Mercurial > libavcodec.hg
view kmvc.c @ 12197:fbf4d5b1b664 libavcodec
Remove FF_MM_SSE2/3 flags for CPUs where this is generally not faster than
regular MMX code. Examples of this are the Core1 CPU. Instead, set a new flag,
FF_MM_SSE2/3SLOW, which can be checked for particular SSE2/3 functions that
have been checked specifically on such CPUs and are actually faster than
their MMX counterparts.
In addition, use this flag to enable particular VP8 and LPC SSE2 functions
that are faster than their MMX counterparts.
Based on a patch by Loren Merritt <lorenm AT u washington edu>.
author | rbultje |
---|---|
date | Mon, 19 Jul 2010 22:38:23 +0000 |
parents | 7dd2a45249a9 |
children |
line wrap: on
line source
/* * KMVC decoder * Copyright (c) 2006 Konstantin Shishkov * * This file is part of FFmpeg. * * FFmpeg 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. * * FFmpeg 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 FFmpeg; if not, write to the Free Software * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA */ /** * @file * Karl Morton's Video Codec decoder */ #include <stdio.h> #include <stdlib.h> #include "avcodec.h" #include "bytestream.h" #define KMVC_KEYFRAME 0x80 #define KMVC_PALETTE 0x40 #define KMVC_METHOD 0x0F /* * Decoder context */ typedef struct KmvcContext { AVCodecContext *avctx; AVFrame pic; int setpal; int palsize; uint32_t pal[256]; uint8_t *cur, *prev; uint8_t *frm0, *frm1; } KmvcContext; typedef struct BitBuf { int bits; int bitbuf; } BitBuf; #define BLK(data, x, y) data[(x) + (y) * 320] #define kmvc_init_getbits(bb, src) bb.bits = 7; bb.bitbuf = *src++; #define kmvc_getbit(bb, src, res) {\ res = 0; \ if (bb.bitbuf & (1 << bb.bits)) res = 1; \ bb.bits--; \ if(bb.bits == -1) { \ bb.bitbuf = *src++; \ bb.bits = 7; \ } \ } static void kmvc_decode_intra_8x8(KmvcContext * ctx, const uint8_t * src, int w, int h) { BitBuf bb; int res, val; int i, j; int bx, by; int l0x, l1x, l0y, l1y; int mx, my; kmvc_init_getbits(bb, src); for (by = 0; by < h; by += 8) for (bx = 0; bx < w; bx += 8) { kmvc_getbit(bb, src, res); if (!res) { // fill whole 8x8 block val = *src++; for (i = 0; i < 64; i++) BLK(ctx->cur, bx + (i & 0x7), by + (i >> 3)) = val; } else { // handle four 4x4 subblocks for (i = 0; i < 4; i++) { l0x = bx + (i & 1) * 4; l0y = by + (i & 2) * 2; kmvc_getbit(bb, src, res); if (!res) { kmvc_getbit(bb, src, res); if (!res) { // fill whole 4x4 block val = *src++; for (j = 0; j < 16; j++) BLK(ctx->cur, l0x + (j & 3), l0y + (j >> 2)) = val; } else { // copy block from already decoded place val = *src++; mx = val & 0xF; my = val >> 4; for (j = 0; j < 16; j++) BLK(ctx->cur, l0x + (j & 3), l0y + (j >> 2)) = BLK(ctx->cur, l0x + (j & 3) - mx, l0y + (j >> 2) - my); } } else { // descend to 2x2 sub-sub-blocks for (j = 0; j < 4; j++) { l1x = l0x + (j & 1) * 2; l1y = l0y + (j & 2); kmvc_getbit(bb, src, res); if (!res) { kmvc_getbit(bb, src, res); if (!res) { // fill whole 2x2 block val = *src++; BLK(ctx->cur, l1x, l1y) = val; BLK(ctx->cur, l1x + 1, l1y) = val; BLK(ctx->cur, l1x, l1y + 1) = val; BLK(ctx->cur, l1x + 1, l1y + 1) = val; } else { // copy block from already decoded place val = *src++; mx = val & 0xF; my = val >> 4; BLK(ctx->cur, l1x, l1y) = BLK(ctx->cur, l1x - mx, l1y - my); BLK(ctx->cur, l1x + 1, l1y) = BLK(ctx->cur, l1x + 1 - mx, l1y - my); BLK(ctx->cur, l1x, l1y + 1) = BLK(ctx->cur, l1x - mx, l1y + 1 - my); BLK(ctx->cur, l1x + 1, l1y + 1) = BLK(ctx->cur, l1x + 1 - mx, l1y + 1 - my); } } else { // read values for block BLK(ctx->cur, l1x, l1y) = *src++; BLK(ctx->cur, l1x + 1, l1y) = *src++; BLK(ctx->cur, l1x, l1y + 1) = *src++; BLK(ctx->cur, l1x + 1, l1y + 1) = *src++; } } } } } } } static void kmvc_decode_inter_8x8(KmvcContext * ctx, const uint8_t * src, int w, int h) { BitBuf bb; int res, val; int i, j; int bx, by; int l0x, l1x, l0y, l1y; int mx, my; kmvc_init_getbits(bb, src); for (by = 0; by < h; by += 8) for (bx = 0; bx < w; bx += 8) { kmvc_getbit(bb, src, res); if (!res) { kmvc_getbit(bb, src, res); if (!res) { // fill whole 8x8 block val = *src++; for (i = 0; i < 64; i++) BLK(ctx->cur, bx + (i & 0x7), by + (i >> 3)) = val; } else { // copy block from previous frame for (i = 0; i < 64; i++) BLK(ctx->cur, bx + (i & 0x7), by + (i >> 3)) = BLK(ctx->prev, bx + (i & 0x7), by + (i >> 3)); } } else { // handle four 4x4 subblocks for (i = 0; i < 4; i++) { l0x = bx + (i & 1) * 4; l0y = by + (i & 2) * 2; kmvc_getbit(bb, src, res); if (!res) { kmvc_getbit(bb, src, res); if (!res) { // fill whole 4x4 block val = *src++; for (j = 0; j < 16; j++) BLK(ctx->cur, l0x + (j & 3), l0y + (j >> 2)) = val; } else { // copy block val = *src++; mx = (val & 0xF) - 8; my = (val >> 4) - 8; for (j = 0; j < 16; j++) BLK(ctx->cur, l0x + (j & 3), l0y + (j >> 2)) = BLK(ctx->prev, l0x + (j & 3) + mx, l0y + (j >> 2) + my); } } else { // descend to 2x2 sub-sub-blocks for (j = 0; j < 4; j++) { l1x = l0x + (j & 1) * 2; l1y = l0y + (j & 2); kmvc_getbit(bb, src, res); if (!res) { kmvc_getbit(bb, src, res); if (!res) { // fill whole 2x2 block val = *src++; BLK(ctx->cur, l1x, l1y) = val; BLK(ctx->cur, l1x + 1, l1y) = val; BLK(ctx->cur, l1x, l1y + 1) = val; BLK(ctx->cur, l1x + 1, l1y + 1) = val; } else { // copy block val = *src++; mx = (val & 0xF) - 8; my = (val >> 4) - 8; BLK(ctx->cur, l1x, l1y) = BLK(ctx->prev, l1x + mx, l1y + my); BLK(ctx->cur, l1x + 1, l1y) = BLK(ctx->prev, l1x + 1 + mx, l1y + my); BLK(ctx->cur, l1x, l1y + 1) = BLK(ctx->prev, l1x + mx, l1y + 1 + my); BLK(ctx->cur, l1x + 1, l1y + 1) = BLK(ctx->prev, l1x + 1 + mx, l1y + 1 + my); } } else { // read values for block BLK(ctx->cur, l1x, l1y) = *src++; BLK(ctx->cur, l1x + 1, l1y) = *src++; BLK(ctx->cur, l1x, l1y + 1) = *src++; BLK(ctx->cur, l1x + 1, l1y + 1) = *src++; } } } } } } } static int decode_frame(AVCodecContext * avctx, void *data, int *data_size, AVPacket *avpkt) { const uint8_t *buf = avpkt->data; int buf_size = avpkt->size; KmvcContext *const ctx = avctx->priv_data; uint8_t *out, *src; int i; int header; int blocksize; if (ctx->pic.data[0]) avctx->release_buffer(avctx, &ctx->pic); ctx->pic.reference = 1; ctx->pic.buffer_hints = FF_BUFFER_HINTS_VALID; if (avctx->get_buffer(avctx, &ctx->pic) < 0) { av_log(avctx, AV_LOG_ERROR, "get_buffer() failed\n"); return -1; } header = *buf++; /* blocksize 127 is really palette change event */ if (buf[0] == 127) { buf += 3; for (i = 0; i < 127; i++) { ctx->pal[i + (header & 0x81)] = AV_RB24(buf); buf += 4; } buf -= 127 * 4 + 3; } if (header & KMVC_KEYFRAME) { ctx->pic.key_frame = 1; ctx->pic.pict_type = FF_I_TYPE; } else { ctx->pic.key_frame = 0; ctx->pic.pict_type = FF_P_TYPE; } /* if palette has been changed, copy it from palctrl */ if (ctx->avctx->palctrl && ctx->avctx->palctrl->palette_changed) { memcpy(ctx->pal, ctx->avctx->palctrl->palette, AVPALETTE_SIZE); ctx->setpal = 1; ctx->avctx->palctrl->palette_changed = 0; } if (header & KMVC_PALETTE) { ctx->pic.palette_has_changed = 1; // palette starts from index 1 and has 127 entries for (i = 1; i <= ctx->palsize; i++) { ctx->pal[i] = bytestream_get_be24(&buf); } } if (ctx->setpal) { ctx->setpal = 0; ctx->pic.palette_has_changed = 1; } /* make the palette available on the way out */ memcpy(ctx->pic.data[1], ctx->pal, 1024); blocksize = *buf++; if (blocksize != 8 && blocksize != 127) { av_log(avctx, AV_LOG_ERROR, "Block size = %i\n", blocksize); return -1; } memset(ctx->cur, 0, 320 * 200); switch (header & KMVC_METHOD) { case 0: case 1: // used in palette changed event memcpy(ctx->cur, ctx->prev, 320 * 200); break; case 3: kmvc_decode_intra_8x8(ctx, buf, avctx->width, avctx->height); break; case 4: kmvc_decode_inter_8x8(ctx, buf, avctx->width, avctx->height); break; default: av_log(avctx, AV_LOG_ERROR, "Unknown compression method %i\n", header & KMVC_METHOD); return -1; } out = ctx->pic.data[0]; src = ctx->cur; for (i = 0; i < avctx->height; i++) { memcpy(out, src, avctx->width); src += 320; out += ctx->pic.linesize[0]; } /* flip buffers */ if (ctx->cur == ctx->frm0) { ctx->cur = ctx->frm1; ctx->prev = ctx->frm0; } else { ctx->cur = ctx->frm0; ctx->prev = ctx->frm1; } *data_size = sizeof(AVFrame); *(AVFrame *) data = ctx->pic; /* always report that the buffer was completely consumed */ return buf_size; } /* * Init kmvc decoder */ static av_cold int decode_init(AVCodecContext * avctx) { KmvcContext *const c = avctx->priv_data; int i; c->avctx = avctx; if (avctx->width > 320 || avctx->height > 200) { av_log(avctx, AV_LOG_ERROR, "KMVC supports frames <= 320x200\n"); return -1; } c->frm0 = av_mallocz(320 * 200); c->frm1 = av_mallocz(320 * 200); c->cur = c->frm0; c->prev = c->frm1; for (i = 0; i < 256; i++) { c->pal[i] = i * 0x10101; } if (avctx->extradata_size < 12) { av_log(NULL, 0, "Extradata missing, decoding may not work properly...\n"); c->palsize = 127; } else { c->palsize = AV_RL16(avctx->extradata + 10); } if (avctx->extradata_size == 1036) { // palette in extradata uint8_t *src = avctx->extradata + 12; for (i = 0; i < 256; i++) { c->pal[i] = AV_RL32(src); src += 4; } c->setpal = 1; if (c->avctx->palctrl) { c->avctx->palctrl->palette_changed = 0; } } avctx->pix_fmt = PIX_FMT_PAL8; return 0; } /* * Uninit kmvc decoder */ static av_cold int decode_end(AVCodecContext * avctx) { KmvcContext *const c = avctx->priv_data; av_freep(&c->frm0); av_freep(&c->frm1); if (c->pic.data[0]) avctx->release_buffer(avctx, &c->pic); return 0; } AVCodec kmvc_decoder = { "kmvc", AVMEDIA_TYPE_VIDEO, CODEC_ID_KMVC, sizeof(KmvcContext), decode_init, NULL, decode_end, decode_frame, CODEC_CAP_DR1, .long_name = NULL_IF_CONFIG_SMALL("Karl Morton's video codec"), };