diff options
author | Diego Biurrun <diego@biurrun.de> | 2013-12-23 22:48:43 +0400 |
---|---|---|
committer | Diego Biurrun <diego@biurrun.de> | 2014-05-29 17:41:15 +0400 |
commit | 65d5d5865845f057cc6530a8d0f34db952d9009c (patch) | |
tree | 87f2ff44f58d550ae4076e741a4c6b3564d98bc0 /libavcodec/ppc/svq1enc_altivec.c | |
parent | 2ea2612df508abdd1f97c6a6ef56275a52c5c41e (diff) |
dsputil: Move SVQ1 encoding specific bits into svq1enc
Diffstat (limited to 'libavcodec/ppc/svq1enc_altivec.c')
-rw-r--r-- | libavcodec/ppc/svq1enc_altivec.c | 80 |
1 files changed, 80 insertions, 0 deletions
diff --git a/libavcodec/ppc/svq1enc_altivec.c b/libavcodec/ppc/svq1enc_altivec.c new file mode 100644 index 0000000000..005239f430 --- /dev/null +++ b/libavcodec/ppc/svq1enc_altivec.c @@ -0,0 +1,80 @@ +/* + * Copyright (c) 2007 Luca Barbato <lu_zero@gentoo.org> + * + * This file is part of Libav. + * + * Libav is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * Libav is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with Libav; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include <stdint.h> + +#include "config.h" +#if HAVE_ALTIVEC_H +#include <altivec.h> +#endif + +#include "libavutil/attributes.h" +#include "libavutil/ppc/types_altivec.h" +#include "libavutil/ppc/util_altivec.h" +#include "libavcodec/svq1enc.h" + +#if HAVE_ALTIVEC +static int ssd_int8_vs_int16_altivec(const int8_t *pix1, const int16_t *pix2, + int size) +{ + int i, size16 = size >> 4; + vector signed char vpix1; + vector signed short vpix2, vdiff, vpix1l, vpix1h; + union { + vector signed int vscore; + int32_t score[4]; + } u = { .vscore = vec_splat_s32(0) }; + + while (size16) { + // score += (pix1[i] - pix2[i]) * (pix1[i] - pix2[i]); + // load pix1 and the first batch of pix2 + + vpix1 = vec_unaligned_load(pix1); + vpix2 = vec_unaligned_load(pix2); + pix2 += 8; + // unpack + vpix1h = vec_unpackh(vpix1); + vdiff = vec_sub(vpix1h, vpix2); + vpix1l = vec_unpackl(vpix1); + // load another batch from pix2 + vpix2 = vec_unaligned_load(pix2); + u.vscore = vec_msum(vdiff, vdiff, u.vscore); + vdiff = vec_sub(vpix1l, vpix2); + u.vscore = vec_msum(vdiff, vdiff, u.vscore); + pix1 += 16; + pix2 += 8; + size16--; + } + u.vscore = vec_sums(u.vscore, vec_splat_s32(0)); + + size %= 16; + for (i = 0; i < size; i++) + u.score[3] += (pix1[i] - pix2[i]) * (pix1[i] - pix2[i]); + + return u.score[3]; +} +#endif /* HAVE_ALTIVEC */ + +av_cold void ff_svq1enc_init_ppc(SVQ1EncContext *c) +{ +#if HAVE_ALTIVEC + c->ssd_int8_vs_int16 = ssd_int8_vs_int16_altivec; +#endif /* HAVE_ALTIVEC */ +} |