[gmx-developers] compilation issues in master with gcc 4.8.2

[David van der Spoel]

On 2014-03-15 11:44, David van der Spoel wrote:
> In file included from
> /home/spoel/GG/master/gromacs/src/gromacs/simd/simd.h:126:0,
>                   from
> /home/spoel/GG/master/gromacs/src/gromacs/simd/tests/simd.h:79,
>                   from
> /home/spoel/GG/master/gromacs/src/gromacs/simd/tests/simd_integer.cpp:39:
> /home/spoel/GG/master/gromacs/src/gromacs/simd/tests/simd_integer.cpp:
> In member function 'virtual void
> gmx::test::{anonymous}::SimdIntegerTest_gmxSimdExtractI_Test::TestBody()':
> /home/spoel/GG/master/gromacs/src/gromacs/simd/impl_x86_sse2/impl_x86_sse2.h:123:92:
> error: the last argument must be an 8-bit immediate
>   #define gmx_simd_extract_fi(x, i)
> _mm_cvtsi128_si32(_mm_srli_si128((x), sizeof(int) * (i)))
>
>                       ^
> /home/spoel/GG/master/gromacs/src/gromacs/simd/simd.h:665:46: note: in
> expansion of macro 'gmx_simd_extract_fi'
>   #    define gmx_simd_extract_i               gmx_simd_extract_fi
>                                                ^
> /home/spoel/GG/master/gromacs/src/gromacs/simd/tests/simd_integer.cpp:154:25:
> note: in expansion of macro 'gmx_simd_extract_i'
>           extracted_int = gmx_simd_extract_i(simd, 1);
>                           ^
> /home/spoel/GG/master/gromacs/src/gromacs/simd/impl_x86_sse2/impl_x86_sse2.h:123:92:
> error: the last argument must be an 8-bit immediate
>   #define gmx_simd_extract_fi(x, i)
> _mm_cvtsi128_si32(_mm_srli_si128((x), sizeof(int) * (i)))
>
>                       ^
> /home/spoel/GG/master/gromacs/src/gromacs/simd/simd.h:665:46: note: in
> expansion of macro 'gmx_simd_extract_fi'
>   #    define gmx_simd_extract_i               gmx_simd_extract_fi
>                                                ^
> /home/spoel/GG/master/gromacs/src/gromacs/simd/tests/simd_integer.cpp:159:25:
> note: in expansion of macro 'gmx_simd_extract_i'
>           extracted_int = gmx_simd_extract_i(simd, 3);
>
> Any clues?
To follow up: it compiles fine with SSE4.1 but not with SSE2.


-- 
David van der Spoel, Ph.D., Professor of Biology
Dept. of Cell & Molec. Biol., Uppsala University.
Box 596, 75124 Uppsala, Sweden. Phone:	+46184714205.
spoel at xray.bmc.uu.se    http://folding.bmc.uu.se