[FFmpeg-devel] [PATCH 9/9] [ARMv6] Add VFP-accelerated version of qmf_32_subbands

[Michael Niedermayer]

On Mon, Jul 15, 2013 at 06:28:17PM +0100, Ben Avison wrote:
>                Before           After
>                Mean    StdDev   Mean    StdDev  Change
> This function   1323.0  98.0      746.2  60.6   +77.3%
> Overall        15400.0 336.4    14147.5 288.4    +8.9%
> 
> Signed-off-by: Ben Avison <bavison at riscosopen.org>
[...]

> +6:
> +        vmsr    FPSCR, OLDFPSCR
> +        ldr     WINDOW, [fp, #3*4]
> +        ldr     OUT, [fp, #4*4]
> +        sub     BUF, BUF, #32*4
> +NOVFP   ldr     SCALEINT, [fp, #6*4]
> +        mov     COUNT, #8
> +VFP     vpush   {SCALE}
> +VFP     sub     sp, sp, #3*4
> +NOVFP   sub     sp, sp, #4*4
> +7:
> +VFP     ldr     a1, [fp, #-7*4]     @ imdct
> +NOVFP   ldr     a1, [fp, #-8*4]
> +        ldmia   fp, {a2-a4}
> +VFP     stmia   sp, {WINDOW, OUT, BUF}
> +NOVFP   stmia   sp, {WINDOW, OUT, BUF, SCALEINT}
> +VFP     vldr    SCALE, [sp, #3*4]
> +        bl      ff_synth_filter_float_vfp
> +        add     OUT, OUT, #32*4
> +        add     BUF, BUF, #32*4
> +        subs    COUNT, COUNT, #1
> +        bne     7b
> +
> +        sub     sp, fp, #(8+8)*4

libavcodec/arm/dcadsp_vfp.S:473: Error: r13 not allowed here -- `sub sp,fp,#(8+8)*4'
make: *** [libavcodec/arm/dcadsp_vfp.o] Error 1

--extra-cflags='-mfpu=neon -mfloat-abi=softfp' --cpu=cortex-a8 --arch=armv7 --target-os=linux --enable-cross-compile

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

It is dangerous to be right in matters on which the established authorities
are wrong. -- Voltaire
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20130715/eed928d1/attachment.asc>