sci-libs/miopen
AMD's Machine Intelligence Library
Screenshots
-
miopen-7.2.0~amd64composable-kernel debug +hipblaslt +rocblas roctracer test +llvm_slot_22 +amdgpu_targets_gfx908 +amdgpu_targets_gfx90a +amdgpu_targets_gfx942 +amdgpu_targets_gfx950 +amdgpu_targets_gfx1030 +amdgpu_targets_gfx1100 +amdgpu_targets_gfx1101 +amdgpu_targets_gfx1200 +amdgpu_targets_gfx1201 amdgpu_targets_gfx803 amdgpu_targets_gfx900 amdgpu_targets_gfx906 amdgpu_targets_gfx940 amdgpu_targets_gfx941 amdgpu_targets_gfx1010 amdgpu_targets_gfx1011 amdgpu_targets_gfx1012 amdgpu_targets_gfx1031 amdgpu_targets_gfx1102 amdgpu_targets_gfx1103 amdgpu_targets_gfx1150 amdgpu_targets_gfx1151
View 
Download 
Browse License: MIT Overlay: gentoo -
miopen-7.1.0-r1~amd64composable-kernel debug +hipblaslt +rocblas roctracer test +llvm_slot_20 +amdgpu_targets_gfx908 +amdgpu_targets_gfx90a +amdgpu_targets_gfx942 +amdgpu_targets_gfx950 +amdgpu_targets_gfx1030 +amdgpu_targets_gfx1100 +amdgpu_targets_gfx1101 +amdgpu_targets_gfx1200 +amdgpu_targets_gfx1201 amdgpu_targets_gfx803 amdgpu_targets_gfx900 amdgpu_targets_gfx906 amdgpu_targets_gfx940 amdgpu_targets_gfx941 amdgpu_targets_gfx1010 amdgpu_targets_gfx1011 amdgpu_targets_gfx1012 amdgpu_targets_gfx1031 amdgpu_targets_gfx1102 amdgpu_targets_gfx1103 amdgpu_targets_gfx1150 amdgpu_targets_gfx1151
View Download Browse License: MIT Overlay: gentoo -
miopen-7.0.2~amd64composable-kernel debug hipblaslt rocblas roctracer test +llvm_slot_20 +amdgpu_targets_gfx908 +amdgpu_targets_gfx90a +amdgpu_targets_gfx942 +amdgpu_targets_gfx950 +amdgpu_targets_gfx1030 +amdgpu_targets_gfx1100 +amdgpu_targets_gfx1101 +amdgpu_targets_gfx1200 +amdgpu_targets_gfx1201 amdgpu_targets_gfx803 amdgpu_targets_gfx900 amdgpu_targets_gfx906 amdgpu_targets_gfx940 amdgpu_targets_gfx941 amdgpu_targets_gfx1010 amdgpu_targets_gfx1011 amdgpu_targets_gfx1012 amdgpu_targets_gfx1031 amdgpu_targets_gfx1102 amdgpu_targets_gfx1103 amdgpu_targets_gfx1150 amdgpu_targets_gfx1151
View Download Browse License: MIT Overlay: gentoo -
miopen-6.4.3-r1~amd64composable-kernel debug hipblaslt rocblas roctracer test +llvm_slot_20 +amdgpu_targets_gfx908 +amdgpu_targets_gfx90a +amdgpu_targets_gfx942 +amdgpu_targets_gfx1030 +amdgpu_targets_gfx1100 +amdgpu_targets_gfx1101 +amdgpu_targets_gfx1200 +amdgpu_targets_gfx1201 amdgpu_targets_gfx803 amdgpu_targets_gfx900 amdgpu_targets_gfx906 amdgpu_targets_gfx940 amdgpu_targets_gfx941 amdgpu_targets_gfx1010 amdgpu_targets_gfx1011 amdgpu_targets_gfx1012 amdgpu_targets_gfx1031 amdgpu_targets_gfx1102 amdgpu_targets_gfx1103 amdgpu_targets_gfx1150 amdgpu_targets_gfx1151
View Download Browse License: MIT Overlay: gentoo -
miopen-6.3.3~amd64composable-kernel debug hipblaslt rocblas roctracer test +llvm_slot_19 +amdgpu_targets_gfx906 +amdgpu_targets_gfx908 +amdgpu_targets_gfx90a +amdgpu_targets_gfx942 +amdgpu_targets_gfx1030 +amdgpu_targets_gfx1100 amdgpu_targets_gfx803 amdgpu_targets_gfx900 amdgpu_targets_gfx940 amdgpu_targets_gfx941 amdgpu_targets_gfx1010 amdgpu_targets_gfx1011 amdgpu_targets_gfx1012 amdgpu_targets_gfx1031 amdgpu_targets_gfx1101 amdgpu_targets_gfx1102
View Download Browse License: MIT Overlay: gentoo
ChangeLog
commit 4de7c7a069be422b39d8c7bd79143b367e865690
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Thu Sep 10 09:49:31 2026 +0200
sci-libs/miopen: drop redundant warning comment
commit dc37c9c1faf5ee68ffe1ae87feb51abb9877f4aa
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Thu Sep 10 09:49:31 2026 +0200
sci-libs/miopen: drop redundant warning comment
commit 4222353a7e534e99f91c8c14840bece84670aa71
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Thu Sep 10 05:33:33 2026 +0200
sci-libs/miopen: trim 7.2.4 comment prose
commit 81f8e247308ccc860cd18d059153654e4244964e
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Thu Sep 10 03:24:58 2026 +0200
sci-libs/miopen: trim 10.0.0 comment prose
commit 48746af2289e66cd706f0b3ec2a184ac9e03aaad
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Thu Sep 3 19:13:54 2026 +0200
sci-libs/miopen: drop unreachable test plumbing in 10.0.0
RESTRICT="test" is unconditional, so src_test can never run. Dropped
with the flag: the gtest DEPEND, MIOPEN_TEST_ALL and MIOPEN_TEST_GDB,
the addpredict /dev/random and check_amdgpu that only the test branch
needed, and src_test itself. BUILD_TESTING is pinned OFF.
miopen-10.0.0-build-all-tests.patch goes as well. Its entire purpose was
to strip EXCLUDE_FROM_ALL from add_test_executable so the test binaries
would be built during the compile phase -- pointless when the test phase
cannot be entered. It had been rebased onto 10.0.0 as recently as
2026-08-30, for a code path that was already unreachable.
commit 48cb003b24a6115f616daffec8ef7e7c23143eb1
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Sun Aug 30 23:39:39 2026 +0200
sci-libs/miopen: drop 7.2.3
Superseded by 7.2.4, the last release of the 7.x series and the rollback
target for the ROCm 10.0 migration. Version retention keeps the last two
plus the last of each previous major series; 7.2.4 and 10.0.0 satisfy both,
so the third version is redundant.
commit 51a40317b9800d02d8e695370fb09e2a1643b4d1
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Sun Aug 30 13:08:21 2026 +0200
sci-libs/miopen: add 10.0.0
Tag moves to therock-10.0, and LLVM_COMPAT goes 22 -> 23 to join the rest of
the subslot-pinned cohort.
Take the assembler, offload bundler and OpenCL compiler from the toolchain
rocm_use_clang just selected rather than from get_llvm_prefix. Otherwise the
GPU code is compiled by one clang and assembled by another; with
hipcc[amd-llvm] that split is fatal, because vanilla clang rejects the gfx
assembly AMD emits. Resolves to the same paths when hipcc points at the system
LLVM.
New patch, ciso646: src/include/miopen/serializable.hpp includes <ciso646>,
which is not a standard header since C++20. GCC 16 libstdc++ says so with a
#warning and MIOpen builds with -Werror, so it is a hard failure. <version> is
what the warning recommends and what the include was being used for. It is the
only file in the tree that includes it.
build-all-tests regenerated: the add_executable line survives verbatim but
moved from 330 to 392.
Dropped the add_test --build sed: 10.0 emits `--build $`,
which is already the $ the sed substituted in; it previously read
CMAKE_CURRENT_BINARY_DIR, one level too deep. Carrying it forward would have
been a silent no-op, since sed exits 0 on no-match. The two surviving
CMakeLists.txt seds now assert their anchors.
dev-build/rocm-cmake gains :$ like its peers.
commit 61a8edf2f20f33033cbaca3edc6532c43cd62f97
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Sun Aug 16 16:55:16 2026 +0200
sci-libs/miopen: add Eigen dep to 7.2.3
AI kernel tuning and immediate-mode fallback link Eigen3 alongside frugally-deep.
Declare Eigen on the same GPU-target paths that enable those features.
commit 3bee23743dcdd4a44ca2720502438868f4382a07
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Sun Aug 16 15:48:25 2026 +0200
sci-libs/miopen: add Eigen dependency for AI tuning
The exact source requires Eigen alongside frugally-deep whenever AI kernel tuning or
immediate-mode fallback is enabled.
commit c013dea92b42d0ee2d0299ef4fc6391d40780d14
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Mon Jul 27 12:45:05 2026 +0200
sci-libs/miopen: declare hipBLAS-common under USE=hipblaslt
CMakeLists.txt issues two find_package(... REQUIRED) calls inside its
if(MIOPEN_USE_HIPBLASLT) block: hipblaslt, then hipblas-common. Only the
first was declared.
sci-libs/hipBLASLt does list hipBLAS-common, but in its own DEPEND, which
does not propagate to consumers. A system that depcleans after building
hipBLASLt therefore loses the hipblas-common cmake config while USE=hipblaslt
is still on, and miopen's configure fails.
Build-time only, so DEPEND rather than RDEPEND - nothing here links against
it. The gap is inherited from upstream's packaging rather than introduced
here; the ebuild is otherwise identical to the one in ::gentoo.
commit 7c384c8cac70105c06770e0ccb7af0fc295f2471
Author: Raukaan Cogbrother <cogbrother@raukaan.local>
Date: Sun May 31 03:29:35 2026 +0200
metadata: normalize tab → 2-space indent in 42 imported files
42 metadata.xml files inherited tab indentation from their origin
overlays (mostly the ROCm cluster, plus a handful of dev-python and
AMD-tooling imports). The overlay convention is 2-space; this brings
them in line with the other 498 files. Maintainer attributions
preserved verbatim.
commit 5b3c001a7da4e04ed7804e3db133c11444942553
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Fri May 29 16:14:28 2026 +0200
sci-libs, dev-libs, dev-util: fix S for ROCm 7.2.4 tarball relayout
The rocm-7.2.4 per-component release-asset tarballs are re-rolled with a
lowercase top-level wrapper dir (hipblas/, rocblas/, rocr-runtime/, ...)
where the 7.2.3 assets exploded flat into WORKDIR. The bumped ebuilds
kept S="$", so src_prepare died applying PATCHES and running
seds against paths that had shifted one level down. Point S at the
wrapper dir each tarball actually unpacks to, matching the layout
hipBLASLt/hipsparselt already use. roct-thunk-interface's libhsakmt now
lives under the rocr-runtime/ wrapper too.
commit 9df29bd0546692f7ddb304ebc0476be9a975c22f
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Fri May 29 14:12:08 2026 +0200
sci-libs/miopen: add 7.2.4
commit c85431c365861ae707e6250b3e8906668356642d
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Fri May 8 20:19:22 2026 +0200
sci-libs/miopen: drop stale patches superseded by 7.2.3
commit cd4c611b94fa7903a79ec5e0f7584a3671839f55
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Wed May 6 10:56:11 2026 +0200
sci-libs/miopen: 7.2.3 fix S= for flat upstream tarball
The 7.2.3 release-asset tarballs from ROCm/rocm-systems and ROCm/rocm-libraries
unpack contents directly into ./ instead of into a wrapping <assetname>/ subdir as in
7.2.0. Adjust S= so src_prepare finds the source tree.
commit 17e0e78c82cad18ae43ebfb88afe190b299ee553
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Wed May 6 01:58:17 2026 +0200
sci-libs/miopen: add 7.2.3
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Thu Sep 10 09:49:31 2026 +0200
sci-libs/miopen: drop redundant warning comment
commit dc37c9c1faf5ee68ffe1ae87feb51abb9877f4aa
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Thu Sep 10 09:49:31 2026 +0200
sci-libs/miopen: drop redundant warning comment
commit 4222353a7e534e99f91c8c14840bece84670aa71
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Thu Sep 10 05:33:33 2026 +0200
sci-libs/miopen: trim 7.2.4 comment prose
commit 81f8e247308ccc860cd18d059153654e4244964e
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Thu Sep 10 03:24:58 2026 +0200
sci-libs/miopen: trim 10.0.0 comment prose
commit 48746af2289e66cd706f0b3ec2a184ac9e03aaad
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Thu Sep 3 19:13:54 2026 +0200
sci-libs/miopen: drop unreachable test plumbing in 10.0.0
RESTRICT="test" is unconditional, so src_test can never run. Dropped
with the flag: the gtest DEPEND, MIOPEN_TEST_ALL and MIOPEN_TEST_GDB,
the addpredict /dev/random and check_amdgpu that only the test branch
needed, and src_test itself. BUILD_TESTING is pinned OFF.
miopen-10.0.0-build-all-tests.patch goes as well. Its entire purpose was
to strip EXCLUDE_FROM_ALL from add_test_executable so the test binaries
would be built during the compile phase -- pointless when the test phase
cannot be entered. It had been rebased onto 10.0.0 as recently as
2026-08-30, for a code path that was already unreachable.
commit 48cb003b24a6115f616daffec8ef7e7c23143eb1
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Sun Aug 30 23:39:39 2026 +0200
sci-libs/miopen: drop 7.2.3
Superseded by 7.2.4, the last release of the 7.x series and the rollback
target for the ROCm 10.0 migration. Version retention keeps the last two
plus the last of each previous major series; 7.2.4 and 10.0.0 satisfy both,
so the third version is redundant.
commit 51a40317b9800d02d8e695370fb09e2a1643b4d1
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Sun Aug 30 13:08:21 2026 +0200
sci-libs/miopen: add 10.0.0
Tag moves to therock-10.0, and LLVM_COMPAT goes 22 -> 23 to join the rest of
the subslot-pinned cohort.
Take the assembler, offload bundler and OpenCL compiler from the toolchain
rocm_use_clang just selected rather than from get_llvm_prefix. Otherwise the
GPU code is compiled by one clang and assembled by another; with
hipcc[amd-llvm] that split is fatal, because vanilla clang rejects the gfx
assembly AMD emits. Resolves to the same paths when hipcc points at the system
LLVM.
New patch, ciso646: src/include/miopen/serializable.hpp includes <ciso646>,
which is not a standard header since C++20. GCC 16 libstdc++ says so with a
#warning and MIOpen builds with -Werror, so it is a hard failure. <version> is
what the warning recommends and what the include was being used for. It is the
only file in the tree that includes it.
build-all-tests regenerated: the add_executable line survives verbatim but
moved from 330 to 392.
Dropped the add_test --build sed: 10.0 emits `--build $`,
which is already the $ the sed substituted in; it previously read
CMAKE_CURRENT_BINARY_DIR, one level too deep. Carrying it forward would have
been a silent no-op, since sed exits 0 on no-match. The two surviving
CMakeLists.txt seds now assert their anchors.
dev-build/rocm-cmake gains :$ like its peers.
commit 61a8edf2f20f33033cbaca3edc6532c43cd62f97
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Sun Aug 16 16:55:16 2026 +0200
sci-libs/miopen: add Eigen dep to 7.2.3
AI kernel tuning and immediate-mode fallback link Eigen3 alongside frugally-deep.
Declare Eigen on the same GPU-target paths that enable those features.
commit 3bee23743dcdd4a44ca2720502438868f4382a07
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Sun Aug 16 15:48:25 2026 +0200
sci-libs/miopen: add Eigen dependency for AI tuning
The exact source requires Eigen alongside frugally-deep whenever AI kernel tuning or
immediate-mode fallback is enabled.
commit c013dea92b42d0ee2d0299ef4fc6391d40780d14
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Mon Jul 27 12:45:05 2026 +0200
sci-libs/miopen: declare hipBLAS-common under USE=hipblaslt
CMakeLists.txt issues two find_package(... REQUIRED) calls inside its
if(MIOPEN_USE_HIPBLASLT) block: hipblaslt, then hipblas-common. Only the
first was declared.
sci-libs/hipBLASLt does list hipBLAS-common, but in its own DEPEND, which
does not propagate to consumers. A system that depcleans after building
hipBLASLt therefore loses the hipblas-common cmake config while USE=hipblaslt
is still on, and miopen's configure fails.
Build-time only, so DEPEND rather than RDEPEND - nothing here links against
it. The gap is inherited from upstream's packaging rather than introduced
here; the ebuild is otherwise identical to the one in ::gentoo.
commit 7c384c8cac70105c06770e0ccb7af0fc295f2471
Author: Raukaan Cogbrother <cogbrother@raukaan.local>
Date: Sun May 31 03:29:35 2026 +0200
metadata: normalize tab → 2-space indent in 42 imported files
42 metadata.xml files inherited tab indentation from their origin
overlays (mostly the ROCm cluster, plus a handful of dev-python and
AMD-tooling imports). The overlay convention is 2-space; this brings
them in line with the other 498 files. Maintainer attributions
preserved verbatim.
commit 5b3c001a7da4e04ed7804e3db133c11444942553
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Fri May 29 16:14:28 2026 +0200
sci-libs, dev-libs, dev-util: fix S for ROCm 7.2.4 tarball relayout
The rocm-7.2.4 per-component release-asset tarballs are re-rolled with a
lowercase top-level wrapper dir (hipblas/, rocblas/, rocr-runtime/, ...)
where the 7.2.3 assets exploded flat into WORKDIR. The bumped ebuilds
kept S="$", so src_prepare died applying PATCHES and running
seds against paths that had shifted one level down. Point S at the
wrapper dir each tarball actually unpacks to, matching the layout
hipBLASLt/hipsparselt already use. roct-thunk-interface's libhsakmt now
lives under the rocr-runtime/ wrapper too.
commit 9df29bd0546692f7ddb304ebc0476be9a975c22f
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Fri May 29 14:12:08 2026 +0200
sci-libs/miopen: add 7.2.4
commit c85431c365861ae707e6250b3e8906668356642d
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Fri May 8 20:19:22 2026 +0200
sci-libs/miopen: drop stale patches superseded by 7.2.3
commit cd4c611b94fa7903a79ec5e0f7584a3671839f55
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Wed May 6 10:56:11 2026 +0200
sci-libs/miopen: 7.2.3 fix S= for flat upstream tarball
The 7.2.3 release-asset tarballs from ROCm/rocm-systems and ROCm/rocm-libraries
unpack contents directly into ./ instead of into a wrapping <assetname>/ subdir as in
7.2.0. Adjust S= so src_prepare finds the source tree.
commit 17e0e78c82cad18ae43ebfb88afe190b299ee553
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Wed May 6 01:58:17 2026 +0200
sci-libs/miopen: add 7.2.3