From patchwork Wed Mar 15 23:34:43 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Michael Niedermayer X-Patchwork-Id: 40690 Delivered-To: ffmpegpatchwork2@gmail.com Received: by 2002:a05:6a20:d046:b0:cd:afd7:272c with SMTP id hv6csp3840341pzb; Wed, 15 Mar 2023 16:34:58 -0700 (PDT) X-Google-Smtp-Source: AK7set++8AI3v+FmsjhAO8xmJakqLOGyjLFuPM8SODcDvWzyIStGVqFcr+CMn1bqdyrLY21+G5AV X-Received: by 2002:a17:906:fe05:b0:930:1391:da7c with SMTP id wy5-20020a170906fe0500b009301391da7cmr1163472ejb.60.1678923298574; Wed, 15 Mar 2023 16:34:58 -0700 (PDT) ARC-Seal: i=1; a=rsa-sha256; t=1678923298; cv=none; d=google.com; s=arc-20160816; b=JIuoZcWX52ktIeUAFweqCEcb5DYpXO/0lFkivAY94FgYZcDI0guhW89wJa4eIEICbc Qall1c1UTQ9hY1rIoOrU5louAyhnxBLsEZd10J35lJxYrUgaRrYAgDk5Pk0rjBMM1LaJ H17dhnrffGcF4jONecXx1tiib2KkR6ZWinAPPRvxxhjgOO7wuo3RP0QW01H4ucOyVwh4 GwiZ3BdXvzDUoiMO+ne1JJYfgS5L+ZaCBccUoJH9GZJyF14wC5Ll/HMEAFfFCD1SIfOt JhVnAR4Y1DMdqPPVbv3xeL6JYpkhTbC15iX3Ury8eGeq+sGXiHxK4urPsAdOj634yCvH QrIw== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=sender:errors-to:content-transfer-encoding:mime-version:reply-to :list-subscribe:list-help:list-post:list-archive:list-unsubscribe :list-id:precedence:subject:message-id:date:to:from:delivered-to; bh=3nVEyDONm2/8VU9F2mHz4eNQ9PUTqxW6dBZsIg6NwXM=; b=q7RbIT1sN30iS4HAP8gGmmo1OU213P6cINeXkqU8Hf03Zn8lrwJ9MYyXNE63cxmmDc aS5i0UQt/SBijyBwjSzHJT/SvwabjKKVp1ji/9I/cD3ninaVpZejNCetYPAwz9Bu8Ten flBZRKyPiFoiibuB3D7w5BHpxbvmCEoS0S4Ni3akDbQAsTwV7J0g/iPQn6Rtpq63E4em bNeRkytsmKd1NwkWif38cL47jlNMDLKTHPw8DxTa6cXzyitoGyrxZkvCQeyzTeboN8K1 NQdADlsUZoWnfSdPjlGvlKFoCRPz1AKfCvozatOsw/11behqCth98wCbiLpIWs98Vnyq bFyw== ARC-Authentication-Results: i=1; mx.google.com; spf=pass (google.com: domain of ffmpeg-devel-bounces@ffmpeg.org designates 79.124.17.100 as permitted sender) smtp.mailfrom=ffmpeg-devel-bounces@ffmpeg.org Return-Path: Received: from ffbox0-bg.mplayerhq.hu (ffbox0-bg.ffmpeg.org. [79.124.17.100]) by mx.google.com with ESMTP id ga16-20020a1709070c1000b0092ec42d6a30si4361181ejc.407.2023.03.15.16.34.57; Wed, 15 Mar 2023 16:34:58 -0700 (PDT) Received-SPF: pass (google.com: domain of ffmpeg-devel-bounces@ffmpeg.org designates 79.124.17.100 as permitted sender) client-ip=79.124.17.100; Authentication-Results: mx.google.com; spf=pass (google.com: domain of ffmpeg-devel-bounces@ffmpeg.org designates 79.124.17.100 as permitted sender) smtp.mailfrom=ffmpeg-devel-bounces@ffmpeg.org Received: from [127.0.1.1] (localhost [127.0.0.1]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id 24EF868BCE6; Thu, 16 Mar 2023 01:34:54 +0200 (EET) X-Original-To: ffmpeg-devel@ffmpeg.org Delivered-To: ffmpeg-devel@ffmpeg.org Received: from relay4-d.mail.gandi.net (relay4-d.mail.gandi.net [217.70.183.196]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTPS id 61FD468AD1B for ; Thu, 16 Mar 2023 01:34:47 +0200 (EET) Received: (Authenticated sender: michael@niedermayer.cc) by mail.gandi.net (Postfix) with ESMTPSA id 37BADE0008 for ; Wed, 15 Mar 2023 23:34:45 +0000 (UTC) From: Michael Niedermayer To: FFmpeg development discussions and patches Date: Thu, 16 Mar 2023 00:34:43 +0100 Message-Id: <20230315233445.5282-1-michael@niedermayer.cc> X-Mailer: git-send-email 2.17.1 Subject: [FFmpeg-devel] [PATCH 1/3] avfilter/vf_uspp: update to new APIs X-BeenThere: ffmpeg-devel@ffmpeg.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: FFmpeg development discussions and patches List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Reply-To: FFmpeg development discussions and patches MIME-Version: 1.0 Errors-To: ffmpeg-devel-bounces@ffmpeg.org Sender: "ffmpeg-devel" X-TUID: OOSeP/Pi93Bb Signed-off-by: Michael Niedermayer --- configure | 1 - libavfilter/vf_uspp.c | 49 ++++++++++++++++++++++++++++++++++++------- 2 files changed, 41 insertions(+), 9 deletions(-) diff --git a/configure b/configure index 03d3c429a5..0370e25577 100755 --- a/configure +++ b/configure @@ -7359,7 +7359,6 @@ enable frame_thread_encoder # they are kept disabled for now, but will be removed if # nobody updates and re-enables them disable mcdeint_filter -disable uspp_filter enabled asm || { arch=c; disable $ARCH_LIST $ARCH_EXT_LIST; } diff --git a/libavfilter/vf_uspp.c b/libavfilter/vf_uspp.c index 051de00771..43114e1b50 100644 --- a/libavfilter/vf_uspp.c +++ b/libavfilter/vf_uspp.c @@ -53,6 +53,7 @@ typedef struct USPPContext { int outbuf_size; uint8_t *outbuf; AVCodecContext *avctx_enc[BLOCK*BLOCK]; + AVCodecContext *avctx_dec[BLOCK*BLOCK]; AVPacket *pkt; AVFrame *frame; AVFrame *frame_dec; @@ -244,7 +245,6 @@ static void filter(USPPContext *p, uint8_t *dst[3], uint8_t *src[3], const int BLOCKc = BLOCK >> p->hsub; int offset; AVPacket *pkt = p->pkt; - int got_pkt_ptr; av_packet_unref(pkt); pkt->data = p->outbuf; @@ -255,14 +255,28 @@ static void filter(USPPContext *p, uint8_t *dst[3], uint8_t *src[3], p->frame->data[2] = p->src[2] + x1c + y1c * p->frame->linesize[2]; p->frame->format = p->avctx_enc[i]->pix_fmt; - ret = avcodec_encode_video2(p->avctx_enc[i], pkt, p->frame, &got_pkt_ptr); + ret = avcodec_send_frame(p->avctx_enc[i], p->frame); if (ret < 0) { - av_log(p->avctx_enc[i], AV_LOG_ERROR, "Encoding failed\n"); + av_log(p->avctx_enc[i], AV_LOG_ERROR, "Error sending a frame for encoding\n"); + continue; + } + ret = avcodec_receive_packet(p->avctx_enc[i], pkt); + if (ret < 0) { + av_log(p->avctx_enc[i], AV_LOG_ERROR, "Error receiving a packet from encoding\n"); continue; } - av_packet_unref(pkt); - p->frame_dec = p->avctx_enc[i]->coded_frame; + ret = avcodec_send_packet(p->avctx_dec[i], pkt); + av_packet_unref(pkt); + if (ret < 0) { + av_log(p->avctx_dec[i], AV_LOG_ERROR, "Error sending a packet for decoding\n"); + continue; + } + ret = avcodec_receive_frame(p->avctx_dec[i], p->frame_dec); + if (ret < 0) { + av_log(p->avctx_dec[i], AV_LOG_ERROR, "Error receiving a frame from decoding\n"); + continue; + } offset = (BLOCK-x1) + (BLOCK-y1) * p->frame_dec->linesize[0]; @@ -315,10 +329,15 @@ static int config_input(AVFilterLink *inlink) int i; const AVCodec *enc = avcodec_find_encoder(AV_CODEC_ID_SNOW); + const AVCodec *dec = avcodec_find_decoder(AV_CODEC_ID_SNOW); if (!enc) { av_log(ctx, AV_LOG_ERROR, "SNOW encoder not found.\n"); return AVERROR(EINVAL); } + if (!dec) { + av_log(ctx, AV_LOG_ERROR, "SNOW decoder not found.\n"); + return AVERROR(EINVAL); + } uspp->hsub = desc->log2_chroma_w; uspp->vsub = desc->log2_chroma_h; @@ -341,15 +360,20 @@ static int config_input(AVFilterLink *inlink) } for (i = 0; i < (1<log2_count); i++) { - AVCodecContext *avctx_enc; + AVCodecContext *avctx_enc, *avctx_dec; AVDictionary *opts = NULL; int ret; if (!(uspp->avctx_enc[i] = avcodec_alloc_context3(NULL))) return AVERROR(ENOMEM); + if (!(uspp->avctx_dec[i] = avcodec_alloc_context3(NULL))) + return AVERROR(ENOMEM); avctx_enc = uspp->avctx_enc[i]; + avctx_dec = uspp->avctx_dec[i]; + avctx_dec->width = avctx_enc->width = width + BLOCK; + avctx_dec->height = avctx_enc->height = height + BLOCK; avctx_enc->time_base = (AVRational){1,25}; // meaningless avctx_enc->gop_size = INT_MAX; @@ -358,17 +382,24 @@ static int config_input(AVFilterLink *inlink) avctx_enc->flags = AV_CODEC_FLAG_QSCALE | AV_CODEC_FLAG_LOW_DELAY; avctx_enc->strict_std_compliance = FF_COMPLIANCE_EXPERIMENTAL; avctx_enc->global_quality = 123; - av_dict_set(&opts, "no_bitstream", "1", 0); ret = avcodec_open2(avctx_enc, enc, &opts); av_dict_free(&opts); if (ret < 0) return ret; av_assert0(avctx_enc->codec); + + + ret = avcodec_open2(avctx_dec, dec, NULL); + if (ret < 0) + return ret; + } uspp->outbuf_size = (width + BLOCK) * (height + BLOCK) * 10; if (!(uspp->frame = av_frame_alloc())) return AVERROR(ENOMEM); + if (!(uspp->frame_dec = av_frame_alloc())) + return AVERROR(ENOMEM); if (!(uspp->pkt = av_packet_alloc())) return AVERROR(ENOMEM); if (!(uspp->outbuf = av_malloc(uspp->outbuf_size))) @@ -460,8 +491,10 @@ static av_cold void uninit(AVFilterContext *ctx) av_freep(&uspp->src[i]); } - for (i = 0; i < (1 << uspp->log2_count); i++) + for (i = 0; i < (1 << uspp->log2_count); i++) { avcodec_free_context(&uspp->avctx_enc[i]); + avcodec_free_context(&uspp->avctx_dec[i]); + } av_freep(&uspp->non_b_qp_table); av_freep(&uspp->outbuf); From patchwork Wed Mar 15 23:34:44 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Michael Niedermayer X-Patchwork-Id: 40691 Delivered-To: ffmpegpatchwork2@gmail.com Received: by 2002:a05:6a20:d046:b0:cd:afd7:272c with SMTP id hv6csp3840413pzb; Wed, 15 Mar 2023 16:35:09 -0700 (PDT) X-Google-Smtp-Source: AK7set+M5yyCXJsvQCOT1LnDV27LQsgeDFmyE8RP5t91WE6V5KpSX7pg6RHRD2KTuXBjM0HuYNFH X-Received: by 2002:a17:907:c386:b0:92f:a00c:ee52 with SMTP id tm6-20020a170907c38600b0092fa00cee52mr2686744ejc.18.1678923308882; Wed, 15 Mar 2023 16:35:08 -0700 (PDT) ARC-Seal: i=1; a=rsa-sha256; t=1678923308; cv=none; d=google.com; s=arc-20160816; b=B6K97tfqhyCnMnbEB/k4Qt7Roe0X0Y1KwMp+Rqo7CSffuuNyveIBZGM2ozP2KGhZnZ Kt6b6SQq1A5f/RWo+PpPnBQtrkmiHH7v2BSNfYSE+hJE2VIqyFIG4eJmLZR84XULT/MR VrfMUvkNPrJOK/89GBO0PcAXYiWIup8rrIbpL45YoZYdRvp1vk+0dKIUcMeNmP/qx2QU t6Fv1Jvhh4ER34r//7lDh+6lMShnry/mhyL1e0FsDyms7T5brU8NvRV9GqzTIA1keXa2 KSA2YmZ8ueZGhT0Knm7k2u+DRzAT9hvT8Kcmjv5v6B0ZGRqRZeeW7Ms/sIRkFq3J5Olt a1lg== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=sender:errors-to:content-transfer-encoding:mime-version:reply-to :list-subscribe:list-help:list-post:list-archive:list-unsubscribe :list-id:precedence:subject:references:in-reply-to:message-id:date :to:from:delivered-to; bh=j9/pRAiK41YYAyDCge0X0tB3lURIEqoKX8NoDP3xA2o=; b=vohJ3rSSZ2nvdyk4BbZX+6AvQ0nlEm5wOBPbNnfvwxuihK6HT36LBoaRFeS9JnT9jn 1HeTYKw+Ik1liahSMGbCYMbAiojb/LalOGXfVikvrtay7JXNUWSAriTgt+rfeq4JLp9+ Edb88aoEzJ8HkAdYt7/rlTADFqxPyMBgvnIRWzMK4ePV55KoTljtTuP4sk1h6oaB8fkN lwRDHX7x8BjFM0rf6nxEJ2Gz/x3bB7nSOqlPJQ5wOouBOstRG915YEFaQOeRihEaLJBa Y+pqrWSIeYPdGPKjAx10FECNxrb13hhLKDl7IMaNeD5y/cnn9XC9BS0lzNmZyEbOtM2j km0Q== ARC-Authentication-Results: i=1; mx.google.com; spf=pass (google.com: domain of ffmpeg-devel-bounces@ffmpeg.org designates 79.124.17.100 as permitted sender) smtp.mailfrom=ffmpeg-devel-bounces@ffmpeg.org Return-Path: Received: from ffbox0-bg.mplayerhq.hu (ffbox0-bg.ffmpeg.org. [79.124.17.100]) by mx.google.com with ESMTP id v21-20020a170906489500b008d490a104b2si379406ejq.523.2023.03.15.16.35.08; Wed, 15 Mar 2023 16:35:08 -0700 (PDT) Received-SPF: pass (google.com: domain of ffmpeg-devel-bounces@ffmpeg.org designates 79.124.17.100 as permitted sender) client-ip=79.124.17.100; Authentication-Results: mx.google.com; spf=pass (google.com: domain of ffmpeg-devel-bounces@ffmpeg.org designates 79.124.17.100 as permitted sender) smtp.mailfrom=ffmpeg-devel-bounces@ffmpeg.org Received: from [127.0.1.1] (localhost [127.0.0.1]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id 68F9F689E37; Thu, 16 Mar 2023 01:34:56 +0200 (EET) X-Original-To: ffmpeg-devel@ffmpeg.org Delivered-To: ffmpeg-devel@ffmpeg.org Received: from relay5-d.mail.gandi.net (relay5-d.mail.gandi.net [217.70.183.197]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTPS id 1286068AD1B for ; Thu, 16 Mar 2023 01:34:48 +0200 (EET) Received: (Authenticated sender: michael@niedermayer.cc) by mail.gandi.net (Postfix) with ESMTPSA id 59A051C0003 for ; Wed, 15 Mar 2023 23:34:46 +0000 (UTC) From: Michael Niedermayer To: FFmpeg development discussions and patches Date: Thu, 16 Mar 2023 00:34:44 +0100 Message-Id: <20230315233445.5282-2-michael@niedermayer.cc> X-Mailer: git-send-email 2.17.1 In-Reply-To: <20230315233445.5282-1-michael@niedermayer.cc> References: <20230315233445.5282-1-michael@niedermayer.cc> Subject: [FFmpeg-devel] [PATCH 2/3] avfilter/vf_uspp: Support any codec X-BeenThere: ffmpeg-devel@ffmpeg.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: FFmpeg development discussions and patches List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Reply-To: FFmpeg development discussions and patches MIME-Version: 1.0 Errors-To: ffmpeg-devel-bounces@ffmpeg.org Sender: "ffmpeg-devel" X-TUID: 3CrD7LeBlSRA Signed-off-by: Michael Niedermayer --- doc/filters.texi | 3 +++ libavfilter/vf_uspp.c | 11 ++++++----- 2 files changed, 9 insertions(+), 5 deletions(-) diff --git a/doc/filters.texi b/doc/filters.texi index 77b594f69c..d634924bfb 100644 --- a/doc/filters.texi +++ b/doc/filters.texi @@ -23364,6 +23364,9 @@ that value the speed drops by a factor of approximately 2. Default value is @item qp Force a constant quantization parameter. If not set, the filter will use the QP from the video stream (if available). + +@item codec +Use specified codec instead of snow. @end table @section v360 diff --git a/libavfilter/vf_uspp.c b/libavfilter/vf_uspp.c index 43114e1b50..a7bf8e3087 100644 --- a/libavfilter/vf_uspp.c +++ b/libavfilter/vf_uspp.c @@ -46,6 +46,7 @@ typedef struct USPPContext { int log2_count; int hsub, vsub; int qp; + char *codec_name; enum AVVideoEncParamsType qscale_type; int temp_stride[3]; uint8_t *src[3]; @@ -68,6 +69,7 @@ static const AVOption uspp_options[] = { { "quality", "set quality", OFFSET(log2_count), AV_OPT_TYPE_INT, {.i64 = 3}, 0, MAX_LEVEL, FLAGS }, { "qp", "force a constant quantizer parameter", OFFSET(qp), AV_OPT_TYPE_INT, {.i64 = 0}, 0, 63, FLAGS }, { "use_bframe_qp", "use B-frames' QP", OFFSET(use_bframe_qp), AV_OPT_TYPE_BOOL,{.i64 = 0}, 0, 1, FLAGS }, + { "codec", "Codec name", OFFSET(codec_name), AV_OPT_TYPE_STRING, {.str = "snow"}, 0, 0, FLAGS }, { NULL } }; @@ -327,15 +329,14 @@ static int config_input(AVFilterLink *inlink) const int width = inlink->w; const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(inlink->format); int i; - - const AVCodec *enc = avcodec_find_encoder(AV_CODEC_ID_SNOW); - const AVCodec *dec = avcodec_find_decoder(AV_CODEC_ID_SNOW); + const AVCodec *enc = avcodec_find_encoder_by_name(uspp->codec_name); + const AVCodec *dec = avcodec_find_decoder_by_name(uspp->codec_name); if (!enc) { - av_log(ctx, AV_LOG_ERROR, "SNOW encoder not found.\n"); + av_log(ctx, AV_LOG_ERROR, "encoder %s not found.\n", uspp->codec_name); return AVERROR(EINVAL); } if (!dec) { - av_log(ctx, AV_LOG_ERROR, "SNOW decoder not found.\n"); + av_log(ctx, AV_LOG_ERROR, "decoder %s not found.\n", uspp->codec_name); return AVERROR(EINVAL); } From patchwork Wed Mar 15 23:34:45 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Michael Niedermayer X-Patchwork-Id: 40692 Delivered-To: ffmpegpatchwork2@gmail.com Received: by 2002:a05:6a20:d046:b0:cd:afd7:272c with SMTP id hv6csp3840493pzb; Wed, 15 Mar 2023 16:35:18 -0700 (PDT) X-Google-Smtp-Source: AK7set+ev9Bm7uf5JELBdcXKS5rg7L2bos+6QnOhBsKx7ErUThOXQoEhpjBwpTDgHaCb2Edeta2Y X-Received: by 2002:aa7:cc8e:0:b0:4fa:39a6:af25 with SMTP id p14-20020aa7cc8e000000b004fa39a6af25mr5051257edt.16.1678923318673; Wed, 15 Mar 2023 16:35:18 -0700 (PDT) ARC-Seal: i=1; a=rsa-sha256; t=1678923318; cv=none; d=google.com; s=arc-20160816; b=A39OyIq8TdrlOldfAqrNbf03YkCXA5meYvPVHs2Kt8Iu02fRmm1GAZZeu+tgW5dg1U Ik/I2B636otz2w2GTUWdpmtVJ2wMJQF77WvnIwB6x3pABwX3lrwOABznkK7q2ujf8Xex W40rlCd7+yOUBpxI0fu1EDkDs4kHVv+oa3rCvQCXRb1sUsTsCtfKA+nInwa8RVd7ORB6 YvsQw0ckEd+ChDlAdvNjRvjOW50GFB8Rjgocb+jwqfx3aWDIoLtVmkMvg4ONxgI82gKA 10MyP3RWAcD0NZ/ESWGfJYGYALy0XDkxwCpZC/G2+UN5ANkoU/vDpkm7jaWFyt+g9AQ7 5XAA== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=sender:errors-to:content-transfer-encoding:mime-version:reply-to :list-subscribe:list-help:list-post:list-archive:list-unsubscribe :list-id:precedence:subject:references:in-reply-to:message-id:date :to:from:delivered-to; bh=qOKTPus5LJ/4PERPEh0QpeDpStbLqVxtYQ+laZn9p+0=; b=GeSEXWvkbJhX8rfNoLyPa9d/AVFRXP4PZdfoivu3iVJqx4b6qJBnJi5W/h/q2m8h+5 PCv6ehPpPwfdQcrHW76UBv0KhH6yUHGA9iIyaMWpk/4qNQpWRq5vJfFXfBmtuu0IRyZX AaR8UDWPxXp8FnmQpOomReskhLhEWTzdvg3Nav308LFwEkBHJOMjpF0Jnr5HFTnbWjLm Riyu1VJaaBke5EI29hzCD89K1o+X4KQB4TXk88UnZPAtYmFxWzE70gz2iYvGz3nLyO5S D7BTc2jINJ54ZmqjQUVFzKK7qsEppDIdH/C+NviVe2Yagm8cChrzlxhu5ExXGErrk8Cz 9UZg== ARC-Authentication-Results: i=1; mx.google.com; spf=pass (google.com: domain of ffmpeg-devel-bounces@ffmpeg.org designates 79.124.17.100 as permitted sender) smtp.mailfrom=ffmpeg-devel-bounces@ffmpeg.org Return-Path: Received: from ffbox0-bg.mplayerhq.hu (ffbox0-bg.ffmpeg.org. [79.124.17.100]) by mx.google.com with ESMTP id q22-20020aa7d456000000b004acc68d5d98si6774994edr.489.2023.03.15.16.35.18; Wed, 15 Mar 2023 16:35:18 -0700 (PDT) Received-SPF: pass (google.com: domain of ffmpeg-devel-bounces@ffmpeg.org designates 79.124.17.100 as permitted sender) client-ip=79.124.17.100; Authentication-Results: mx.google.com; spf=pass (google.com: domain of ffmpeg-devel-bounces@ffmpeg.org designates 79.124.17.100 as permitted sender) smtp.mailfrom=ffmpeg-devel-bounces@ffmpeg.org Received: from [127.0.1.1] (localhost [127.0.0.1]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id 7BFB268BD67; Thu, 16 Mar 2023 01:34:57 +0200 (EET) X-Original-To: ffmpeg-devel@ffmpeg.org Delivered-To: ffmpeg-devel@ffmpeg.org Received: from relay8-d.mail.gandi.net (relay8-d.mail.gandi.net [217.70.183.201]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTPS id 445C968BCE8 for ; Thu, 16 Mar 2023 01:34:49 +0200 (EET) Received: (Authenticated sender: michael@niedermayer.cc) by mail.gandi.net (Postfix) with ESMTPSA id 6296B1BF207 for ; Wed, 15 Mar 2023 23:34:48 +0000 (UTC) From: Michael Niedermayer To: FFmpeg development discussions and patches Date: Thu, 16 Mar 2023 00:34:45 +0100 Message-Id: <20230315233445.5282-3-michael@niedermayer.cc> X-Mailer: git-send-email 2.17.1 In-Reply-To: <20230315233445.5282-1-michael@niedermayer.cc> References: <20230315233445.5282-1-michael@niedermayer.cc> Subject: [FFmpeg-devel] [PATCH 3/3] avfilter/vf_uspp: 1000% faster with threads X-BeenThere: ffmpeg-devel@ffmpeg.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: FFmpeg development discussions and patches List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Reply-To: FFmpeg development discussions and patches MIME-Version: 1.0 Errors-To: ffmpeg-devel-bounces@ffmpeg.org Sender: "ffmpeg-devel" X-TUID: Teuhz2aZt2TL Signed-off-by: Michael Niedermayer --- libavfilter/vf_uspp.c | 181 +++++++++++++++++++++++------------------- 1 file changed, 99 insertions(+), 82 deletions(-) diff --git a/libavfilter/vf_uspp.c b/libavfilter/vf_uspp.c index a7bf8e3087..0a992df898 100644 --- a/libavfilter/vf_uspp.c +++ b/libavfilter/vf_uspp.c @@ -44,6 +44,7 @@ typedef struct USPPContext { const AVClass *av_class; int log2_count; + int count; int hsub, vsub; int qp; char *codec_name; @@ -55,12 +56,13 @@ typedef struct USPPContext { uint8_t *outbuf; AVCodecContext *avctx_enc[BLOCK*BLOCK]; AVCodecContext *avctx_dec[BLOCK*BLOCK]; - AVPacket *pkt; - AVFrame *frame; - AVFrame *frame_dec; + AVPacket *pkt [BLOCK*BLOCK]; + AVFrame *frame [BLOCK*BLOCK]; + AVFrame *frame_dec [BLOCK*BLOCK]; int8_t *non_b_qp_table; int non_b_qp_stride; int use_bframe_qp; + int quality; } USPPContext; #define OFFSET(x) offsetof(USPPContext, x) @@ -188,13 +190,87 @@ static void store_slice_c(uint8_t *dst, const uint16_t *src, } } -static void filter(USPPContext *p, uint8_t *dst[3], uint8_t *src[3], +static int filter_1phase(AVFilterContext *ctx, void *arg, int i, int nb_jobs) +{ + USPPContext *p = ctx->priv; + int ret, x, y; + int width = ctx->inputs[0]->w; + int height = ctx->inputs[0]->h; + + const int x1 = offset[i+nb_jobs-1][0]; + const int y1 = offset[i+nb_jobs-1][1]; + const int x1c = x1 >> p->hsub; + const int y1c = y1 >> p->vsub; + const int BLOCKc = BLOCK >> p->hsub; + int offset; + AVPacket *pkt = p->pkt[i]; + + av_packet_unref(pkt); + pkt->data = p->outbuf; + pkt->size = p->outbuf_size; + + p->frame[i]->linesize[0] = p->temp_stride[0]; + p->frame[i]->linesize[1] = p->temp_stride[1]; + p->frame[i]->linesize[2] = p->temp_stride[2]; + p->frame[i]->height = height + BLOCK; + p->frame[i]->width = width + BLOCK; + p->frame[i]->data[0] = p->src[0] + x1 + y1 * p->frame[i]->linesize[0]; + p->frame[i]->data[1] = p->src[1] + x1c + y1c * p->frame[i]->linesize[1]; + p->frame[i]->data[2] = p->src[2] + x1c + y1c * p->frame[i]->linesize[2]; + p->frame[i]->format = p->avctx_enc[i]->pix_fmt; + p->frame[i]->quality = p->quality; + + ret = avcodec_send_frame(p->avctx_enc[i], p->frame[i]); + if (ret < 0) { + av_log(p->avctx_enc[i], AV_LOG_ERROR, "Error sending a frame for encoding\n"); + return ret; + } + ret = avcodec_receive_packet(p->avctx_enc[i], pkt); + if (ret < 0) { + av_log(p->avctx_enc[i], AV_LOG_ERROR, "Error receiving a packet from encoding\n"); + return ret; + } + + ret = avcodec_send_packet(p->avctx_dec[i], pkt); + av_packet_unref(pkt); + if (ret < 0) { + av_log(p->avctx_dec[i], AV_LOG_ERROR, "Error sending a packet for decoding\n"); + return ret; + } + ret = avcodec_receive_frame(p->avctx_dec[i], p->frame_dec[i]); + if (ret < 0) { + av_log(p->avctx_dec[i], AV_LOG_ERROR, "Error receiving a frame from decoding\n"); + return ret; + } + + offset = (BLOCK-x1) + (BLOCK-y1) * p->frame_dec[i]->linesize[0]; + + for (y = 0; y < height; y++) + for (x = 0; x < width; x++) + p->temp[0][x + y * p->temp_stride[0]] += p->frame_dec[i]->data[0][x + y * p->frame_dec[i]->linesize[0] + offset]; + + + if (!p->frame_dec[i]->data[2] || !p->temp[2]) + return 0; + + offset = (BLOCKc-x1c) + (BLOCKc-y1c) * p->frame_dec[i]->linesize[1]; + + for (y = 0; y < AV_CEIL_RSHIFT(height, p->vsub); y++) { + for (x = 0; x < AV_CEIL_RSHIFT(width, p->hsub); x++) { + p->temp[1][x + y * p->temp_stride[1]] += p->frame_dec[i]->data[1][x + y * p->frame_dec[i]->linesize[1] + offset]; + p->temp[2][x + y * p->temp_stride[2]] += p->frame_dec[i]->data[2][x + y * p->frame_dec[i]->linesize[2] + offset]; + } + } + + return 0; +} + +static void filter(AVFilterContext *ctx, uint8_t *dst[3], uint8_t *src[3], int dst_stride[3], int src_stride[3], int width, int height, uint8_t *qp_store, int qp_stride) { + USPPContext *p = ctx->priv; int x, y, i, j; - const int count = 1<log2_count; - int ret; for (i = 0; i < 3; i++) { int is_chroma = !!i; @@ -219,12 +295,11 @@ static void filter(USPPContext *p, uint8_t *dst[3], uint8_t *src[3], memcpy(p->src[i] + (h+block +y) * stride, p->src[i] + (h-y+block-1) * stride, stride); } - p->frame->linesize[i] = stride; memset(p->temp[i], 0, (h + 2 * block) * stride * sizeof(int16_t)); } if (p->qp) - p->frame->quality = p->qp * FF_QP2LAMBDA; + p->quality = p->qp * FF_QP2LAMBDA; else { int qpsum=0; int qpcount = (height>>4) * (height>>4); @@ -233,71 +308,11 @@ static void filter(USPPContext *p, uint8_t *dst[3], uint8_t *src[3], for (x = 0; x < (width>>4); x++) qpsum += qp_store[x + y * qp_stride]; } - p->frame->quality = ff_norm_qscale((qpsum + qpcount/2) / qpcount, p->qscale_type) * FF_QP2LAMBDA; + p->quality = ff_norm_qscale((qpsum + qpcount/2) / qpcount, p->qscale_type) * FF_QP2LAMBDA; } // init per MB qscale stuff FIXME - p->frame->height = height + BLOCK; - p->frame->width = width + BLOCK; - - for (i = 0; i < count; i++) { - const int x1 = offset[i+count-1][0]; - const int y1 = offset[i+count-1][1]; - const int x1c = x1 >> p->hsub; - const int y1c = y1 >> p->vsub; - const int BLOCKc = BLOCK >> p->hsub; - int offset; - AVPacket *pkt = p->pkt; - - av_packet_unref(pkt); - pkt->data = p->outbuf; - pkt->size = p->outbuf_size; - - p->frame->data[0] = p->src[0] + x1 + y1 * p->frame->linesize[0]; - p->frame->data[1] = p->src[1] + x1c + y1c * p->frame->linesize[1]; - p->frame->data[2] = p->src[2] + x1c + y1c * p->frame->linesize[2]; - p->frame->format = p->avctx_enc[i]->pix_fmt; - - ret = avcodec_send_frame(p->avctx_enc[i], p->frame); - if (ret < 0) { - av_log(p->avctx_enc[i], AV_LOG_ERROR, "Error sending a frame for encoding\n"); - continue; - } - ret = avcodec_receive_packet(p->avctx_enc[i], pkt); - if (ret < 0) { - av_log(p->avctx_enc[i], AV_LOG_ERROR, "Error receiving a packet from encoding\n"); - continue; - } - - ret = avcodec_send_packet(p->avctx_dec[i], pkt); - av_packet_unref(pkt); - if (ret < 0) { - av_log(p->avctx_dec[i], AV_LOG_ERROR, "Error sending a packet for decoding\n"); - continue; - } - ret = avcodec_receive_frame(p->avctx_dec[i], p->frame_dec); - if (ret < 0) { - av_log(p->avctx_dec[i], AV_LOG_ERROR, "Error receiving a frame from decoding\n"); - continue; - } - - offset = (BLOCK-x1) + (BLOCK-y1) * p->frame_dec->linesize[0]; - - for (y = 0; y < height; y++) - for (x = 0; x < width; x++) - p->temp[0][x + y * p->temp_stride[0]] += p->frame_dec->data[0][x + y * p->frame_dec->linesize[0] + offset]; - - if (!src[2] || !dst[2]) - continue; - - offset = (BLOCKc-x1c) + (BLOCKc-y1c) * p->frame_dec->linesize[1]; - for (y = 0; y < AV_CEIL_RSHIFT(height, p->vsub); y++) { - for (x = 0; x < AV_CEIL_RSHIFT(width, p->hsub); x++) { - p->temp[1][x + y * p->temp_stride[1]] += p->frame_dec->data[1][x + y * p->frame_dec->linesize[1] + offset]; - p->temp[2][x + y * p->temp_stride[2]] += p->frame_dec->data[2][x + y * p->frame_dec->linesize[2] + offset]; - } - } - } + ff_filter_execute(ctx, filter_1phase, NULL, NULL, p->count); for (j = 0; j < 3; j++) { int is_chroma = !!j; @@ -342,6 +357,7 @@ static int config_input(AVFilterLink *inlink) uspp->hsub = desc->log2_chroma_w; uspp->vsub = desc->log2_chroma_h; + uspp->count = 1<log2_count; for (i = 0; i < 3; i++) { int is_chroma = !!i; @@ -360,7 +376,7 @@ static int config_input(AVFilterLink *inlink) return AVERROR(ENOMEM); } - for (i = 0; i < (1<log2_count); i++) { + for (i = 0; i < uspp->count; i++) { AVCodecContext *avctx_enc, *avctx_dec; AVDictionary *opts = NULL; int ret; @@ -394,15 +410,15 @@ static int config_input(AVFilterLink *inlink) if (ret < 0) return ret; + if (!(uspp->frame[i] = av_frame_alloc())) + return AVERROR(ENOMEM); + if (!(uspp->frame_dec[i] = av_frame_alloc())) + return AVERROR(ENOMEM); + if (!(uspp->pkt[i] = av_packet_alloc())) + return AVERROR(ENOMEM); } uspp->outbuf_size = (width + BLOCK) * (height + BLOCK) * 10; - if (!(uspp->frame = av_frame_alloc())) - return AVERROR(ENOMEM); - if (!(uspp->frame_dec = av_frame_alloc())) - return AVERROR(ENOMEM); - if (!(uspp->pkt = av_packet_alloc())) - return AVERROR(ENOMEM); if (!(uspp->outbuf = av_malloc(uspp->outbuf_size))) return AVERROR(ENOMEM); @@ -464,7 +480,7 @@ static int filter_frame(AVFilterLink *inlink, AVFrame *in) out->height = in->height; } - filter(uspp, out->data, in->data, out->linesize, in->linesize, + filter(ctx, out->data, in->data, out->linesize, in->linesize, inlink->w, inlink->h, qp_table, qp_stride); } } @@ -492,15 +508,16 @@ static av_cold void uninit(AVFilterContext *ctx) av_freep(&uspp->src[i]); } - for (i = 0; i < (1 << uspp->log2_count); i++) { + for (i = 0; i < uspp->count; i++) { avcodec_free_context(&uspp->avctx_enc[i]); avcodec_free_context(&uspp->avctx_dec[i]); + av_frame_free(&uspp->frame[i]); + av_frame_free(&uspp->frame_dec[i]); + av_packet_free(&uspp->pkt[i]); } av_freep(&uspp->non_b_qp_table); av_freep(&uspp->outbuf); - av_packet_free(&uspp->pkt); - av_frame_free(&uspp->frame); } static const AVFilterPad uspp_inputs[] = { @@ -528,5 +545,5 @@ const AVFilter ff_vf_uspp = { FILTER_OUTPUTS(uspp_outputs), FILTER_PIXFMTS_ARRAY(pix_fmts), .priv_class = &uspp_class, - .flags = AVFILTER_FLAG_SUPPORT_TIMELINE_INTERNAL, + .flags = AVFILTER_FLAG_SUPPORT_TIMELINE_INTERNAL | AVFILTER_FLAG_SLICE_THREADS, };