[FFmpeg-devel] [PATCH] sbrdsp.asm: convert all instructions to float/SSE ones.

[Jason Garrett-Glaser]

On Wed, Mar 7, 2012 at 12:35 PM, Reimar Döffinger
<Reimar.Doeffinger at gmx.de> wrote:
> Since the values are floats, using the float operations
> makes sense, improves performance on some CPUs and
> makes the code SSE compatible instead of needing SSE2.
>
> Based on suggestion by Jason.
>
> Signed-off-by: Reimar Döffinger <Reimar.Doeffinger at gmx.de>
> ---
>  libavcodec/x86/sbrdsp.asm |   16 ++++++++--------
>  1 files changed, 8 insertions(+), 8 deletions(-)
>
> diff --git a/libavcodec/x86/sbrdsp.asm b/libavcodec/x86/sbrdsp.asm
> index c3b559b..31a1c8b 100644
> --- a/libavcodec/x86/sbrdsp.asm
> +++ b/libavcodec/x86/sbrdsp.asm
> @@ -82,14 +82,14 @@ cglobal sbr_hf_g_filt, 5, 6, 5
>     lea         r0, [r0 + r3*8]
>     neg         r3
>  .loop4:
> -    movq        m0, [r2 + 4*r3 + 0]
> -    movq        m1, [r2 + 4*r3 + 8]
> -    movq        m2, [r1 + 0*STEP]
> -    movq        m3, [r1 + 2*STEP]
> +    movlps      m0, [r2 + 4*r3 + 0]
> +    movlps      m1, [r2 + 4*r3 + 8]
> +    movlps      m2, [r1 + 0*STEP]
> +    movlps      m3, [r1 + 2*STEP]
>     movhps      m2, [r1 + 1*STEP]
>     movhps      m3, [r1 + 3*STEP]
> -    punpckldq   m0, m0
> -    punpckldq   m1, m1
> +    unpcklps    m0, m0
> +    unpcklps    m1, m1

Suggestion (not required for this patch) -- if you do an SSE3 version,
use movddup instead of movlps + unpcklps.

Jason