recent versions of pytest
+RESTRICT="test"
+
+RDEPEND="${PYTHON_DEPS}
+	sys-devel/clang:${LLVM_MAX_SLOT}
+	dev-python/pyyaml[${PYTHON_USEDEP}]
+	dev-python/msgpack[${PYTHON_USEDEP}]
+	>=dev-util/rocm-smi-4.3.0
+"
+DEPEND="${RDEPEND}
+	dev-util/hip
+"
+
+PATCHES=( "${FILESDIR}"/${PN}-4.3.0-output-commands.patch
+		  "${FILESDIR}"/${PN}-5.3.3-gfx1031.patch
+		  "${FILESDIR}"/${PN}-5.3.3-fix-arch-parse.patch
+		  "${FILESDIR}"/${PN}-5.4.2-use-ninja.patch
+	  )
+
+CMAKE_USE_DIR="${WORKDIR}/Source"
+
+src_prepare() {
+	distutils-r1_src_prepare
+	sed -e "s,\@LLVM_PATH\@,$(get_llvm_prefix ${LLVM_MAX_SLOT}),g" \
+		"${FILESDIR}"/${PN}-5.4.2-gentoopath.patch > "${S}"/gentoopath.patch || die
+	eapply $(prefixify_ro "${S}"/gentoopath.patch)
+
+	pushd ${PN} || die
+
+	sed -e "/ROCM_SMI_ROOT/s,lib,$(get_libdir)," \
+		-i Source/cmake/FindROCmSMI.cmake || die
+	sed -r -e "/TENSILE_USE_LLVM/s/ON/OFF/" \
+		-i Source/CMakeLists.txt || die
+	sed -e "/chmod 755/d" -i Source/TensileCreateLibrary.cmake || die # remove chmod 755 on
+
+	# ${Tensile_ROOT}/bin does not exists; call command directly
+	sed -e "s,\${Tensile_ROOT}/bin/,,g" -i Source/TensileCreateLibrary.cmake cmake/TensileConfig.cmake || die
+
+	local Tensile_share_dir="\"${EPREFIX}/usr/share/${PN}\""
+	sed -e "/HipClangVersion/s/0.0.0/$(hipconfig -v)/" -i Common.py || die
+
+	sed -e "s,os.path.dirname(os.path.realpath(__file__)),${Tensile_share_dir},g" \
+		-i ReplacementKernels.py Common.py ${PN}.py || die
+
+	sed -e "s|os\.path\.dirname.*$|\"${EPREFIX}/usr/share/Tensile/Source\", end='')|" -i __init__.py || die
+
+	popd || die
+
+	sed -e "/package_data/d" -e "/data_files/d" -i setup.py || die
+}
+
+python_install() {
+	distutils-r1_python_install
+
+	python_moduleinto Tensile
+	pushd Tensile || die
+	python_domodule Components
+	python_newexe Utilities/merge.py ${PN}-merge
+}
+
+src_install() {
+	distutils-r1_src_install
+
+	pushd ${PN} || die
+	insinto /usr/share/${PN}
+	doins -r Configs Perf ReplacementKernels ReplacementKernels-cov3 Source CustomKernels
+	insinto /usr/$(get_libdir)/cmake/${PN}
+	doins cmake/*.cmake
+}
diff --git a/dev-util/Tensile/files/Tensile-5.4.2-gentoopath.patch b/dev-util/Tensile/files/Tensile-5.4.2-gentoopath.patch
new file mode 100644
index 000000000000..2aed8082de05
--- /dev/null
+++ b/dev-util/Tensile/files/Tensile-5.4.2-gentoopath.patch
@@ -0,0 +1,67 @@
+Fix incorrect path in scripts and CMakeLists, in order to build tensile_client 
+===================================================================
+Index: Tensile-rocm-5.4.2/Tensile/Source/lib/CMakeLists.txt
+===================================================================
+--- Tensile-rocm-5.4.2.orig/Tensile/Source/lib/CMakeLists.txt
++++ Tensile-rocm-5.4.2/Tensile/Source/lib/CMakeLists.txt
+@@ -51,7 +51,7 @@ set(tensile_sources  ${tensile_sources}
+     )
+ 
+ if(TENSILE_USE_LLVM)
+-    find_package(LLVM 13.0 QUIET CONFIG)
++    find_package(LLVM PATHS @LLVM_PATH@ QUIET CONFIG)
+     if(NOT LLVM_FOUND)
+         find_package(LLVM 12.0 QUIET CONFIG)
+         if(NOT LLVM_FOUND)
+Index: Tensile-rocm-5.4.2/Tensile/Common.py
+===================================================================
+--- Tensile-rocm-5.4.2.orig/Tensile/Common.py
++++ Tensile-rocm-5.4.2/Tensile/Common.py
+@@ -1950,7 +1950,7 @@ def assignGlobalParameters( config ):
+     else:
+       print2(" %24s: %8s (unspecified)" % (key, defaultValue))
+ 
+-  globalParameters["ROCmPath"] = "/opt/rocm"
++  globalParameters["ROCmPath"] = "@GENTOO_PORTAGE_EPREFIX@/usr"
+   if "ROCM_PATH" in os.environ:
+     globalParameters["ROCmPath"] = os.environ.get("ROCM_PATH")
+   if "TENSILE_ROCM_PATH" in os.environ:
+Index: Tensile-rocm-5.4.2/Tensile/ClientWriter.py
+===================================================================
+--- Tensile-rocm-5.4.2.orig/Tensile/ClientWriter.py
++++ Tensile-rocm-5.4.2/Tensile/ClientWriter.py
+@@ -180,7 +180,7 @@ def runClient(libraryLogicPath, forBench
+ def getBuildClientLibraryScript(buildPath, libraryLogicPath):
+   callCreateLibraryCmd = ["python"] if os.name == "nt" else []
+ 
+-  callCreateLibraryCmd += [os.path.join(globalParameters["ScriptPath"] , "bin", "TensileCreateLibrary")]
++  callCreateLibraryCmd += "TensileCreateLibrary"
+ 
+   if globalParameters["MergeFiles"]:
+     callCreateLibraryCmd += ["--merge-files"]
+Index: Tensile-rocm-5.4.2/Tensile/GenerateSummations.py
+===================================================================
+--- Tensile-rocm-5.4.2.orig/Tensile/GenerateSummations.py
++++ Tensile-rocm-5.4.2/Tensile/GenerateSummations.py
+@@ -59,7 +59,7 @@ def createLibraryForBenchmark(logicPath,
+     Selection.
+     """
+ 
+-    pythonExePath = os.path.join(os.path.dirname(os.path.realpath(__file__)), "bin", "TensileCreateLibrary")
++    pythonExePath = "TensileCreateLibrary"
+     args = [pythonExePath, \
+         "--merge-files", "--new-client-only", "--no-short-file-names", "--no-library-print-debug", \
+         "--architecture=all", "--code-object-version=V3", "--cxx-compiler=hipcc", "--library-format=yaml", \
+Index: Tensile-rocm-5.4.2/HostLibraryTests/CMakeLists.txt
+===================================================================
+--- Tensile-rocm-5.4.2.orig/HostLibraryTests/CMakeLists.txt
++++ Tensile-rocm-5.4.2/HostLibraryTests/CMakeLists.txt
+@@ -139,7 +139,7 @@ set(test_sources ${test_sources}
+ )
+ 
+ if(TENSILE_USE_LLVM)
+-    find_package(LLVM 13.0 QUIET CONFIG)
++    find_package(LLVM PATHS @LLVM_PATH@ QUIET CONFIG)
+     if(NOT LLVM_FOUND)
+         find_package(LLVM 12.0 QUIET CONFIG)
+         if(NOT LLVM_FOUND)
diff --git a/dev-util/Tensile/files/Tensile-5.4.2-use-ninja.patch b/dev-util/Tensile/files/Tensile-5.4.2-use-ninja.patch
new file mode 100644
index 000000000000..f88bf00665cb
--- /dev/null
+++ b/dev-util/Tensile/files/Tensile-5.4.2-use-ninja.patch
@@ -0,0 +1,21 @@
+Index: Tensile-rocm-5.4.2/Tensile/ClientExecutable.py
+===================================================================
+--- Tensile-rocm-5.4.2.orig/Tensile/ClientExecutable.py
++++ Tensile-rocm-5.4.2/Tensile/ClientExecutable.py
+@@ -44,6 +44,7 @@ class CMakeEnvironment:
+         args = ['cmake']
+         args += ['-G', 'Ninja'] if (os.name == 'nt') else []
+         args += itertools.chain.from_iterable([ ['-D{}={}'.format(key, value)] for key,value in self.options.items()])
++        args += itertools.chain.from_iterable([ ['-G', 'Ninja'], [ '-D', 'CMAKE_EXPORT_COMPILE_COMMANDS=ON' ] ])
+         args += [self.sourceDir]
+         args = [cmake_path(arg) for arg in args]
+ 
+@@ -53,7 +54,7 @@ class CMakeEnvironment:
+             subprocess.check_output(args, stderr=subprocess.STDOUT, cwd=Common.ensurePath(self.buildDir))
+ 
+     def build(self):
+-        args = [('ninja' if (os.name == "nt") else 'make'), f'-j{CPUThreadCount()}']
++        args = [('ninja'), f'-j{CPUThreadCount()}']
+         Common.print2(' '.join(args))
+         with Common.ClientExecutionLock():
+             # change to use  check_output to force windows cmd block util command finish