Skip to content

xt::equal fails for int64_t due to SIMD issues. #2815

Open
Open
xt::equal fails for int64_t due to SIMD issues.#2815
Labels
AnsweredQuestion

Description

@Ivorforce
Ivorforce
opened on Nov 24, 2024

System info

macOS 13.7, intel.
cmake version 3.30.5 (CMake; JetBrains IDE bundle; build 22)
Apple clang version 15.0.0 (clang-1500.1.0.2.5)

Reproduction

#define XTENSOR_USE_XSIMD
#include <iostream>
#include <set>

#include "xtensor/xarray.hpp"
#include "xtensor/xtensor_forward.hpp"
#include "xtensor/xbuilder.hpp"
#include "xtensor/xio.hpp"

int main()
{
	xt::xarray<int64_t> x = xt::arange(50);
	xt::xarray<bool> res = xt::equal(x, x);

	std::cout << res << std::endl;
}

When removing XTENSOR_USE_XSIMD, it runs fine.
It also works when using xt::xarray<bool> res = xt::cast<uint8_t>(xt::equal(x, y));.
I'm pretty sure it fails for assignment to bool from all int types.

Error

====================[ Build | xtensor_tests | Release ]=========================
/Users/lukas/Applications/CLion.app/Contents/bin/cmake/mac/x64/bin/cmake --build /Users/lukas/dev/xtensor-tests/cmake-build-release --target xtensor_tests -j 14
[1/2] Building CXX object CMakeFiles/xtensor_tests.dir/main.cpp.o
FAILED: CMakeFiles/xtensor_tests.dir/main.cpp.o 
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/c++  -I/Users/lukas/dev/xtensor-tests/xtensor/include -I/Users/lukas/dev/xtensor-tests/xtl/include -I/Users/lukas/dev/xtensor-tests/xsimd/include -O3 -DNDEBUG -std=gnu++20 -isysroot /Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX14.2.sdk -mmacosx-version-min=13.7 -fcolor-diagnostics -MD -MT CMakeFiles/xtensor_tests.dir/main.cpp.o -MF CMakeFiles/xtensor_tests.dir/main.cpp.o.d -o CMakeFiles/xtensor_tests.dir/main.cpp.o -c /Users/lukas/dev/xtensor-tests/main.cpp
In file included from /Users/lukas/dev/xtensor-tests/main.cpp:5:
In file included from /Users/lukas/dev/xtensor-tests/xtensor/include/xtensor/xarray.hpp:19:
In file included from /Users/lukas/dev/xtensor-tests/xtensor/include/xtensor/xbuffer_adaptor.hpp:21:
In file included from /Users/lukas/dev/xtensor-tests/xtensor/include/xtensor/xstorage.hpp:21:
In file included from /Users/lukas/dev/xtensor-tests/xtensor/include/xtensor/xexception.hpp:24:
In file included from /Users/lukas/dev/xtensor-tests/xtensor/include/xtensor/xtensor_config.hpp:61:
In file included from /Users/lukas/dev/xtensor-tests/xsimd/include/xsimd/xsimd.hpp:67:
/Users/lukas/dev/xtensor-tests/xsimd/include/xsimd/types/xsimd_api.hpp:2253:9: error: no matching function for call to 'store'
        kernel::store<A>(src, dst, A {});
        ^~~~~~~~~~~~~~~~
/Users/lukas/dev/xtensor-tests/xtensor/include/xtensor/xcontainer.hpp:756:18: note: in instantiation of function template specialization 'xsimd::store_as<xsimd::sse4_1, long long>' requested here
        xt_simd::store_as(std::addressof(storage()[i]), e, align_mode());
                 ^
/Users/lukas/dev/xtensor-tests/xtensor/include/xtensor/xassign.hpp:784:25: note: in instantiation of function template specialization 'xt::xcontainer<xt::xarray_container<xt::uvector<bool, xsimd::aligned_allocator<bool, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>>>::store_simd<xsimd::aligned_mode, xsimd::batch_bool<long long>>' requested here
            e1.template store_simd<lhs_align_mode>(i, e2.template load_simd<rhs_align_mode, value_type>(i));
                        ^
/Users/lukas/dev/xtensor-tests/xtensor/include/xtensor/xassign.hpp:466:47: note: in instantiation of function template specialization 'xt::linear_assigner<true>::run<xt::xarray_container<xt::uvector<bool, xsimd::aligned_allocator<bool, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>>, xt::xfunction<xt::detail::equal_to, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &>>' requested here
                linear_assigner<simd_assign>::run(de1, de2);
                                              ^
/Users/lukas/dev/xtensor-tests/xtensor/include/xtensor/xassign.hpp:488:20: note: in instantiation of function template specialization 'xt::xexpression_assigner_base<xt::xtensor_expression_tag>::assign_data<xt::xarray_container<xt::uvector<bool, xsimd::aligned_allocator<bool, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>>, xt::xfunction<xt::detail::equal_to, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &>>' requested here
        base_type::assign_data(e1, e2, trivial_broadcast);
                   ^
/Users/lukas/dev/xtensor-tests/xtensor/include/xtensor/xassign.hpp:226:44: note: in instantiation of function template specialization 'xt::xexpression_assigner<xt::xtensor_expression_tag>::assign_xexpression<xt::xexpression<xt::xarray_container<xt::uvector<bool, xsimd::aligned_allocator<bool, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>>>, xt::xexpression<xt::xfunction<xt::detail::equal_to, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &>>>' requested here
                xexpression_assigner<tag>::assign_xexpression(e1, e2);
                                           ^
/Users/lukas/dev/xtensor-tests/xtensor/include/xtensor/xsemantic.hpp:661:13: note: in instantiation of function template specialization 'xt::assign_xexpression<xt::xarray_container<xt::uvector<bool, xsimd::aligned_allocator<bool, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>>, xt::xfunction<xt::detail::equal_to, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &>>' requested here
        xt::assign_xexpression(*this, e);
            ^
/Users/lukas/dev/xtensor-tests/xtensor/include/xtensor/xsemantic.hpp:513:37: note: in instantiation of function template specialization 'xt::xcontainer_semantic<xt::xarray_container<xt::uvector<bool, xsimd::aligned_allocator<bool, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>>>::assign_xexpression<xt::xfunction<xt::detail::equal_to, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &>>' requested here
        return this->derived_cast().assign_xexpression(e);
                                    ^
/Users/lukas/dev/xtensor-tests/xtensor/include/xtensor/xarray.hpp:510:24: note: in instantiation of function template specialization 'xt::xsemantic_base<xt::xarray_container<xt::uvector<bool, xsimd::aligned_allocator<bool, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>>>::assign<xt::xfunction<xt::detail::equal_to, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &>>' requested here
        semantic_base::assign(e);
                       ^
/Users/lukas/dev/xtensor-tests/main.cpp:62:25: note: in instantiation of function template specialization 'xt::xarray_container<xt::uvector<bool, xsimd::aligned_allocator<bool, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>>::xarray_container<xt::xfunction<xt::detail::equal_to, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &>>' requested here
        xt::xarray<bool> res = xt::equal(x, x);
                               ^
/Users/lukas/dev/xtensor-tests/xsimd/include/xsimd/arch/./generic/xsimd_generic_memory.hpp:518:27: note: candidate template ignored: could not match 'xsimd::sse4_1' against 'long long'
        XSIMD_INLINE void store(batch_bool<T, A> const& self, bool* mem, requires_arch<generic>) noexcept
                          ^
1 error generated.
ninja: build stopped: subcommand failed.

I'm not sure why it defaults to SSE4.1, but if I whack it down to core2 (macOS minimum) I still get a similar error:

====================[ Build | xtensor_tests | Release ]=========================
/Users/lukas/Applications/CLion.app/Contents/bin/cmake/mac/x64/bin/cmake --build /Users/lukas/dev/xtensor-tests/cmake-build-release --target xtensor_tests -j 14
[1/2] Building CXX object CMakeFiles/xtensor_tests.dir/main.cpp.o
FAILED: CMakeFiles/xtensor_tests.dir/main.cpp.o 
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/c++ -DXTENSOR_USE_XSIMD -isystem /Users/lukas/dev/xtensor-tests/prefix/include -march=core2 -O3 -DNDEBUG -std=gnu++20 -isysroot /Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX14.2.sdk -mmacosx-version-min=13.7 -fcolor-diagnostics -MD -MT CMakeFiles/xtensor_tests.dir/main.cpp.o -MF CMakeFiles/xtensor_tests.dir/main.cpp.o.d -o CMakeFiles/xtensor_tests.dir/main.cpp.o -c /Users/lukas/dev/xtensor-tests/main.cpp
In file included from /Users/lukas/dev/xtensor-tests/main.cpp:4:
In file included from /Users/lukas/dev/xtensor-tests/prefix/include/xtensor/xarray.hpp:19:
In file included from /Users/lukas/dev/xtensor-tests/prefix/include/xtensor/xbuffer_adaptor.hpp:21:
In file included from /Users/lukas/dev/xtensor-tests/prefix/include/xtensor/xstorage.hpp:21:
In file included from /Users/lukas/dev/xtensor-tests/prefix/include/xtensor/xexception.hpp:24:
In file included from /Users/lukas/dev/xtensor-tests/prefix/include/xtensor/xtensor_config.hpp:61:
In file included from /Users/lukas/dev/xtensor-tests/prefix/include/xsimd/xsimd.hpp:67:
/Users/lukas/dev/xtensor-tests/prefix/include/xsimd/types/xsimd_api.hpp:2253:9: error: no matching function for call to 'store'
        kernel::store<A>(src, dst, A {});
        ^~~~~~~~~~~~~~~~
/Users/lukas/dev/xtensor-tests/prefix/include/xtensor/xcontainer.hpp:756:18: note: in instantiation of function template specialization 'xsimd::store_as<xsimd::ssse3, long long>' requested here
        xt_simd::store_as(std::addressof(storage()[i]), e, align_mode());
                 ^
/Users/lukas/dev/xtensor-tests/prefix/include/xtensor/xassign.hpp:784:25: note: in instantiation of function template specialization 'xt::xcontainer<xt::xarray_container<xt::uvector<bool, xsimd::aligned_allocator<bool, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>>>::store_simd<xsimd::aligned_mode, xsimd::batch_bool<long long>>' requested here
            e1.template store_simd<lhs_align_mode>(i, e2.template load_simd<rhs_align_mode, value_type>(i));
                        ^
/Users/lukas/dev/xtensor-tests/prefix/include/xtensor/xassign.hpp:466:47: note: in instantiation of function template specialization 'xt::linear_assigner<true>::run<xt::xarray_container<xt::uvector<bool, xsimd::aligned_allocator<bool, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>>, xt::xfunction<xt::detail::equal_to, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &>>' requested here
                linear_assigner<simd_assign>::run(de1, de2);
                                              ^
/Users/lukas/dev/xtensor-tests/prefix/include/xtensor/xassign.hpp:488:20: note: in instantiation of function template specialization 'xt::xexpression_assigner_base<xt::xtensor_expression_tag>::assign_data<xt::xarray_container<xt::uvector<bool, xsimd::aligned_allocator<bool, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>>, xt::xfunction<xt::detail::equal_to, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &>>' requested here
        base_type::assign_data(e1, e2, trivial_broadcast);
                   ^
/Users/lukas/dev/xtensor-tests/prefix/include/xtensor/xassign.hpp:226:44: note: in instantiation of function template specialization 'xt::xexpression_assigner<xt::xtensor_expression_tag>::assign_xexpression<xt::xexpression<xt::xarray_container<xt::uvector<bool, xsimd::aligned_allocator<bool, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>>>, xt::xexpression<xt::xfunction<xt::detail::equal_to, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &>>>' requested here
                xexpression_assigner<tag>::assign_xexpression(e1, e2);
                                           ^
/Users/lukas/dev/xtensor-tests/prefix/include/xtensor/xsemantic.hpp:661:13: note: in instantiation of function template specialization 'xt::assign_xexpression<xt::xarray_container<xt::uvector<bool, xsimd::aligned_allocator<bool, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>>, xt::xfunction<xt::detail::equal_to, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &>>' requested here
        xt::assign_xexpression(*this, e);
            ^
/Users/lukas/dev/xtensor-tests/prefix/include/xtensor/xsemantic.hpp:513:37: note: in instantiation of function template specialization 'xt::xcontainer_semantic<xt::xarray_container<xt::uvector<bool, xsimd::aligned_allocator<bool, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>>>::assign_xexpression<xt::xfunction<xt::detail::equal_to, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &>>' requested here
        return this->derived_cast().assign_xexpression(e);
                                    ^
/Users/lukas/dev/xtensor-tests/prefix/include/xtensor/xarray.hpp:510:24: note: in instantiation of function template specialization 'xt::xsemantic_base<xt::xarray_container<xt::uvector<bool, xsimd::aligned_allocator<bool, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>>>::assign<xt::xfunction<xt::detail::equal_to, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &>>' requested here
        semantic_base::assign(e);
                       ^
/Users/lukas/dev/xtensor-tests/main.cpp:12:25: note: in instantiation of function template specialization 'xt::xarray_container<xt::uvector<bool, xsimd::aligned_allocator<bool, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>>::xarray_container<xt::xfunction<xt::detail::equal_to, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &, const xt::xarray_container<xt::uvector<long long, xsimd::aligned_allocator<long long, 16>>, xt::layout_type::row_major, xt::svector<unsigned long>> &>>' requested here
        xt::xarray<bool> res = xt::equal(x, x);
                               ^
/Users/lukas/dev/xtensor-tests/prefix/include/xsimd/arch/./generic/xsimd_generic_memory.hpp:518:27: note: candidate template ignored: could not match 'xsimd::ssse3' against 'long long'
        XSIMD_INLINE void store(batch_bool<T, A> const& self, bool* mem, requires_arch<generic>) noexcept
                          ^
1 error generated.
ninja: build stopped: subcommand failed.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    AnsweredQuestion

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions