[vlc-devel] commit: Win32: use IsProcessorFeaturePresent() to detect available instructions (Geoffroy Couprie )

[Geoffroy Couprie]

SSE4.1 and SSE.42 are not recognized by IsProcessorFeaturePresent, but
there's some code at
http://software.intel.com/en-us/articles/using-cpuid-to-detect-the-presence-of-sse-41-and-sse-42-instruction-sets/
that we can use. I would rather add code using cpuid than forking and
testing for a SIGILL on Windows.

Any thoughts?

On Sat, Jan 16, 2010 at 6:47 PM, git version control <git at videolan.org> wrote:
> vlc | branch: master | Geoffroy Couprie <geal at videolan.org> | Sat Jan 16 17:15:32 2010 +0100| [4b5229c80508c84dbb28b834c5955e1177d9f779] | committer: Geoffroy Couprie
>
> Win32: use IsProcessorFeaturePresent() to detect available instructions
>
>> http://git.videolan.org/gitweb.cgi/vlc.git/?a=commit;h=4b5229c80508c84dbb28b834c5955e1177d9f779
> ---
>
>  src/misc/cpu.c |   44 +++++++++++++++++++++++++++++++++++++++++++-
>  1 files changed, 43 insertions(+), 1 deletions(-)
>
> diff --git a/src/misc/cpu.c b/src/misc/cpu.c
> index b09debb..3640865 100644
> --- a/src/misc/cpu.c
> +++ b/src/misc/cpu.c
> @@ -40,6 +40,9 @@
>  #include <signal.h>
>  #else
>  #include <errno.h>
> +#include <windows.h>
> +#include <winbase.h>
> +#define PF_SSE3_INSTRUCTIONS_AVAILABLE 13
>  #endif
>
>  #include "libvlc.h"
> @@ -169,12 +172,12 @@ uint32_t CPUCapabilities( void )
>
>     /* test for the MMX flag */
>     cpuid( 0x00000001 );
> +
>  # if !defined (__MMX__)
>     if( ! (i_edx & 0x00800000) )
>         goto out;
>  # endif
>     i_capabilities |= CPU_CAPABILITY_MMX;
> -
>  # if defined (__SSE__)
>     i_capabilities |= CPU_CAPABILITY_MMXEXT | CPU_CAPABILITY_SSE;
>  # else
> @@ -183,9 +186,14 @@ uint32_t CPUCapabilities( void )
>         i_capabilities |= CPU_CAPABILITY_MMXEXT;
>
>  #   ifdef CAN_COMPILE_SSE
> +#   ifdef WIN32
> +        if( IsProcessorFeaturePresent( PF_XMMI_INSTRUCTIONS_AVAILABLE ) )
> +            i_capabilities |= CPU_CAPABILITY_SSE;
> +#   else
>         check_capability( "SSE", CPU_CAPABILITY_SSE,
>                           "xorps %%xmm0,%%xmm0\n" );
>  #   endif
> +#   endif
>     }
>  # endif
>
> @@ -193,40 +201,66 @@ uint32_t CPUCapabilities( void )
>     i_capabilities |= CPU_CAPABILITY_SSE2;
>  # elif defined (CAN_COMPILE_SSE2)
>     if( i_edx & 0x04000000 )
> +    {
> +#   ifdef WIN32
> +        if( IsProcessorFeaturePresent( PF_XMMI64_INSTRUCTIONS_AVAILABLE ) )
> +            i_capabilities |= CPU_CAPABILITY_SSE2;
> +#   else
>         check_capability( "SSE2", CPU_CAPABILITY_SSE2,
>                           "movupd %%xmm0, %%xmm0\n" );
> +#   endif
> +    }
>  # endif
>
>  # if defined (__SSE3__)
>     i_capabilities |= CPU_CAPABILITY_SSE3;
>  # elif defined (CAN_COMPILE_SSE3)
>     if( i_ecx & 0x00000001 )
> +    {
> +#   ifdef WIN32
> +        if( IsProcessorFeaturePresent( PF_SSE3_INSTRUCTIONS_AVAILABLE ) )
> +            i_capabilities |= CPU_CAPABILITY_SSE3;
> +#   else
>         check_capability( "SSE3", CPU_CAPABILITY_SSE3,
>                           "movsldup %%xmm1, %%xmm0\n" );
> +#   endif
> +    }
>  # endif
>
>  # if defined (__SSSE3__)
>     i_capabilities |= CPU_CAPABILITY_SSSE3;
>  # elif defined (CAN_COMPILE_SSSE3)
> +#   ifdef WIN32
> +    /* FIXME: IsProcessorFeaturePresent can't check for SSSE3 */
> +#   else
>     if( i_ecx & 0x00000200 )
>         check_capability( "SSSE3", CPU_CAPABILITY_SSSE3,
>                           "pabsw %%xmm1, %%xmm0\n" );
> +#   endif
>  # endif
>
>  # if defined (__SSE4_1__)
>     i_capabilities |= CPU_CAPABILITY_SSE4_1;
>  # elif defined (CAN_COMPILE_SSE4_1)
> +#   ifdef WIN32
> +    /* FIXME: IsProcessorFeaturePresent can't check for SSE4.1 */
> +#   else
>     if( i_ecx & 0x00080000 )
>         check_capability( "SSE4.1", CPU_CAPABILITY_SSE4_1,
>                           "pmaxsb %%xmm1, %%xmm0\n" );
> +#   endif
>  # endif
>
>  # if defined (__SSE4_2__)
>     i_capabilities |= CPU_CAPABILITY_SSE4_2;
>  # elif defined (CAN_COMPILE_SSE4_2)
> +#   ifdef WIN32
> +    /* FIXME: IsProcessorFeaturePresent can't check for SSE4.2 */
> +#   else
>     if( i_ecx & 0x00100000 )
>         check_capability( "SSE4.2", CPU_CAPABILITY_SSE4_2,
>                           "pcmpgtq %%xmm1, %%xmm0\n" );
> +#   endif
>  # endif
>
>     /* test for additional capabilities */
> @@ -241,9 +275,17 @@ uint32_t CPUCapabilities( void )
>  # if defined (__3dNOW__)
>     i_capabilities |= CPU_CAPABILITY_3DNOW;
>  # elif defined (CAN_COMPILE_3DNOW)
> +
>     if( i_edx & 0x80000000 )
> +    {
> +#   ifdef WIN32
> +        if( IsProcessorFeaturePresent( PF_3DNOW_INSTRUCTIONS_AVAILABLE ) )
> +            i_capabilities |= CPU_CAPABILITY_3DNOW;
> +#   else
>         check_capability( "3D Now!", CPU_CAPABILITY_3DNOW,
>                           "pfadd %%mm0,%%mm0\n" "femms\n" );
> +#   endif
> +    }
>  # endif
>
>     if( b_amd && ( i_edx & 0x00400000 ) )
>
> _______________________________________________
> vlc-devel mailing list
> To unsubscribe or modify your subscription options:
> http://mailman.videolan.org/listinfo/vlc-devel
>