public inbox for gentoo-commits@lists.gentoo.org
 help / color / mirror / Atom feed
From: "Andreas Sturmlechner" <asturm@gentoo.org>
To: gentoo-commits@lists.gentoo.org
Subject: [gentoo-commits] repo/gentoo:master commit in: dev-util/Tensile/, dev-util/Tensile/files/
Date: Mon, 19 Dec 2022 14:00:42 +0000 (UTC)	[thread overview]
Message-ID: <1671458402.3a572264e28322d9b09f6f9475cca0eede4d5f3f.asturm@gentoo> (raw)

commit:     3a572264e28322d9b09f6f9475cca0eede4d5f3f
Author:     Andreas Sturmlechner <asturm <AT> gentoo <DOT> org>
AuthorDate: Sun Dec 18 13:09:21 2022 +0000
Commit:     Andreas Sturmlechner <asturm <AT> gentoo <DOT> org>
CommitDate: Mon Dec 19 14:00:02 2022 +0000
URL:        https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=3a572264

dev-util/Tensile: drop 4.3.0-r1

Signed-off-by: Andreas Sturmlechner <asturm <AT> gentoo.org>

 dev-util/Tensile/Manifest                          |   1 -
 dev-util/Tensile/Tensile-4.3.0-r1.ebuild           |  86 ---------------
 .../files/Tensile-4.3.0-fix-arch-parse.patch       |  12 ---
 .../Tensile/files/Tensile-4.3.0-gentoopath.patch   |  62 -----------
 dev-util/Tensile/files/Tensile-4.3.0-gfx1031.patch | 119 ---------------------
 ...ensile-4.3.0-hsaco-compile-specified-arch.patch |  96 -----------------
 .../Tensile/files/Tensile-4.3.0-use-ninja.patch    |  21 ----
 7 files changed, 397 deletions(-)

diff --git a/dev-util/Tensile/Manifest b/dev-util/Tensile/Manifest
index 3b51748fa904..c3f43afdabe4 100644
--- a/dev-util/Tensile/Manifest
+++ b/dev-util/Tensile/Manifest
@@ -1,4 +1,3 @@
 DIST Tensile-5.0.2-PR1419.patch.gz 43754 BLAKE2B 3b1d7333ed0115f77816806e7735e25d897dbae3f7a40ece61d655c8f016d9d1c804b586b4e0e75661e20c517f5937a61d5ead25a545f0ce2e4a178bcbb3e7be SHA512 0154d8e14d335f4ecd86e5db335289f8662cede7e5d4b93bf6c56d8e48c1d7d0457271c078c9abbbca1f4aa91f028f6ce6a89efb783ba0207d0f39fd5de8a42c
-DIST rocm-Tensile-4.3.0.tar.gz 14250149 BLAKE2B 030138eaca2a0aadd96801e6dbd72e510716dd90553ef3795c5e04e00a34a05ecae82b24f755e4033a4acfcdb1cf26291da1e7902bb090f89a010d403e832beb SHA512 126db0b413c716fba8d5be9bff7a44fd1badacbf32f3db8d0db649819177db37ebd56fd22dd3c809655f5d29675be115e698cd10bc3d0b4b23878ae3726fce47
 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

diff --git a/dev-util/Tensile/Tensile-4.3.0-r1.ebuild b/dev-util/Tensile/Tensile-4.3.0-r1.ebuild
deleted file mode 100644
index 7cb7e94cd256..000000000000
--- a/dev-util/Tensile/Tensile-4.3.0-r1.ebuild
+++ /dev/null
@@ -1,86 +0,0 @@
-# Copyright 1999-2022 Gentoo Authors
-# Distributed under the terms of the GNU General Public License v2
-
-EAPI=8
-
-PYTHON_COMPAT=( python3_{8..10} )
-DISTUTILS_USE_PEP517=setuptools
-inherit distutils-r1 prefix
-
-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="${PYTHON_DEPS}
-	dev-python/pyyaml[${PYTHON_USEDEP}]
-	dev-python/msgpack[${PYTHON_USEDEP}]
-	>=dev-util/rocm-smi-4.3.0
-"
-DEPEND="${RDEPEND}
-	dev-util/hip:${SLOT}
-"
-
-PATCHES=( "${FILESDIR}"/${PN}-4.3.0-output-commands.patch
-		  "${FILESDIR}"/${PN}-4.3.0-hsaco-compile-specified-arch.patch
-		  "${FILESDIR}"/${PN}-4.3.0-gfx1031.patch
-		  "${FILESDIR}"/${PN}-4.3.0-fix-arch-parse.patch
-		  "${FILESDIR}"/${PN}-4.3.0-use-ninja.patch
-	  )
-
-src_prepare() {
-	distutils-r1_src_prepare
-	eapply $(prefixify_ro "${FILESDIR}"/${PN}-4.3.0-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
-	sed -e "s,\${Tensile_ROOT}/bin/,,g" -i Source/TensileCreateLibrary.cmake cmake/TensileConfig.cmake || die # ${Tensile_ROOT}/bin does not exists; call command directly
-
-	local Tensile_share_dir="\"${EPREFIX}/usr/share/${PN}\""
-	sed -e "/HipClangVersion/s/0,0,0/$(hipconfig -v)/" \
-		-e "/SourcePath/s,globalParameters\[\"ScriptPath\"\],${Tensile_share_dir}," \
-		-i Common.py || die
-
-	sed  -e "/CMAKE_CXX_COMPILER/s,globalParameters\[\"ROCmBinPath\"\],\"${EPREFIX}/usr/lib/hip/bin\"," -i ClientExecutable.py || die
-
-	sed -e "/scriptDir/s,os.path.dirname(os.path.realpath(__file__)),${Tensile_share_dir}," -i ReplacementKernels.py || die
-
-	sed -e "s,os.path.dirname(os.path.realpath(__file__)),${Tensile_share_dir},g" -i ${PN}.py || die
-
-	sed -e "s|os\.path\.dirname.*$|\"${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
-	insinto /usr/$(get_libdir)/cmake/${PN}
-	doins cmake/*.cmake
-}

diff --git a/dev-util/Tensile/files/Tensile-4.3.0-fix-arch-parse.patch b/dev-util/Tensile/files/Tensile-4.3.0-fix-arch-parse.patch
deleted file mode 100644
index 01b94b32cb6e..000000000000
--- a/dev-util/Tensile/files/Tensile-4.3.0-fix-arch-parse.patch
+++ /dev/null
@@ -1,12 +0,0 @@
-Index: Tensile-rocm-4.3.0/Tensile/Tensile.py
-===================================================================
---- Tensile-rocm-4.3.0.orig/Tensile/Tensile.py
-+++ Tensile-rocm-4.3.0/Tensile/Tensile.py
-@@ -97,7 +97,6 @@ def addCommonArguments(argParser):
-     """
- 
-     (key, value) = par.split("=")
--    value = eval(value)
-     return (key, value)
- 
-   argParser.add_argument("-d", "--device", dest="device", type=int, \

diff --git a/dev-util/Tensile/files/Tensile-4.3.0-gentoopath.patch b/dev-util/Tensile/files/Tensile-4.3.0-gentoopath.patch
deleted file mode 100644
index da135e3b38ee..000000000000
--- a/dev-util/Tensile/files/Tensile-4.3.0-gentoopath.patch
+++ /dev/null
@@ -1,62 +0,0 @@
-Index: Tensile-rocm-4.3.0/Tensile/Source/client/CMakeLists.txt
-===================================================================
---- Tensile-rocm-4.3.0.orig/Tensile/Source/client/CMakeLists.txt
-+++ Tensile-rocm-4.3.0/Tensile/Source/client/CMakeLists.txt
-@@ -50,3 +50,5 @@ endif()
- foreach(arch IN LISTS TENSILE_GPU_ARCHS)
-     target_link_libraries(tensile_client PRIVATE "--amdgpu-target=${arch}")
- endforeach(arch)
-+
-+target_link_libraries(tensile_client PRIVATE LLVMSupport)
-Index: Tensile-rocm-4.3.0/Tensile/Source/lib/CMakeLists.txt
-===================================================================
---- Tensile-rocm-4.3.0.orig/Tensile/Source/lib/CMakeLists.txt
-+++ Tensile-rocm-4.3.0/Tensile/Source/lib/CMakeLists.txt
-@@ -48,7 +48,7 @@ set(tensile_sources  ${tensile_sources}
-     )
- 
- if(TENSILE_USE_LLVM)
--    find_package(LLVM 13.0 QUIET CONFIG)
-+    find_package(LLVM PATHS @GENTOO_PORTAGE_EPREFIX@/usr/lib/llvm/roc QUIET CONFIG)
-     if(NOT LLVM_FOUND)
-         find_package(LLVM 12.0 QUIET CONFIG)
-         if(NOT LLVM_FOUND)
-Index: Tensile-rocm-4.3.0/Tensile/Common.py
-===================================================================
---- Tensile-rocm-4.3.0.orig/Tensile/Common.py
-+++ Tensile-rocm-4.3.0/Tensile/Common.py
-@@ -1667,7 +1667,7 @@ 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-4.3.0/Tensile/ClientWriter.py
-===================================================================
---- Tensile-rocm-4.3.0.orig/Tensile/ClientWriter.py
-+++ Tensile-rocm-4.3.0/Tensile/ClientWriter.py
-@@ -284,7 +284,7 @@ def getBuildNewClientLibraryScript(build
-   import io
-   runScriptFile = io.StringIO()
- 
--  callCreateLibraryCmd = globalParameters["ScriptPath"] + "/bin/TensileCreateLibrary"
-+  callCreateLibraryCmd = "TensileCreateLibrary"
- 
- 
-   if globalParameters["MergeFiles"]:
-Index: Tensile-rocm-4.3.0/Tensile/GenerateSummations.py
-===================================================================
---- Tensile-rocm-4.3.0.orig/Tensile/GenerateSummations.py
-+++ Tensile-rocm-4.3.0/Tensile/GenerateSummations.py
-@@ -57,7 +57,7 @@ def createLibraryForBenchmark(logicPath,
-     Selection.
-     """
- 
--    pythonExePath = os.path.join(os.path.dirname(os.path.realpath(__file__)), "bin", "TensileCreateLibrary")
-+    pythonExePath = "TensileCreateLibrary"
-     args = [pythonExePath, \
-         "--merge-files", "--no-legacy-components", \
-         "--new-client-only", "--no-short-file-names", "--no-library-print-debug", "--architecture=all", \

diff --git a/dev-util/Tensile/files/Tensile-4.3.0-gfx1031.patch b/dev-util/Tensile/files/Tensile-4.3.0-gfx1031.patch
deleted file mode 100644
index 99abceb7c44d..000000000000
--- a/dev-util/Tensile/files/Tensile-4.3.0-gfx1031.patch
+++ /dev/null
@@ -1,119 +0,0 @@
---- Tensile-rocm-4.3.0/pytest.ini	2021-04-22 22:35:33.000000000 +0800
-+++ Tensile-rocm-4.3.0/pytest.ini	2021-09-30 17:32:21.382806851 +0800
-@@ -88,6 +88,7 @@
-  xfail-gfx1011: architecture
-  xfail-gfx1012: architecture
-  xfail-gfx1030: architecture
-+ xfail-gfx1031: architecture
-  skip-gfx000:  architecture
-  skip-gfx900:  architecture
-  skip-gfx906:  architecture
-@@ -97,3 +98,4 @@
-  skip-gfx1011: architecture
-  skip-gfx1012: architecture
-  skip-gfx1030: architecture
-+ skip-gfx1031: architecture
---- Tensile-rocm-4.3.0/Tensile/Common.py	2021-10-05 14:41:22.490754124 +0800
-+++ Tensile-rocm-4.3.0/Tensile/Common.py	2021-09-30 17:33:19.041724015 +0800
-@@ -200,7 +200,7 @@
- globalParameters["MergeFiles"] = True             # F=store every solution and kernel in separate file; T=store all solutions in single file
- 
- globalParameters["MaxFileName"] = 64              # If a file name would be longer than this, shorten it with a hash.
-+globalParameters["SupportedISA"] = [(8,0,3), (9,0,0), (9,0,6), (9,0,8), (9,0,10), (10,1,0), (10,1,1), (10,1,2), (10,3,0),(10,3,1)] # assembly kernels writer supports these architectures
--globalParameters["SupportedISA"] = [(8,0,3), (9,0,0), (9,0,6), (9,0,8), (9,0,10), (10,1,0), (10,1,1), (10,1,2), (10,3,0)] # assembly kernels writer supports these architectures
- 
- globalParameters["GenerateManifestAndExit"] = False               # Output manifest file with list of expected library objects and exit
- globalParameters["ClientBuildPath"] = "0_Build"                   # subdirectory for host code build directory
-@@ -265,7 +265,7 @@
-   'gfx906':'vega20', 'gfx906:xnack+':'vega20', 'gfx906:xnack-':'vega20',
-   'gfx908':'arcturus','gfx908:xnack+':'arcturus', 'gfx908:xnack-':'arcturus',
-   'gfx90a':'aldebaran', 'gfx90a:xnack+':'aldebaran', 'gfx90a:xnack-':'aldebaran',
-+  'gfx1010':'navi10', 'gfx1011':'navi11', 'gfx1012':'navi12', 'gfx1030':'navi21', 'gfx1031':'navi22',
--  'gfx1010':'navi10', 'gfx1011':'navi11', 'gfx1012':'navi12', 'gfx1030':'navi21'
- }
- 
- def getArchitectureName(gfxName):
---- Tensile-rocm-4.3.0/Tensile/KernelWriterAssembly.py	2021-04-22 22:35:33.000000000 +0800
-+++ Tensile-rocm-4.3.0/Tensile/KernelWriterAssembly.py	2021-09-30 17:27:06.696711144 +0800
-@@ -2257,7 +2257,7 @@
- 
-     # integer i8x4
-     elif kernel["ProblemType"]["DataType"].isInt8x4():
-+      if self.version == (9,0,6) or self.version == (9,0,8) or self.version == (9,0,10) or self.version == (10,3,0) or self.version == (10,3,1):
--      if self.version == (9,0,6) or self.version == (9,0,8) or self.version == (9,0,10) or self.version == (10,3,0):
-         for b in range(0, kernel["ThreadTile1"]):
-           for a in range(0, kernel["ThreadTile0"]):
-             for iui in range(0, innerUnroll):
---- Tensile-rocm-4.3.0/Tensile/Source/CMakeLists.txt	2021-04-22 22:35:33.000000000 +0800
-+++ Tensile-rocm-4.3.0/Tensile/Source/CMakeLists.txt	2021-09-30 17:29:12.846345541 +0800
-@@ -51,7 +51,7 @@
-     endif()
- 
-     if(CMAKE_CXX_COMPILER STREQUAL "hipcc")
-+      set(TENSILE_GPU_ARCHS gfx803 gfx900 gfx906:xnack- gfx908:xnack- gfx90a:xnack- gfx1010 gfx1011 gfx1012 gfx1030 gfx1031 CACHE STRING "GPU architectures")
--      set(TENSILE_GPU_ARCHS gfx803 gfx900 gfx906:xnack- gfx908:xnack- gfx90a:xnack- gfx1010 gfx1011 gfx1012 gfx1030 CACHE STRING "GPU architectures")
-     else()
-       set(TENSILE_GPU_ARCHS gfx803 gfx900 gfx906 gfx908 gfx90a CACHE STRING "GPU architectures")
-     endif()
-@@ -104,7 +104,7 @@
-     set_property( CACHE Tensile_CODE_OBJECT_VERSION PROPERTY STRINGS V2 V3)
-     set_property( CACHE Tensile_COMPILER PROPERTY STRINGS hipcc)
-     set(Tensile_ARCHITECTURE all CACHE STRING "Which GPU architecture to use")
-+    set_property( CACHE Tensile_ARCHITECTURE PROPERTY STRINGS all gfx000 gfx900 gfx906:xnack- gfx908:xnack- gfx90a:xnack- gfx1010 gfx1011 gfx1012 gfx1030 gfx1031)
--    set_property( CACHE Tensile_ARCHITECTURE PROPERTY STRINGS all gfx000 gfx900 gfx906:xnack- gfx908:xnack- gfx90a:xnack- gfx1010 gfx1011 gfx1012 gfx1030)
-     set_property( CACHE Tensile_LIBRARY_FORMAT PROPERTY STRINGS yaml msgpack)
- 
-     message( STATUS "Tensile_CODE_OBJECT_VERSION from Tensile/Source/CMakeLists.txt: ${Tensile_CODE_OBJECT_VERSION}")
-@@ -206,7 +206,7 @@
-         ${Tensile_RUNTIME_LANGUAGE}     # HIP or OCL
-         ${Tensile_COMPILER}             # hipcc
-         ${Tensile_CODE_OBJECT_VERSION}  # V2 or V3
-+        ${Tensile_ARCHITECTURE}         # all, gfx000, gfx803, gfx900, gfx906:xnack-, gfx908:xnack-, gfx90a:xnack-, gfx1010, gfx1011, gfx1012, gfx1030, gfx1031
--        ${Tensile_ARCHITECTURE}         # all, gfx000, gfx803, gfx900, gfx906:xnack-, gfx908:xnack-, gfx90a:xnack-, gfx1010, gfx1011, gfx1012, gfx1030
-         ${Tensile_LIBRARY_FORMAT}       # yaml or msgpack
-         ${Tensile_MERGE_FILES}          # ON or OFF
-         ${Tensile_SHORT_FILE_NAMES}     # ON or OFF
---- Tensile-rocm-4.3.0/Tensile/Source/lib/include/Tensile/AMDGPU.hpp	2021-04-22 22:35:33.000000000 +0800
-+++ Tensile-rocm-4.3.0/Tensile/Source/lib/include/Tensile/AMDGPU.hpp	2021-09-30 17:48:11.277066515 +0800
-@@ -62,7 +62,8 @@
-             gfx1010 = 1010,
-             gfx1011 = 1011,
-             gfx1012 = 1012,
-+            gfx1030 = 1030,
-+            gfx1031 = 1031
--            gfx1030 = 1030
-         };
- 
-         AMDGPU();
---- Tensile-rocm-4.3.0/Tensile/Source/lib/include/Tensile/Serialization/Predicates.hpp	2021-04-22 22:35:33.000000000 +0800
-+++ Tensile-rocm-4.3.0/Tensile/Source/lib/include/Tensile/Serialization/Predicates.hpp	2021-09-30 17:31:26.075845249 +0800
-@@ -215,6 +215,7 @@
-                 iot::enumCase(io, value, "gfx1011", AMDGPU::Processor::gfx1011);
-                 iot::enumCase(io, value, "gfx1012", AMDGPU::Processor::gfx1012);
-                 iot::enumCase(io, value, "gfx1030", AMDGPU::Processor::gfx1030);
-+                iot::enumCase(io, value, "gfx1031", AMDGPU::Processor::gfx1031);
-             }
-         };
-     } // namespace Serialization
---- Tensile-rocm-4.3.0/Tensile/Source/lib/source/AMDGPU.cpp	2021-04-22 22:35:33.000000000 +0800
-+++ Tensile-rocm-4.3.0/Tensile/Source/lib/source/AMDGPU.cpp	2021-09-30 17:31:45.195486318 +0800
-@@ -82,6 +82,8 @@
-             return stream << "gfx1012";
-         case AMDGPU::Processor::gfx1030:
-             return stream << "gfx1030";
-+        case AMDGPU::Processor::gfx1031:
-+            return stream << "gfx1031";
-         }
-         return stream;
-     }
---- Tensile-rocm-4.3.0/Tensile/Source/TensileCreateLibrary.cmake	2021-04-22 22:35:33.000000000 +0800
-+++ Tensile-rocm-4.3.0/Tensile/Source/TensileCreateLibrary.cmake	2021-09-30 17:29:24.766121912 +0800
-@@ -138,7 +138,7 @@
-     add_library(Tensile ${options} ${Tensile_SOURCE_FILES})
-     # specify gpu targets
-     if( Tensile_ARCHITECTURE MATCHES "all" )
-+      set( Tensile_HIP_ISA "gfx803" "gfx900" "gfx906" "gfx908" "gfx1010" "gfx1011" "gfx1012" "gfx1030", "gfx1031")
--      set( Tensile_HIP_ISA "gfx803" "gfx900" "gfx906" "gfx908" "gfx1010" "gfx1011" "gfx1012" "gfx1030")
-     else()
-       set( Tensile_HIP_ISA ${Tensile_ARCHITECTURE})
-     endif()

diff --git a/dev-util/Tensile/files/Tensile-4.3.0-hsaco-compile-specified-arch.patch b/dev-util/Tensile/files/Tensile-4.3.0-hsaco-compile-specified-arch.patch
deleted file mode 100644
index 8e6753781a2a..000000000000
--- a/dev-util/Tensile/files/Tensile-4.3.0-hsaco-compile-specified-arch.patch
+++ /dev/null
@@ -1,96 +0,0 @@
-https://github.com/ROCmSoftwarePlatform/Tensile/issues/1395
-https://github.com/ROCmSoftwarePlatform/Tensile/pull/1398
-
---- a/Tensile/TensileCreateLibrary.py
-+++ b/Tensile/TensileCreateLibrary.py
-@@ -136,6 +136,35 @@ def which(p):
-                 return candidate
-     return None
- 
-+def splitArchs():
-+  # Helper for architecture
-+  def isSupported(arch):
-+    return globalParameters["AsmCaps"][arch]["SupportedISA"] and \
-+           globalParameters["AsmCaps"][arch]["SupportedSource"]
-+
-+  if ";" in globalParameters["Architecture"]:
-+    wantedArchs = globalParameters["Architecture"].split(";")
-+  else:
-+    wantedArchs = globalParameters["Architecture"].split("_")
-+  archs = []
-+  cmdlineArchs = []
-+  if "all" in wantedArchs:
-+    for arch in globalParameters['SupportedISA']:
-+      if isSupported(arch):
-+        if (arch == (9,0,6) or arch == (9,0,8) or arch == (9,0,10)):
-+          if (arch == (9,0,10)):
-+            archs += [gfxName(arch) + '-xnack+']
-+            cmdlineArchs += [gfxName(arch) + ':xnack+']
-+          archs += [gfxName(arch) + '-xnack-']
-+          cmdlineArchs += [gfxName(arch) + ':xnack-']
-+        else:
-+          archs += [gfxName(arch)]
-+          cmdlineArchs += [gfxName(arch)]
-+  else:
-+    for arch in wantedArchs:
-+      archs += [re.sub(":", "-", arch)]
-+      cmdlineArchs += [arch]
-+  return archs, cmdlineArchs
- 
- def buildSourceCodeObjectFile(CxxCompiler, outputPath, kernelFile):
-     buildPath = ensurePath(os.path.join(globalParameters['WorkingPath'], 'code_object_tmp'))
-@@ -149,24 +178,8 @@ def buildSourceCodeObjectFile(CxxCompiler, outputPath, kernelFile):
-     objectFilename = base + '.o'
-     soFilename = base + '.so'
- 
--    def isSupported(arch):
--        return globalParameters["AsmCaps"][arch]["SupportedISA"] and \
--               globalParameters["AsmCaps"][arch]["SupportedSource"]
--
-     if (CxxCompiler == "hipcc"):
--      archs = []
--      cmdlineArchs = []
--      for arch in globalParameters['SupportedISA']:
--        if isSupported(arch):
--          if (arch == (9,0,6) or arch == (9,0,8) or arch == (9,0,10)):
--            if (arch == (9,0,10)):
--              archs += [gfxName(arch) + '-xnack+']
--              cmdlineArchs += [gfxName(arch) + ':xnack+']
--            archs += [gfxName(arch) + '-xnack-']
--            cmdlineArchs += [gfxName(arch) + ':xnack-']
--          else:
--            archs += [gfxName(arch)]
--            cmdlineArchs += [gfxName(arch)]
-+      archs, cmdlineArchs = splitArchs()
- 
-       archFlags = ['--offload-arch=' + arch for arch in cmdlineArchs]
- 
-@@ -1063,11 +1076,6 @@ def buildObjectFileNames(solutionWriter, kernelWriterSource, kernelWriterAssembl
-   sourceKernels = list([k for k in kernels if k['KernelLanguage'] == 'Source'])
-   asmKernels = list([k for k in kernels if k['KernelLanguage'] == 'Assembly'])
- 
--  # Helper for architecture
--  def isSupported(arch):
--        return globalParameters["AsmCaps"][arch]["SupportedISA"] and \
--               globalParameters["AsmCaps"][arch]["SupportedSource"]
--
-   # Build a list of kernel object names.
-   for kernel in sourceKernels:
-     sourceKernelNames += [kernelWriterSource.getKernelFileBase(kernel)]
-@@ -1081,15 +1089,7 @@ def buildObjectFileNames(solutionWriter, kernelWriterSource, kernelWriterAssembl
- 
-   # Source based kernels are built for all supported architectures
-   if (cxxCompiler == 'hipcc'):
--    sourceArchs = []
--    for arch in globalParameters['SupportedISA']:
--      if isSupported(arch):
--        if (arch == (9,0,6) or arch == (9,0,8) or arch == (9,0,10)):
--          if (arch == (9,0,10)):
--            sourceArchs += [gfxName(arch) + '-xnack+']
--          sourceArchs += [gfxName(arch) + '-xnack-']
--        else:
--          sourceArchs += [gfxName(arch)]
-+    sourceArchs, _ = splitArchs()
-   else:
-     raise RuntimeError("Unknown compiler %s" % cxxCompiler)
- 

diff --git a/dev-util/Tensile/files/Tensile-4.3.0-use-ninja.patch b/dev-util/Tensile/files/Tensile-4.3.0-use-ninja.patch
deleted file mode 100644
index 6645efd72fb1..000000000000
--- a/dev-util/Tensile/files/Tensile-4.3.0-use-ninja.patch
+++ /dev/null
@@ -1,21 +0,0 @@
-Index: Tensile-rocm-4.3.0/Tensile/ClientExecutable.py
-===================================================================
---- Tensile-rocm-4.3.0.orig/Tensile/ClientExecutable.py
-+++ Tensile-rocm-4.3.0/Tensile/ClientExecutable.py
-@@ -36,6 +36,7 @@ class CMakeEnvironment:
- 
-         args = ['cmake']
-         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]
- 
-         Common.print2(' '.join(args))
-@@ -43,7 +44,7 @@ class CMakeEnvironment:
-             subprocess.check_call(args, cwd=Common.ensurePath(self.buildDir))
- 
-     def build(self):
--        args = ['make', '-j']
-+        args = ['ninja']
-         Common.print2(' '.join(args))
-         with Common.ClientExecutionLock():
-             subprocess.check_call(args, cwd=self.buildDir)


             reply	other threads:[~2022-12-19 14:00 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-12-19 14:00 Andreas Sturmlechner [this message]
  -- strict thread matches above, loose matches on Subject: below --
2025-02-13  8:38 [gentoo-commits] repo/gentoo:master commit in: dev-util/Tensile/, dev-util/Tensile/files/ Sam James
2023-11-27 14:35 Benda XU
2023-03-30 22:41 Marek Szuba
2023-02-01  1:52 Benda XU
2023-01-31 14:30 Benda XU
2022-03-05 20:51 Andrew Ammerlaan

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1671458402.3a572264e28322d9b09f6f9475cca0eede4d5f3f.asturm@gentoo \
    --to=asturm@gentoo.org \
    --cc=gentoo-commits@lists.gentoo.org \
    --cc=gentoo-dev@lists.gentoo.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox