From patchwork Fri Sep 23 23:14:59 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Lynne X-Patchwork-Id: 38200 Delivered-To: ffmpegpatchwork2@gmail.com Received: by 2002:a05:6a20:3b1c:b0:96:9ee8:5cfd with SMTP id c28csp790483pzh; Fri, 23 Sep 2022 16:15:35 -0700 (PDT) X-Google-Smtp-Source: AMsMyM7yMf+ZsJxiq6wJo3tZS+FOMw4tUWaEE0QJHkX5ipQS1fqV7UdY2XEvuxxJ7BEmVEjfK3Lp X-Received: by 2002:a05:6402:27d0:b0:451:b381:e0a1 with SMTP id c16-20020a05640227d000b00451b381e0a1mr10609592ede.4.1663974935514; Fri, 23 Sep 2022 16:15:35 -0700 (PDT) ARC-Seal: i=1; a=rsa-sha256; t=1663974935; cv=none; d=google.com; s=arc-20160816; b=fas2nAOlafctGtdQRH6Ym4gaPzQPBonzBSqOX8oPc5L/lP3f5uVvhq1sGAmcrwauLJ XjHBkbsQ8Yqu/PAj+89tE6K++vLBulNf6lAy/14hytXAP+fbiv8fDkSYsbt4eLSi5pc3 JKUxTe7hI20iDbodDaBxlbfBu/J/MS5IqCi7bUJ9iroc2kt3uJCqogIq1EEGjJSq3tU7 rtJF6WexpFfpCBOpgRbNHD3alxZa8RvgGq27givfrAf/pTGkLXVKQmSiWlWalJY1tv8+ LoxGjD+jceyNgE7OfWIalPEkqGLFFY+Grp+6XxPtC007FOZK0msBMeAaTXX5MbgcJADE hesA== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=sender:errors-to:reply-to:list-subscribe:list-help:list-post :list-archive:list-unsubscribe:list-id:precedence:subject :mime-version:message-id:to:from:date:dkim-signature:delivered-to; bh=4ccC3YdrtKKT19tYh7Dm4paHinXa/unckGxEPzmBVUI=; b=Liz0/+qCJWN6pZPcs+JByxJDhBW/tPlW7FdhINFhkVa1kah9ZTwMEfUN/BsaKPyhIU ffD/dzdrb2KvCf0SU2+q7Nj//OvvUHfsf2F1JWTPBJbkbkUy4UyPb8qLD5l1ZxwTpF7w JMlZa2+VfNiysGOfFEHhW07UO2x1G7gNPfcWgN5Enaj1MDktQypnakkjLUMRdvaZp0E7 5/IQLKyfC9sAlizgVvb7y9reapCWYyXsbZmpW1UgM0QTAM0jt8s5oQJNF22MJDSHVDaA E23mlM95usN8FDeDnCiUf8JQwgEbfH/xIY/sKDUJVlozVzWJbS1iNEZQw1cwVJ2pzxJ2 DI6A== ARC-Authentication-Results: i=1; mx.google.com; dkim=neutral (body hash did not verify) header.i=@lynne.ee header.s=s1 header.b="H/h4tiM9"; 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; dmarc=fail (p=NONE sp=NONE dis=NONE) header.from=lynne.ee Return-Path: Received: from ffbox0-bg.mplayerhq.hu (ffbox0-bg.ffmpeg.org. [79.124.17.100]) by mx.google.com with ESMTP id fs36-20020a170907602400b007305f9cf344si6689828ejc.853.2022.09.23.16.15.12; Fri, 23 Sep 2022 16:15:35 -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; dkim=neutral (body hash did not verify) header.i=@lynne.ee header.s=s1 header.b="H/h4tiM9"; 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; dmarc=fail (p=NONE sp=NONE dis=NONE) header.from=lynne.ee Received: from [127.0.1.1] (localhost [127.0.0.1]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id 66B0D68BB5D; Sat, 24 Sep 2022 02:15:08 +0300 (EEST) X-Original-To: ffmpeg-devel@ffmpeg.org Delivered-To: ffmpeg-devel@ffmpeg.org Received: from w4.tutanota.de (w4.tutanota.de [81.3.6.165]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTPS id D8AB568BADF for ; Sat, 24 Sep 2022 02:15:01 +0300 (EEST) Received: from w3.tutanota.de (unknown [192.168.1.164]) by w4.tutanota.de (Postfix) with ESMTP id 1E7CA10600E8 for ; Fri, 23 Sep 2022 23:15:00 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; t=1663974899; s=s1; d=lynne.ee; h=From:From:To:To:Subject:Subject:Content-Description:Content-ID:Content-Type:Content-Type:Content-Transfer-Encoding:Cc:Date:Date:In-Reply-To:MIME-Version:MIME-Version:Message-ID:Message-ID:Reply-To:References:Sender; bh=G/05UJ2RmzGGSTcsX1gOCVGtwiU1w8KuzdLEMGxbWJA=; b=H/h4tiM9q69km0vZpIv97kd5p3P7PPArPddKJBVA487V/+nAlU7eQbyYwq8+X9lM 1GJEgj4Vtq4UdFkFdBROpESdUdvOHqhUxyQxRumKMqKrAJRstMMtNUvF50qWUWApG7y /MHupkHY+K93fnyJs0BAlrvOUvNiZ0cpbL22T1U4aie7MYi0GJuZhMAkXUhd4Xo1o6j nPSFHlv/vgC/11362wB9uPd3Po/Sn/vruoNmYSNJLtD/nwzAD+LUBfQpI5/6FoCE71M TSmSY5caaRUx4Pr3EwfdS7k122F4yHoCYg03gyeG9qjBcnN1HwF6bmwEXOur18xa204 //BEIUZbyQ== Date: Sat, 24 Sep 2022 01:14:59 +0200 (CEST) From: Lynne To: Ffmpeg Devel Message-ID: MIME-Version: 1.0 Subject: [FFmpeg-devel] [PATCH 1/6] opus: convert encoder and decoder to lavu/tx 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 Errors-To: ffmpeg-devel-bounces@ffmpeg.org Sender: "ffmpeg-devel" X-TUID: 8xnvyokVCKPf This commit changes both the encoder and decoder to use the new lavu/tx code, which has faster C transforms and more assembly optimizations. Patch attached. From d4fdda5b57ab1e0f08eb3d78dac6b003060dfd41 Mon Sep 17 00:00:00 2001 From: Lynne Date: Sat, 24 Sep 2022 00:46:44 +0200 Subject: [PATCH 1/6] opus: convert encoder and decoder to lavu/tx This commit changes both the encoder and decoder to use the new lavu/tx code, which has faster C transforms and more assembly optimizations. --- libavcodec/opus_celt.c | 20 ++++++++++++-------- libavcodec/opus_celt.h | 5 +++-- libavcodec/opusenc.c | 15 +++++++++------ libavcodec/opusenc_psy.c | 13 ++++++++----- libavcodec/opusenc_psy.h | 4 +++- 5 files changed, 35 insertions(+), 22 deletions(-) diff --git a/libavcodec/opus_celt.c b/libavcodec/opus_celt.c index 9dbeff1927..f1fb88a56d 100644 --- a/libavcodec/opus_celt.c +++ b/libavcodec/opus_celt.c @@ -323,7 +323,8 @@ int ff_celt_decode_frame(CeltFrame *f, OpusRangeCoder *rc, { int i, j, downmix = 0; int consumed; // bits of entropy consumed thus far for this frame - MDCT15Context *imdct; + AVTXContext *imdct; + av_tx_fn imdct_fn; if (channels != 1 && channels != 2) { av_log(f->avctx, AV_LOG_ERROR, "Invalid number of coded channels: %d\n", @@ -385,7 +386,8 @@ int ff_celt_decode_frame(CeltFrame *f, OpusRangeCoder *rc, f->blocks = f->transient ? 1 << f->size : 1; f->blocksize = frame_size / f->blocks; - imdct = f->imdct[f->transient ? 0 : f->size]; + imdct = f->tx[f->transient ? 0 : f->size]; + imdct_fn = f->tx_fn[f->transient ? 0 : f->size]; if (channels == 1) { for (i = 0; i < CELT_MAX_BANDS; i++) @@ -440,8 +442,8 @@ int ff_celt_decode_frame(CeltFrame *f, OpusRangeCoder *rc, for (j = 0; j < f->blocks; j++) { float *dst = block->buf + 1024 + j * f->blocksize; - imdct->imdct_half(imdct, dst + CELT_OVERLAP / 2, f->block[i].coeffs + j, - f->blocks); + imdct_fn(imdct, dst + CELT_OVERLAP / 2, f->block[i].coeffs + j, + sizeof(float)*f->blocks); f->dsp->vector_fmul_window(dst, dst, dst + CELT_OVERLAP / 2, ff_celt_window, CELT_OVERLAP / 2); } @@ -526,8 +528,8 @@ void ff_celt_free(CeltFrame **f) if (!frm) return; - for (i = 0; i < FF_ARRAY_ELEMS(frm->imdct); i++) - ff_mdct15_uninit(&frm->imdct[i]); + for (i = 0; i < FF_ARRAY_ELEMS(frm->tx); i++) + av_tx_uninit(&frm->tx[i]); ff_celt_pvq_uninit(&frm->pvq); @@ -555,9 +557,11 @@ int ff_celt_init(AVCodecContext *avctx, CeltFrame **f, int output_channels, frm->output_channels = output_channels; frm->apply_phase_inv = apply_phase_inv; - for (i = 0; i < FF_ARRAY_ELEMS(frm->imdct); i++) - if ((ret = ff_mdct15_init(&frm->imdct[i], 1, i + 3, -1.0f/32768)) < 0) + for (i = 0; i < FF_ARRAY_ELEMS(frm->tx); i++) { + const float scale = -1.0f/32768; + if ((ret = av_tx_init(&frm->tx[i], &frm->tx_fn[i], AV_TX_FLOAT_MDCT, 1, 15 << (i + 3), &scale, 0)) < 0) goto fail; + } if ((ret = ff_celt_pvq_init(&frm->pvq, 0)) < 0) goto fail; diff --git a/libavcodec/opus_celt.h b/libavcodec/opus_celt.h index 661ca251de..291a544298 100644 --- a/libavcodec/opus_celt.h +++ b/libavcodec/opus_celt.h @@ -30,10 +30,10 @@ #include "opus_pvq.h" #include "opusdsp.h" -#include "mdct15.h" #include "libavutil/float_dsp.h" #include "libavutil/libm.h" #include "libavutil/mem_internal.h" +#include "libavutil/tx.h" #define CELT_VECTORS 11 #define CELT_ALLOC_STEPS 6 @@ -93,7 +93,8 @@ typedef struct CeltBlock { struct CeltFrame { // constant values that do not change during context lifetime AVCodecContext *avctx; - MDCT15Context *imdct[4]; + AVTXContext *tx[4]; + av_tx_fn tx_fn[4]; AVFloatDSPContext *dsp; CeltBlock block[2]; CeltPVQ *pvq; diff --git a/libavcodec/opusenc.c b/libavcodec/opusenc.c index a7a9d3a5f5..8cdd27d930 100644 --- a/libavcodec/opusenc.c +++ b/libavcodec/opusenc.c @@ -40,7 +40,8 @@ typedef struct OpusEncContext { AVCodecContext *avctx; AudioFrameQueue afq; AVFloatDSPContext *dsp; - MDCT15Context *mdct[CELT_BLOCK_NB]; + AVTXContext *tx[CELT_BLOCK_NB]; + av_tx_fn tx_fn[CELT_BLOCK_NB]; CeltPVQ *pvq; struct FFBufQueue bufqueue; @@ -204,7 +205,7 @@ static void celt_frame_mdct(OpusEncContext *s, CeltFrame *f) s->dsp->vector_fmul_reverse(&win[CELT_OVERLAP], src2, ff_celt_window - 8, 128); src1 = src2; - s->mdct[0]->mdct(s->mdct[0], b->coeffs + t, win, f->blocks); + s->tx_fn[0](s->tx[0], b->coeffs + t, win, sizeof(float)*f->blocks); } } } else { @@ -226,7 +227,7 @@ static void celt_frame_mdct(OpusEncContext *s, CeltFrame *f) ff_celt_window - 8, 128); memcpy(win + lap_dst + blk_len, temp, CELT_OVERLAP*sizeof(float)); - s->mdct[f->size]->mdct(s->mdct[f->size], b->coeffs, win, 1); + s->tx_fn[f->size](s->tx[f->size], b->coeffs, win, sizeof(float)); } } @@ -612,7 +613,7 @@ static av_cold int opus_encode_end(AVCodecContext *avctx) OpusEncContext *s = avctx->priv_data; for (int i = 0; i < CELT_BLOCK_NB; i++) - ff_mdct15_uninit(&s->mdct[i]); + av_tx_uninit(&s->tx[i]); ff_celt_pvq_uninit(&s->pvq); av_freep(&s->dsp); @@ -668,9 +669,11 @@ static av_cold int opus_encode_init(AVCodecContext *avctx) return AVERROR(ENOMEM); /* I have no idea why a base scaling factor of 68 works, could be the twiddles */ - for (int i = 0; i < CELT_BLOCK_NB; i++) - if ((ret = ff_mdct15_init(&s->mdct[i], 0, i + 3, 68 << (CELT_BLOCK_NB - 1 - i)))) + for (int i = 0; i < CELT_BLOCK_NB; i++) { + const float scale = 68 << (CELT_BLOCK_NB - 1 - i); + if ((ret = av_tx_init(&s->tx[i], &s->tx_fn[i], AV_TX_FLOAT_MDCT, 0, 15 << (i + 3), &scale, 0))) return AVERROR(ENOMEM); + } /* Zero out previous energy (matters for inter first frame) */ for (int ch = 0; ch < s->channels; ch++) diff --git a/libavcodec/opusenc_psy.c b/libavcodec/opusenc_psy.c index 1c8f69269c..3bff57d347 100644 --- a/libavcodec/opusenc_psy.c +++ b/libavcodec/opusenc_psy.c @@ -22,7 +22,6 @@ #include "opusenc_psy.h" #include "opus_pvq.h" #include "opustab.h" -#include "mdct15.h" #include "libavutil/qsort.h" static float pvq_band_cost(CeltPVQ *pvq, CeltFrame *f, OpusRangeCoder *rc, int band, @@ -99,7 +98,8 @@ static void step_collect_psy_metrics(OpusPsyContext *s, int index) s->dsp->vector_fmul(s->scratch, s->scratch, s->window[s->bsize_analysis], (OPUS_BLOCK_SIZE(s->bsize_analysis) << 1)); - s->mdct[s->bsize_analysis]->mdct(s->mdct[s->bsize_analysis], st->coeffs[ch], s->scratch, 1); + s->mdct_fn[s->bsize_analysis](s->mdct[s->bsize_analysis], st->coeffs[ch], + s->scratch, sizeof(float)); for (i = 0; i < CELT_MAX_BANDS; i++) st->bands[ch][i] = &st->coeffs[ch][ff_celt_freq_bands[i] << s->bsize_analysis]; @@ -558,13 +558,16 @@ av_cold int ff_opus_psy_init(OpusPsyContext *s, AVCodecContext *avctx, for (i = 0; i < CELT_BLOCK_NB; i++) { float tmp; const int len = OPUS_BLOCK_SIZE(i); + const float scale = 68 << (CELT_BLOCK_NB - 1 - i); s->window[i] = av_malloc(2*len*sizeof(float)); if (!s->window[i]) { ret = AVERROR(ENOMEM); goto fail; } generate_window_func(s->window[i], 2*len, WFUNC_SINE, &tmp); - if ((ret = ff_mdct15_init(&s->mdct[i], 0, i + 3, 68 << (CELT_BLOCK_NB - 1 - i)))) + ret = av_tx_init(&s->mdct[i], &s->mdct_fn[i], AV_TX_FLOAT_MDCT, + 0, 15 << (i + 3), &scale, 0); + if (ret < 0) goto fail; } @@ -575,7 +578,7 @@ fail: av_freep(&s->dsp); for (i = 0; i < CELT_BLOCK_NB; i++) { - ff_mdct15_uninit(&s->mdct[i]); + av_tx_uninit(&s->mdct[i]); av_freep(&s->window[i]); } @@ -598,7 +601,7 @@ av_cold int ff_opus_psy_end(OpusPsyContext *s) av_freep(&s->dsp); for (i = 0; i < CELT_BLOCK_NB; i++) { - ff_mdct15_uninit(&s->mdct[i]); + av_tx_uninit(&s->mdct[i]); av_freep(&s->window[i]); } diff --git a/libavcodec/opusenc_psy.h b/libavcodec/opusenc_psy.h index d4fb096a3d..0a7cdb6f2c 100644 --- a/libavcodec/opusenc_psy.h +++ b/libavcodec/opusenc_psy.h @@ -22,6 +22,7 @@ #ifndef AVCODEC_OPUSENC_PSY_H #define AVCODEC_OPUSENC_PSY_H +#include "libavutil/tx.h" #include "libavutil/mem_internal.h" #include "opusenc.h" @@ -70,7 +71,8 @@ typedef struct OpusPsyContext { int max_steps; float *window[CELT_BLOCK_NB]; - MDCT15Context *mdct[CELT_BLOCK_NB]; + AVTXContext *mdct[CELT_BLOCK_NB]; + av_tx_fn mdct_fn[CELT_BLOCK_NB]; int bsize_analysis; DECLARE_ALIGNED(32, float, scratch)[2048]; -- 2.37.2.609.g9ff673ca1a From patchwork Fri Sep 23 23:15:58 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Lynne X-Patchwork-Id: 38201 Delivered-To: ffmpegpatchwork2@gmail.com Received: by 2002:a05:6a20:3b1c:b0:96:9ee8:5cfd with SMTP id c28csp790688pzh; Fri, 23 Sep 2022 16:16:08 -0700 (PDT) X-Google-Smtp-Source: AMsMyM6oYukN9J6tuKVZaf2Mudj930oeEDx7UP9VRKFktwQgmCZzTi/tBifLddQ3ac2mMTkcMgFA X-Received: by 2002:a05:6402:3806:b0:450:bad8:8cd5 with SMTP id es6-20020a056402380600b00450bad88cd5mr10954880edb.305.1663974968219; Fri, 23 Sep 2022 16:16:08 -0700 (PDT) ARC-Seal: i=1; a=rsa-sha256; t=1663974968; cv=none; d=google.com; s=arc-20160816; b=kjOJwaefeOd9/O/oim6YzS1FDKDkRbIkX0zE5oA8p/9EhWWJhv7zcfhSdCUEcSpZqr KntnKustklddnwq5YjU6AXuVnTugKKWBJChVU74uW5AaYEI24eBUs5QYdQBAifdLO1T/ 4HYzIDt91ozr6DM+KnqGx7uxJMT6kvTPJgINwuPwSQWAFu1hNM7aq1pDJYK2NlHn3XEI L3tRv3OQtZB8mR6Zv0diAsVVFKSGh/Lo3iHg8CncnayeKtqvQ+2C9sX7ZG3/tUQtLzlk EEFReZGegVJxn0e4k61a4c1AYo2LqZsNWMXsrEy64dnmMLkw9BLjRULddH5baGo//brF +TCQ== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=sender:errors-to:reply-to:list-subscribe:list-help:list-post :list-archive:list-unsubscribe:list-id:precedence:subject :mime-version:references:in-reply-to:message-id:to:from:date :dkim-signature:delivered-to; bh=id4tUUvm70qyAvkcdwIowvzHaVXaYiVCPaFt/UurQfA=; b=FgUfPnpdf8m/x8lVT/5yqM5gn49U73Xb1miob2V7so3rR31wsBAC/VjWti4CgbUsCp Yw+P77NhW/L02h3aVPIe8y3P5Bxq1BiR5cHHKJ7Jq8cTr9RMbo+O/ln3uHOsfdCP++y1 9Q5t2AUL9pKd0vhNGXZKoxmWVdoUCRnfB97XKFxlnB4ie7WLBFH4dv/KsqQpI9VRd0+5 myF1BgbCHbD1B84OongwNbEtPsf71KErdytvlg6WAVXEGeKgngOAoBLEuKWRhmzkQBfr dnqSuwGzlbXT+u91utlv7UaSlWNCAJKgC2+EocGPjhSNPV6nUBb54b6SYWuLpCNqiWzJ aaEg== ARC-Authentication-Results: i=1; mx.google.com; dkim=neutral (body hash did not verify) header.i=@lynne.ee header.s=s1 header.b=CX4mhC0E; 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; dmarc=fail (p=NONE sp=NONE dis=NONE) header.from=lynne.ee Return-Path: Received: from ffbox0-bg.mplayerhq.hu (ffbox0-bg.ffmpeg.org. [79.124.17.100]) by mx.google.com with ESMTP id o15-20020a509b0f000000b004462de364fesi7928572edi.597.2022.09.23.16.16.07; Fri, 23 Sep 2022 16:16:08 -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; dkim=neutral (body hash did not verify) header.i=@lynne.ee header.s=s1 header.b=CX4mhC0E; 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; dmarc=fail (p=NONE sp=NONE dis=NONE) header.from=lynne.ee Received: from [127.0.1.1] (localhost [127.0.0.1]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id D380E68BB94; Sat, 24 Sep 2022 02:16:05 +0300 (EEST) X-Original-To: ffmpeg-devel@ffmpeg.org Delivered-To: ffmpeg-devel@ffmpeg.org Received: from w4.tutanota.de (w4.tutanota.de [81.3.6.165]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTPS id D8D4D68BB5C for ; Sat, 24 Sep 2022 02:15:58 +0300 (EEST) Received: from w3.tutanota.de (unknown [192.168.1.164]) by w4.tutanota.de (Postfix) with ESMTP id 91443106016E for ; Fri, 23 Sep 2022 23:15:58 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; t=1663974958; s=s1; d=lynne.ee; h=From:From:To:To:Subject:Subject:Content-Description:Content-ID:Content-Type:Content-Type:Content-Transfer-Encoding:Cc:Date:Date:In-Reply-To:In-Reply-To:MIME-Version:MIME-Version:Message-ID:Message-ID:Reply-To:References:References:Sender; bh=Hw8jz/Iw2RL6HshncpZWHa/tWbt4zYVIgiarIKImuSM=; b=CX4mhC0E6zYweQ6FQOvUHGyBc2vz8XkzSDD2CVuiPs0GNewakggAj2qBPPdKuydh 9RR7X3IYgjwiL4pn7qK96a+WOed9LwIDzm7C9LFAMsZxaFvUNfKUTrXQyCjpy6j0Pkh rZLJemGO6vUjEeoDlxAuhPcQwVaLjBx/ZJk/AZ9XpEjI/TXcidiZxh/NIYsa24XnQw0 p+SlHPrx0WGxiFApnvNvEH/oywDrY8hpUx/cjD91O7NKpL72+bUKsSu2994fK04kCwA TTYJGKsC4WWhnX2/pwNcp5iiKikmC4KDuxn+nnjYHy64WWdjpwy+LLV8b66Ljxl9zZp 1p0AXEEe4Q== Date: Sat, 24 Sep 2022 01:15:58 +0200 (CEST) From: Lynne To: FFmpeg development discussions and patches Message-ID: In-Reply-To: References: MIME-Version: 1.0 Subject: [FFmpeg-devel] [PATCH 2/6] atrac9dec: switch to lavu/tx 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 Errors-To: ffmpeg-devel-bounces@ffmpeg.org Sender: "ffmpeg-devel" X-TUID: HI/zs3T6XL00 Patch attached. From 5a310246569e19efd50b37016a80fe6171df0329 Mon Sep 17 00:00:00 2001 From: Lynne Date: Sat, 24 Sep 2022 00:51:18 +0200 Subject: [PATCH 2/6] atrac9dec: switch to lavu/tx --- libavcodec/atrac9dec.c | 21 +++++++++++++-------- 1 file changed, 13 insertions(+), 8 deletions(-) diff --git a/libavcodec/atrac9dec.c b/libavcodec/atrac9dec.c index d3a5d05799..60962b1676 100644 --- a/libavcodec/atrac9dec.c +++ b/libavcodec/atrac9dec.c @@ -25,8 +25,8 @@ #include "codec_internal.h" #include "decode.h" #include "get_bits.h" -#include "fft.h" #include "atrac9tab.h" +#include "libavutil/tx.h" #include "libavutil/lfg.h" #include "libavutil/float_dsp.h" #include "libavutil/mem_internal.h" @@ -86,7 +86,8 @@ typedef struct ATRAC9BlockData { typedef struct ATRAC9Context { AVCodecContext *avctx; AVFloatDSPContext *fdsp; - FFTContext imdct; + AVTXContext *tx; + av_tx_fn tx_fn; ATRAC9BlockData block[5]; AVLFG lfg; @@ -101,7 +102,7 @@ typedef struct ATRAC9Context { uint8_t alloc_curve[48][48]; DECLARE_ALIGNED(32, float, imdct_win)[256]; - DECLARE_ALIGNED(32, float, temp)[256]; + DECLARE_ALIGNED(32, float, temp)[2048]; } ATRAC9Context; static VLC sf_vlc[2][8]; /* Signed/unsigned, length */ @@ -778,7 +779,7 @@ imdct: const ptrdiff_t offset = wsize*frame_idx*sizeof(float); float *dst = (float *)(frame->extended_data[dst_idx] + offset); - s->imdct.imdct_half(&s->imdct, s->temp, c->coeffs); + s->tx_fn(s->tx, s->temp, c->coeffs, sizeof(float)); s->fdsp->vector_fmul_window(dst, c->prev_win, s->temp, s->imdct_win, wsize >> 1); memcpy(c->prev_win, s->temp + (wsize >> 1), sizeof(float)*wsize >> 1); @@ -834,7 +835,7 @@ static av_cold int atrac9_decode_close(AVCodecContext *avctx) { ATRAC9Context *s = avctx->priv_data; - ff_mdct_end(&s->imdct); + av_tx_uninit(&s->tx); av_freep(&s->fdsp); return 0; @@ -896,10 +897,11 @@ static av_cold void atrac9_init_static(void) static av_cold int atrac9_decode_init(AVCodecContext *avctx) { + float scale; static AVOnce static_table_init = AV_ONCE_INIT; GetBitContext gb; ATRAC9Context *s = avctx->priv_data; - int version, block_config_idx, superframe_idx, alloc_c_len; + int err, version, block_config_idx, superframe_idx, alloc_c_len; s->avctx = avctx; @@ -959,8 +961,11 @@ static av_cold int atrac9_decode_init(AVCodecContext *avctx) s->frame_count = 1 << superframe_idx; s->frame_log2 = at9_tab_sri_frame_log2[s->samplerate_idx]; - if (ff_mdct_init(&s->imdct, s->frame_log2 + 1, 1, 1.0f / 32768.0f)) - return AVERROR(ENOMEM); + scale = 1.0f / 32768.0; + err = av_tx_init(&s->tx, &s->tx_fn, AV_TX_FLOAT_MDCT, 1, + 1 << s->frame_log2, &scale, 0); + if (err < 0) + return err; s->fdsp = avpriv_float_dsp_alloc(avctx->flags & AV_CODEC_FLAG_BITEXACT); if (!s->fdsp) -- 2.37.2.609.g9ff673ca1a From patchwork Fri Sep 23 23:18:19 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Lynne X-Patchwork-Id: 38202 Delivered-To: ffmpegpatchwork2@gmail.com Received: by 2002:a05:6a20:3b1c:b0:96:9ee8:5cfd with SMTP id c28csp791556pzh; Fri, 23 Sep 2022 16:18:30 -0700 (PDT) X-Google-Smtp-Source: AMsMyM7ByzouvweL82LsydE4riXE+2PQok0UiJTkO9ekBqag2mjSzOb9efKS90eDFIj3SbCkD7Z0 X-Received: by 2002:a50:fa8c:0:b0:456:cf8e:40fd with SMTP id w12-20020a50fa8c000000b00456cf8e40fdmr4527041edr.365.1663975109905; Fri, 23 Sep 2022 16:18:29 -0700 (PDT) ARC-Seal: i=1; a=rsa-sha256; t=1663975109; cv=none; d=google.com; s=arc-20160816; b=BSaKvSFHGwi1h6UtSp0tiSCdPWzzxDPK/vU1Vd/0iQzZScsb2ObPwqf3K3Hxl47Ic5 qVsGYmSW9teDEq8aZAoXqawyJOG6qzVQ9nfzWPhrzo6Zyuucc+M8qj70thvUG3ABm4ec 1UE5pazdokAxehIKVX+mZ9gjPZOeAvxs9V+JPM+EAeyBVZP4r89JswQBsLhwcK7TylRE qRJgshVSmQibEq65LQeftDGY8IBV8BTcnwveutUBffIj7YUxdf2oAZePB8OX7FZfrhFE 3g1OJH5EM7u6aplNzguqZT3IyoXuR5BaPCkxxCAPAv/Cgy0UmlE+kFMPH8BzK681qjv3 4GiA== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=sender:errors-to:reply-to:list-subscribe:list-help:list-post :list-archive:list-unsubscribe:list-id:precedence:subject :mime-version:references:in-reply-to:message-id:to:from:date :dkim-signature:delivered-to; bh=51U8J93uo3B3nB/hjlyhEYLF6y2GCZKp4dB81Pq1XIw=; b=BqfhKxbA6seZar2PDUtOI6FndyZX9DQ8Kltb/qUkfV1kRUG7nqWPcsa95gZwPGutBU iQJvlLfuI7KDrdc6XTrLDAXyYzWJXVnnPoF3iCRvfmq8WWKi5p7fcuecIN7bl5ZbOWKy kD3daK5EIHdscy6Dgn8HfKXZss0tQW3VCVTtlcT44PqbLoZQqahvxjhqLJfvMk7tW6HU /blyZPL/tnwJ/ZcnyOmUpEoPjK3lTZo8JMJQy5odX1j6unpRAmN0pk45v1bPwc9twfOi i+twQyS5yLE1eMeyUER9sGm/IB+dQbtOibazOzXQBuQY5NnGOb6Z2j11uL9OfojDni2t utgg== ARC-Authentication-Results: i=1; mx.google.com; dkim=neutral (body hash did not verify) header.i=@lynne.ee header.s=s1 header.b=c2kAtucV; 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; dmarc=fail (p=NONE sp=NONE dis=NONE) header.from=lynne.ee Return-Path: Received: from ffbox0-bg.mplayerhq.hu (ffbox0-bg.ffmpeg.org. [79.124.17.100]) by mx.google.com with ESMTP id u28-20020a50d51c000000b00451280d355bsi8573701edi.520.2022.09.23.16.18.29; Fri, 23 Sep 2022 16:18:29 -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; dkim=neutral (body hash did not verify) header.i=@lynne.ee header.s=s1 header.b=c2kAtucV; 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; dmarc=fail (p=NONE sp=NONE dis=NONE) header.from=lynne.ee Received: from [127.0.1.1] (localhost [127.0.0.1]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id 1685368BB94; Sat, 24 Sep 2022 02:18:27 +0300 (EEST) X-Original-To: ffmpeg-devel@ffmpeg.org Delivered-To: ffmpeg-devel@ffmpeg.org Received: from w4.tutanota.de (w4.tutanota.de [81.3.6.165]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTPS id 5131B68BB5C for ; Sat, 24 Sep 2022 02:18:20 +0300 (EEST) Received: from w3.tutanota.de (unknown [192.168.1.164]) by w4.tutanota.de (Postfix) with ESMTP id 05E8D106016E for ; Fri, 23 Sep 2022 23:18:20 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; t=1663975099; s=s1; d=lynne.ee; h=From:From:To:To:Subject:Subject:Content-Description:Content-ID:Content-Type:Content-Type:Content-Transfer-Encoding:Cc:Date:Date:In-Reply-To:In-Reply-To:MIME-Version:MIME-Version:Message-ID:Message-ID:Reply-To:References:References:Sender; bh=rvZm5sCnAXlz7fkAu4xPBviK3yp/BgsIY5VowbJZabM=; b=c2kAtucVM0rUksja6bj8WUHEla04+Pg/kUMgIzgkfnFPgsG7yQiVTLbq8Gr1dJ+z TLXz5b5+Cg3vklv+WqNFl1fxXbPUN50kh4+4mxapmrzRmCAYWn0sj4ORQe4S7D7Gnr1 Xpq0qMCbGhVJruIDcRFeSUAgAyguSBGCg9zz5WQ0ldRk0Sp9UQRqac1ZUNW8NXx8I6i kdHNlPaFvcKTU1sJq2dgnv3gn11amLBbc9FAncw/gzP34s/LZkgKlSrIoPz3AJ32cwR jO2CZ368AWzN6twtLemyTVkvdOnbKZU57KK+/fZtHY9LPLOTEy+dOPsEst14r3s2hI8 emCqjf3k0Q== Date: Sat, 24 Sep 2022 01:18:19 +0200 (CEST) From: Lynne To: FFmpeg development discussions and patches Message-ID: In-Reply-To: References: MIME-Version: 1.0 Subject: [FFmpeg-devel] [PATCH 3/6] ac3: convert encoder and decoder to lavu/tx 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 Errors-To: ffmpeg-devel-bounces@ffmpeg.org Sender: "ffmpeg-devel" X-TUID: U+RGFa7qbaQT The fixed-point transforms are much better, faster, and more accurate. Patch attached. From e54775cc93ceb27d9faabe1ddf9b1eacb269826b Mon Sep 17 00:00:00 2001 From: Lynne Date: Sat, 24 Sep 2022 01:05:19 +0200 Subject: [PATCH 3/6] ac3: convert encoder and decoder to lavu/tx The fixed-point transforms are much better, faster, and more accurate. --- libavcodec/ac3dec.c | 22 +++++++++++++--------- libavcodec/ac3dec.h | 6 +++--- libavcodec/ac3dec_fixed.c | 3 ++- libavcodec/ac3dec_float.c | 1 + libavcodec/ac3enc.c | 2 +- libavcodec/ac3enc.h | 7 ++++--- libavcodec/ac3enc_fixed.c | 19 ++++--------------- libavcodec/ac3enc_float.c | 18 +++--------------- libavcodec/ac3enc_template.c | 4 ++-- 9 files changed, 33 insertions(+), 49 deletions(-) diff --git a/libavcodec/ac3dec.c b/libavcodec/ac3dec.c index aba8e0fb7f..cd3320caa0 100644 --- a/libavcodec/ac3dec.c +++ b/libavcodec/ac3dec.c @@ -217,13 +217,17 @@ static av_cold int ac3_decode_init(AVCodecContext *avctx) { static AVOnce init_static_once = AV_ONCE_INIT; AC3DecodeContext *s = avctx->priv_data; + const float scale = 1.0f; int i, ret; s->avctx = avctx; - if ((ret = ff_mdct_init(&s->imdct_256, 8, 1, 1.0)) < 0 || - (ret = ff_mdct_init(&s->imdct_512, 9, 1, 1.0)) < 0) + if ((ret = av_tx_init(&s->tx_128, &s->tx_fn_128, IMDCT_TYPE, 1, 128, &scale, 0))) return ret; + + if ((ret = av_tx_init(&s->tx_256, &s->tx_fn_256, IMDCT_TYPE, 1, 256, &scale, 0))) + return ret; + AC3_RENAME(ff_kbd_window_init)(s->window, 5.0, 256); ff_bswapdsp_init(&s->bdsp); @@ -721,10 +725,10 @@ static inline void do_imdct(AC3DecodeContext *s, int channels, int offset) for (ch = 1; ch <= channels; ch++) { if (s->block_switch[ch]) { int i; - FFTSample *x = s->tmp_output + 128; + INTFLOAT *x = s->tmp_output + 128; for (i = 0; i < 128; i++) x[i] = s->transform_coeffs[ch][2 * i]; - s->imdct_256.imdct_half(&s->imdct_256, s->tmp_output, x); + s->tx_fn_128(s->tx_128, s->tmp_output, x, sizeof(INTFLOAT)); #if USE_FIXED s->fdsp->vector_fmul_window_scaled(s->outptr[ch - 1], s->delay[ch - 1 + offset], s->tmp_output, s->window, 128, 8); @@ -734,9 +738,9 @@ static inline void do_imdct(AC3DecodeContext *s, int channels, int offset) #endif for (i = 0; i < 128; i++) x[i] = s->transform_coeffs[ch][2 * i + 1]; - s->imdct_256.imdct_half(&s->imdct_256, s->delay[ch - 1 + offset], x); + s->tx_fn_256(s->tx_256, s->delay[ch - 1 + offset], x, sizeof(INTFLOAT)); } else { - s->imdct_512.imdct_half(&s->imdct_512, s->tmp_output, s->transform_coeffs[ch]); + s->tx_fn_256(s->tx_256, s->tmp_output, s->transform_coeffs[ch], sizeof(INTFLOAT)); #if USE_FIXED s->fdsp->vector_fmul_window_scaled(s->outptr[ch - 1], s->delay[ch - 1 + offset], s->tmp_output, s->window, 128, 8); @@ -744,7 +748,7 @@ static inline void do_imdct(AC3DecodeContext *s, int channels, int offset) s->fdsp->vector_fmul_window(s->outptr[ch - 1], s->delay[ch - 1 + offset], s->tmp_output, s->window, 128); #endif - memcpy(s->delay[ch - 1 + offset], s->tmp_output + 128, 128 * sizeof(FFTSample)); + memcpy(s->delay[ch - 1 + offset], s->tmp_output + 128, 128 * sizeof(INTFLOAT)); } } } @@ -1865,8 +1869,8 @@ skip: static av_cold int ac3_decode_end(AVCodecContext *avctx) { AC3DecodeContext *s = avctx->priv_data; - ff_mdct_end(&s->imdct_512); - ff_mdct_end(&s->imdct_256); + av_tx_uninit(&s->tx_256); + av_tx_uninit(&s->tx_128); av_freep(&s->fdsp); av_freep(&s->downmix_coeffs[0]); diff --git a/libavcodec/ac3dec.h b/libavcodec/ac3dec.h index 88651ae61f..138b462abb 100644 --- a/libavcodec/ac3dec.h +++ b/libavcodec/ac3dec.h @@ -50,6 +50,7 @@ #ifndef AVCODEC_AC3DEC_H #define AVCODEC_AC3DEC_H +#include "libavutil/tx.h" #include "libavutil/float_dsp.h" #include "libavutil/fixed_dsp.h" #include "libavutil/lfg.h" @@ -60,7 +61,6 @@ #include "avcodec.h" #include "bswapdsp.h" #include "get_bits.h" -#include "fft.h" #include "fmtconvert.h" #define AC3_OUTPUT_LFEON 8 @@ -223,8 +223,8 @@ typedef struct AC3DecodeContext { ///@name IMDCT int block_switch[AC3_MAX_CHANNELS]; ///< block switch flags (blksw) - FFTContext imdct_512; ///< for 512 sample IMDCT - FFTContext imdct_256; ///< for 256 sample IMDCT + AVTXContext *tx_128, *tx_256; + av_tx_fn tx_fn_128, tx_fn_256; ///@} ///@name Optimization diff --git a/libavcodec/ac3dec_fixed.c b/libavcodec/ac3dec_fixed.c index 0a7ae6cfbf..c9e5cda69c 100644 --- a/libavcodec/ac3dec_fixed.c +++ b/libavcodec/ac3dec_fixed.c @@ -47,11 +47,12 @@ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA */ -#define FFT_FLOAT 0 #define USE_FIXED 1 #include "ac3dec.h" #include "codec_internal.h" +#define IMDCT_TYPE AV_TX_INT32_MDCT +#include "ac3dec.h" static const int end_freq_inv_tab[8] = { diff --git a/libavcodec/ac3dec_float.c b/libavcodec/ac3dec_float.c index 8c1adb3e01..b8868d8ee1 100644 --- a/libavcodec/ac3dec_float.c +++ b/libavcodec/ac3dec_float.c @@ -29,6 +29,7 @@ */ #include "config_components.h" +#define IMDCT_TYPE AV_TX_FLOAT_MDCT #include "ac3dec.h" #include "codec_internal.h" diff --git a/libavcodec/ac3enc.c b/libavcodec/ac3enc.c index a090576823..fbedf40d20 100644 --- a/libavcodec/ac3enc.c +++ b/libavcodec/ac3enc.c @@ -2203,7 +2203,7 @@ av_cold int ff_ac3_encode_close(AVCodecContext *avctx) av_freep(&block->cpl_coord_mant); } - s->mdct_end(s); + av_tx_uninit(&s->tx); return 0; } diff --git a/libavcodec/ac3enc.h b/libavcodec/ac3enc.h index f0dc006759..55e88d69e4 100644 --- a/libavcodec/ac3enc.h +++ b/libavcodec/ac3enc.h @@ -31,12 +31,13 @@ #include #include "libavutil/opt.h" +#include "libavutil/tx.h" + #include "ac3.h" #include "ac3defs.h" #include "ac3dsp.h" #include "avcodec.h" #include "codec_internal.h" -#include "fft.h" #include "mathops.h" #include "me_cmp.h" #include "put_bits.h" @@ -167,7 +168,8 @@ typedef struct AC3EncodeContext { #endif MECmpContext mecc; AC3DSPContext ac3dsp; ///< AC-3 optimized functions - FFTContext mdct; ///< FFT context for MDCT calculation + AVTXContext *tx; ///< FFT context for MDCT calculation + av_tx_fn tx_fn; const SampleType *mdct_window; ///< MDCT window function array AC3Block blocks[AC3_MAX_BLOCKS]; ///< per-block info @@ -257,7 +259,6 @@ typedef struct AC3EncodeContext { int warned_alternate_bitstream; /* fixed vs. float function pointers */ - void (*mdct_end)(struct AC3EncodeContext *s); int (*mdct_init)(struct AC3EncodeContext *s); /* fixed vs. float templated function pointers */ diff --git a/libavcodec/ac3enc_fixed.c b/libavcodec/ac3enc_fixed.c index a22d3b4abf..76e5392733 100644 --- a/libavcodec/ac3enc_fixed.c +++ b/libavcodec/ac3enc_fixed.c @@ -27,7 +27,7 @@ */ #define AC3ENC_FLOAT 0 -#define FFT_FLOAT 0 +#include "internal.h" #include "audiodsp.h" #include "ac3enc.h" #include "codec_internal.h" @@ -66,20 +66,8 @@ static CoefType calc_cpl_coord(CoefSumType energy_ch, CoefSumType energy_cpl) } } - #include "ac3enc_template.c" - -/** - * Finalize MDCT and free allocated memory. - * - * @param s AC-3 encoder private context - */ -static av_cold void ac3_fixed_mdct_end(AC3EncodeContext *s) -{ - ff_mdct_end(&s->mdct); -} - /** * Initialize MDCT tables. * @@ -89,6 +77,7 @@ static av_cold void ac3_fixed_mdct_end(AC3EncodeContext *s) static av_cold int ac3_fixed_mdct_init(AC3EncodeContext *s) { float fwin[AC3_BLOCK_SIZE]; + const float scale = -1.0f; int32_t *iwin = av_malloc_array(AC3_BLOCK_SIZE, sizeof(*iwin)); if (!iwin) @@ -104,7 +93,8 @@ static av_cold int ac3_fixed_mdct_init(AC3EncodeContext *s) if (!s->fdsp) return AVERROR(ENOMEM); - return ff_mdct_init(&s->mdct, 9, 0, -1.0); + return av_tx_init(&s->tx, &s->tx_fn, AV_TX_INT32_MDCT, 0, + AC3_BLOCK_SIZE, &scale, 0); } @@ -112,7 +102,6 @@ static av_cold int ac3_fixed_encode_init(AVCodecContext *avctx) { AC3EncodeContext *s = avctx->priv_data; s->fixed_point = 1; - s->mdct_end = ac3_fixed_mdct_end; s->mdct_init = ac3_fixed_mdct_init; s->allocate_sample_buffers = allocate_sample_buffers; return ff_ac3_encode_init(avctx); diff --git a/libavcodec/ac3enc_float.c b/libavcodec/ac3enc_float.c index 6238980690..8a3f605b48 100644 --- a/libavcodec/ac3enc_float.c +++ b/libavcodec/ac3enc_float.c @@ -75,21 +75,8 @@ static void sum_square_butterfly(AC3EncodeContext *s, float sum[4], s->ac3dsp.sum_square_butterfly_float(sum, coef0, coef1, len); } - #include "ac3enc_template.c" - -/** - * Finalize MDCT and free allocated memory. - * - * @param s AC-3 encoder private context - */ -static av_cold void ac3_float_mdct_end(AC3EncodeContext *s) -{ - ff_mdct_end(&s->mdct); -} - - /** * Initialize MDCT tables. * @@ -98,6 +85,7 @@ static av_cold void ac3_float_mdct_end(AC3EncodeContext *s) */ static av_cold int ac3_float_mdct_init(AC3EncodeContext *s) { + const float scale = -2.0 / AC3_WINDOW_SIZE; float *window = av_malloc_array(AC3_BLOCK_SIZE, sizeof(*window)); if (!window) { av_log(s->avctx, AV_LOG_ERROR, "Cannot allocate memory.\n"); @@ -107,14 +95,14 @@ static av_cold int ac3_float_mdct_init(AC3EncodeContext *s) ff_kbd_window_init(window, 5.0, AC3_BLOCK_SIZE); s->mdct_window = window; - return ff_mdct_init(&s->mdct, 9, 0, -2.0 / AC3_WINDOW_SIZE); + return av_tx_init(&s->tx, &s->tx_fn, AV_TX_FLOAT_MDCT, 0, + AC3_BLOCK_SIZE, &scale, 0); } av_cold int ff_ac3_float_encode_init(AVCodecContext *avctx) { AC3EncodeContext *s = avctx->priv_data; - s->mdct_end = ac3_float_mdct_end; s->mdct_init = ac3_float_mdct_init; s->allocate_sample_buffers = allocate_sample_buffers; s->fdsp = avpriv_float_dsp_alloc(avctx->flags & AV_CODEC_FLAG_BITEXACT); diff --git a/libavcodec/ac3enc_template.c b/libavcodec/ac3enc_template.c index de66964d0d..be4ecebc9c 100644 --- a/libavcodec/ac3enc_template.c +++ b/libavcodec/ac3enc_template.c @@ -98,8 +98,8 @@ static void apply_mdct(AC3EncodeContext *s) &input_samples[AC3_BLOCK_SIZE], s->mdct_window, AC3_BLOCK_SIZE); - s->mdct.mdct_calc(&s->mdct, block->mdct_coef[ch+1], - s->windowed_samples); + s->tx_fn(s->tx, block->mdct_coef[ch+1], + s->windowed_samples, sizeof(float)); } } } -- 2.37.2.609.g9ff673ca1a From patchwork Fri Sep 23 23:18:54 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Lynne X-Patchwork-Id: 38204 Delivered-To: ffmpegpatchwork2@gmail.com Received: by 2002:a05:6a20:3b1c:b0:96:9ee8:5cfd with SMTP id c28csp791793pzh; Fri, 23 Sep 2022 16:19:03 -0700 (PDT) X-Google-Smtp-Source: AMsMyM4EqW4/Wts/WfL8IpT65rA2yohiaFACVwKuaCrQBlxEryoiBr9dl44R9get5Kjg6S0IrZjI X-Received: by 2002:a17:907:75c7:b0:779:bf7:9be7 with SMTP id jl7-20020a17090775c700b007790bf79be7mr9132797ejc.432.1663975143348; Fri, 23 Sep 2022 16:19:03 -0700 (PDT) ARC-Seal: i=1; a=rsa-sha256; t=1663975143; cv=none; d=google.com; s=arc-20160816; b=0VBfzaDHlAx20r2kEwpMlTO6Zn9ex1k6AtAGjC5Ka2GpnLHXGXrbr54yL0Xa672cwn LVBO6BWd6G5aDWeUWmlUwl6KdsjXkjqyg2m1BnCrSmqw03EY/lMwL+6ENQSAcfl6Y3jQ IJ/ya3ga28xltAIvpkS18lbOPMli6WBzj48mO/mfdvBVlYAvOkNaYKfoOf8JaUQ9nsE6 5HbVZr56LNG9UgQb7VP/r1mpXWg5568PUA9AbhNch0GeUg/MfyI3EtDCoZlsBanpVayJ /1igOWXsp61z/LSBHoZBc5FuI6sCKlx3M//q/O5TrtIClWaw0kuNMnqhpN0CFuKW6C3F FGYA== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=sender:errors-to:reply-to:list-subscribe:list-help:list-post :list-archive:list-unsubscribe:list-id:precedence:subject :mime-version:references:in-reply-to:message-id:to:from:date :dkim-signature:delivered-to; bh=THNwkYtyaJxGONL0EHyO2Ko/S6snm6UlqhJ9LB/CSMQ=; b=m1g80uIkOpMF9a5wcPbFH8H0cY+sPM8dm1OFAlojLWzeIuhDym+aqFHra9lP/3Byq3 TCHEL0C3UTTXbZQOUYY0Ye39NkL8w4WAL72ULbuz9yYU44atKjp9SoTWvxzAqHEKhHhO 4MudLc9QxoibxCqhiK6TWkAZmBt3y5XInetJ2QG2NX1vw+prl0dOVP/601EvG4XVUHx+ E8g4VlEfyX+8SuFGhNKa8I1aiuib7mUTIsIF7TrdPfrBLdzBaPxTAkg0Der2nNKm1SlO F8kDxyj86rAr2CwIHVHr+yrVKh4nuwCDtch6glarSCKYqM2Y5st46HiBwiZwAEow9sVf +2Lg== ARC-Authentication-Results: i=1; mx.google.com; dkim=neutral (body hash did not verify) header.i=@lynne.ee header.s=s1 header.b="U3ALqST/"; 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; dmarc=fail (p=NONE sp=NONE dis=NONE) header.from=lynne.ee Return-Path: Received: from ffbox0-bg.mplayerhq.hu (ffbox0-bg.ffmpeg.org. [79.124.17.100]) by mx.google.com with ESMTP id hc41-20020a17090716a900b007823f12eaf7si7829612ejc.510.2022.09.23.16.19.03; Fri, 23 Sep 2022 16:19:03 -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; dkim=neutral (body hash did not verify) header.i=@lynne.ee header.s=s1 header.b="U3ALqST/"; 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; dmarc=fail (p=NONE sp=NONE dis=NONE) header.from=lynne.ee Received: from [127.0.1.1] (localhost [127.0.0.1]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id 2195768BBAB; Sat, 24 Sep 2022 02:19:01 +0300 (EEST) X-Original-To: ffmpeg-devel@ffmpeg.org Delivered-To: ffmpeg-devel@ffmpeg.org Received: from w4.tutanota.de (w4.tutanota.de [81.3.6.165]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTPS id 70B5B68BB67 for ; Sat, 24 Sep 2022 02:18:54 +0300 (EEST) Received: from w3.tutanota.de (unknown [192.168.1.164]) by w4.tutanota.de (Postfix) with ESMTP id 299D5106016E for ; Fri, 23 Sep 2022 23:18:54 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; t=1663975134; s=s1; d=lynne.ee; h=From:From:To:To:Subject:Subject:Content-Description:Content-ID:Content-Type:Content-Type:Content-Transfer-Encoding:Cc:Date:Date:In-Reply-To:In-Reply-To:MIME-Version:MIME-Version:Message-ID:Message-ID:Reply-To:References:References:Sender; bh=VtN3OIV+VgUO9ihCVzwaFy7YRloGVeXBmmJiZwMTVW0=; b=U3ALqST/M3XJKUFjVyoxOG5ujHnfeOq8rPkVLgDDVAHFq0THf810Yxm1m8g/XK2X vhCRets0yfXo9JJ7IrSZNIoSRnKRZc7DVbrxl5kUTjeFFWej61MSVQViK5Oduly6iwZ BbzwtQtTIU7H6/4YBRrDkv39FvDq89kRxMMFBHO9VqHBnPhsbtB0obXsqf//8D9IjUO SqO/W2IGLdGCidverCE+GBhc6xOS8dm++xEilopwTzKGU2ocdas9xzFkLYS2n+WZG8U MCwTrUgVjRu0+JA+FWH3v8DRxqkNDV4OtqlpVl8IfF/uUrzDFx72zfqZM6A0XEN2rcg N01IN+qeMA== Date: Sat, 24 Sep 2022 01:18:54 +0200 (CEST) From: Lynne To: FFmpeg development discussions and patches Message-ID: In-Reply-To: References: MIME-Version: 1.0 Subject: [FFmpeg-devel] [PATCH 4/6] vorbisdec: convert to lavu/tx 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 Errors-To: ffmpeg-devel-bounces@ffmpeg.org Sender: "ffmpeg-devel" X-TUID: 1mbbP+gdLCNi Patch attached. From 1334c8c26a8d1c3f8e2aa98b902b2dab6e524a84 Mon Sep 17 00:00:00 2001 From: Lynne Date: Sat, 24 Sep 2022 01:07:15 +0200 Subject: [PATCH 4/6] vorbisdec: convert to lavu/tx --- libavcodec/vorbisdec.c | 29 ++++++++++++++++++++--------- 1 file changed, 20 insertions(+), 9 deletions(-) diff --git a/libavcodec/vorbisdec.c b/libavcodec/vorbisdec.c index 0d04e7c2c4..44c76d0da2 100644 --- a/libavcodec/vorbisdec.c +++ b/libavcodec/vorbisdec.c @@ -29,6 +29,7 @@ #include #include +#include "libavutil/tx.h" #include "libavutil/avassert.h" #include "libavutil/float_dsp.h" @@ -36,7 +37,6 @@ #include "avcodec.h" #include "codec_internal.h" #include "decode.h" -#include "fft.h" #include "get_bits.h" #include "vorbis.h" #include "vorbisdsp.h" @@ -129,7 +129,8 @@ typedef struct vorbis_context_s { VorbisDSPContext dsp; AVFloatDSPContext *fdsp; - FFTContext mdct[2]; + AVTXContext *tx[2]; + av_tx_fn tx_fn[2]; uint8_t first_frame; int64_t initial_pts; uint32_t version; @@ -201,8 +202,8 @@ static void vorbis_free(vorbis_context *vc) av_freep(&vc->residues); av_freep(&vc->modes); - ff_mdct_end(&vc->mdct[0]); - ff_mdct_end(&vc->mdct[1]); + av_tx_uninit(&vc->tx[0]); + av_tx_uninit(&vc->tx[1]); if (vc->codebooks) for (i = 0; i < vc->codebook_count; ++i) { @@ -961,6 +962,8 @@ static int vorbis_parse_setup_hdr(vorbis_context *vc) static int vorbis_parse_id_hdr(vorbis_context *vc) { + int ret; + const float mdct_scale = -1.0f; GetBitContext *gb = &vc->gb; unsigned bl0, bl1; @@ -1008,8 +1011,14 @@ static int vorbis_parse_id_hdr(vorbis_context *vc) vc->previous_window = -1; - ff_mdct_init(&vc->mdct[0], bl0, 1, -1.0); - ff_mdct_init(&vc->mdct[1], bl1, 1, -1.0); + if ((ret = av_tx_init(&vc->tx[0], &vc->tx_fn[0], AV_TX_FLOAT_MDCT, 1, 1 << (bl0 - 1), + &mdct_scale, 0))) + return ret; + + if ((ret = av_tx_init(&vc->tx[1], &vc->tx_fn[1], AV_TX_FLOAT_MDCT, 1, 1 << (bl1 - 1), + &mdct_scale, 0))) + return ret; + vc->fdsp = avpriv_float_dsp_alloc(vc->avctx->flags & AV_CODEC_FLAG_BITEXACT); if (!vc->fdsp) return AVERROR(ENOMEM); @@ -1584,7 +1593,8 @@ static inline int vorbis_residue_decode(vorbis_context *vc, vorbis_residue *vr, static int vorbis_parse_audio_packet(vorbis_context *vc, float **floor_ptr) { GetBitContext *gb = &vc->gb; - FFTContext *mdct; + AVTXContext *tx; + av_tx_fn tx_fn; int previous_window = vc->previous_window; unsigned mode_number, blockflag, blocksize; int i, j; @@ -1706,12 +1716,13 @@ static int vorbis_parse_audio_packet(vorbis_context *vc, float **floor_ptr) // Dotproduct, MDCT - mdct = &vc->mdct[blockflag]; + tx = vc->tx[blockflag]; + tx_fn = vc->tx_fn[blockflag]; for (j = vc->audio_channels-1;j >= 0; j--) { ch_res_ptr = vc->channel_residues + res_chan[j] * blocksize / 2; vc->fdsp->vector_fmul(floor_ptr[j], floor_ptr[j], ch_res_ptr, blocksize / 2); - mdct->imdct_half(mdct, ch_res_ptr, floor_ptr[j]); + tx_fn(tx, ch_res_ptr, floor_ptr[j], sizeof(float)); } // Overlap/add, save data for next overlapping -- 2.37.2.609.g9ff673ca1a From patchwork Fri Sep 23 23:19:36 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Lynne X-Patchwork-Id: 38205 Delivered-To: ffmpegpatchwork2@gmail.com Received: by 2002:a05:6a20:3b1c:b0:96:9ee8:5cfd with SMTP id c28csp792089pzh; Fri, 23 Sep 2022 16:19:45 -0700 (PDT) X-Google-Smtp-Source: AMsMyM6jrPBphnAuZkjoWwvnhRNm42pKyiiFF4b4E20qfNfed0/jDTRnVEFuD+3RWQ8UwxaIZRNH X-Received: by 2002:a17:906:3a15:b0:73d:80bf:542c with SMTP id z21-20020a1709063a1500b0073d80bf542cmr9223484eje.633.1663975185600; Fri, 23 Sep 2022 16:19:45 -0700 (PDT) ARC-Seal: i=1; a=rsa-sha256; t=1663975185; cv=none; d=google.com; s=arc-20160816; b=AOynTcxntbKk4LnHV2qveXd7BFrrGgvpgUKAXkp2PilByLz/bly+fEcD2OqIor1TWg p7mcN0Q0ahhMxvJJgJdbodGMDf549KL7yPlFZqr3UKqyrSavJPVw5QmpcuqoUNGjUZws l82QixoehxWdJfbTiH57muLfwijqFv3dexxbYz5rDPwot7zYmJDnVFvF5EJpWbRxwv6G l3pPwZHLJQStlST0fT/1defcak+JEJYJZSL69dpEjZKYs73/GEQslBbfNpFp6HvARvdm JvWOljxdNqMzDPkrC/OET39sU3X+3Wo6G7KY6QmNh5j3VHk1mSgKRN7N1c4HFxW6D1Nh 5tkg== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=sender:errors-to:reply-to:list-subscribe:list-help:list-post :list-archive:list-unsubscribe:list-id:precedence:subject :mime-version:references:in-reply-to:message-id:to:from:date :dkim-signature:delivered-to; bh=/Sky0ef2X/nLfl1lgutIkNvT+whHev3Wulxylu1Liys=; b=VfpS7ACvw4mGf2idLHaBRW5pdVpH7YdXU3q68WP6bVg7x0v9Be9mb5nd72po17vexa 6PYiYWf/ZPviPICo2/khBmXfxNX8zj9g3KCO1Ir8c94bq8wwdQMT+/G2jXTlhj3VTeBY ShpwigvKjycKZJZzWyurB+9IMq7OBAgMFpvXlLy+uOzlz0/YYit2xWImsuNBExKxlnLc fDsljp7TPQ2Ypvv19Om9KVAPUS1j4aiW6cIZPdSf+LL5LW9TwyMmCgHSJvQA9+ttDvSO MlTqRGogLqh9HPmRS2M9ltvcf+NWLrvi+Y+GW1mQgqH5oVcFAuAyklK9+Ocrs5Y1yWms BmOg== ARC-Authentication-Results: i=1; mx.google.com; dkim=neutral (body hash did not verify) header.i=@lynne.ee header.s=s1 header.b=O3fpzz6G; 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; dmarc=fail (p=NONE sp=NONE dis=NONE) header.from=lynne.ee Return-Path: Received: from ffbox0-bg.mplayerhq.hu (ffbox0-bg.ffmpeg.org. [79.124.17.100]) by mx.google.com with ESMTP id bx19-20020a170906a1d300b00780e1dc484esi8875677ejb.776.2022.09.23.16.19.45; Fri, 23 Sep 2022 16:19:45 -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; dkim=neutral (body hash did not verify) header.i=@lynne.ee header.s=s1 header.b=O3fpzz6G; 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; dmarc=fail (p=NONE sp=NONE dis=NONE) header.from=lynne.ee Received: from [127.0.1.1] (localhost [127.0.0.1]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id 2D60868BBB3; Sat, 24 Sep 2022 02:19:43 +0300 (EEST) X-Original-To: ffmpeg-devel@ffmpeg.org Delivered-To: ffmpeg-devel@ffmpeg.org Received: from w4.tutanota.de (w4.tutanota.de [81.3.6.165]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTPS id 80AD468BB86 for ; Sat, 24 Sep 2022 02:19:36 +0300 (EEST) Received: from w3.tutanota.de (unknown [192.168.1.164]) by w4.tutanota.de (Postfix) with ESMTP id 36BA910600E8 for ; Fri, 23 Sep 2022 23:19:36 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; t=1663975176; s=s1; d=lynne.ee; h=From:From:To:To:Subject:Subject:Content-Description:Content-ID:Content-Type:Content-Type:Content-Transfer-Encoding:Cc:Date:Date:In-Reply-To:In-Reply-To:MIME-Version:MIME-Version:Message-ID:Message-ID:Reply-To:References:References:Sender; bh=r7WOyX1t2S++ciUFYJFICvISg1srXl2myuCKLUIKE2o=; b=O3fpzz6GaqoAITxKcYa4QHzZj7oFNorkmWRCmUBDW9gzzxeciP97Nk/lUNPmKhCa vTUzUCctT3sK05KWXZuebXXZWXFa+eeUXvPWGxM8VWp9lYkO/zdAVBVxwAoWZHw+wSO A72qj3C0E/1lxbr7qK+TbYx5rSh0seaob1J7l97ekVLOoH34U0vIE5RzoiZ0kXNNUfE /oKNi/iNzcBYbE6RWxGHvM8I8QCwmpzssFO7sFKOtHvFL6+apYsZRZhDcB271kTD8Km ipNHh+sek7VmLRKgcDY4vHsv2z27yWdYde7Sz/Ot+EN36GUwWvhtszFGHsw0DR1PCA3 iUuoR0vCPw== Date: Sat, 24 Sep 2022 01:19:36 +0200 (CEST) From: Lynne To: FFmpeg development discussions and patches Message-ID: In-Reply-To: References: MIME-Version: 1.0 Subject: [FFmpeg-devel] [PATCH 5/6] twinvq: convert to lavu/tx 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 Errors-To: ffmpeg-devel-bounces@ffmpeg.org Sender: "ffmpeg-devel" X-TUID: oSQw0jHjlDFD Patch attached. From 685ac65ce0f391fd1d3a06e191c9659dacd375be Mon Sep 17 00:00:00 2001 From: Lynne Date: Sat, 24 Sep 2022 01:07:44 +0200 Subject: [PATCH 5/6] twinvq: convert to lavu/tx --- libavcodec/twinvq.c | 12 +++++++----- libavcodec/twinvq.h | 6 ++++-- 2 files changed, 11 insertions(+), 7 deletions(-) diff --git a/libavcodec/twinvq.c b/libavcodec/twinvq.c index da10923d78..8cd3c91e14 100644 --- a/libavcodec/twinvq.c +++ b/libavcodec/twinvq.c @@ -328,7 +328,8 @@ static const uint8_t wtype_to_wsize[] = { 0, 0, 2, 2, 2, 1, 0, 1, 1 }; static void imdct_and_window(TwinVQContext *tctx, enum TwinVQFrameType ftype, int wtype, float *in, float *prev, int ch) { - FFTContext *mdct = &tctx->mdct_ctx[ftype]; + AVTXContext *tx = tctx->tx[ftype]; + av_tx_fn tx_fn = tctx->tx_fn[ftype]; const TwinVQModeTab *mtab = tctx->mtab; int bsize = mtab->size / mtab->fmode[ftype].sub; int size = mtab->size; @@ -357,7 +358,7 @@ static void imdct_and_window(TwinVQContext *tctx, enum TwinVQFrameType ftype, wsize = types_sizes[wtype_to_wsize[sub_wtype]]; - mdct->imdct_half(mdct, buf1 + bsize * j, in + bsize * j); + tx_fn(tx, buf1 + bsize * j, in + bsize * j, sizeof(float)); tctx->fdsp->vector_fmul_window(out2, prev_buf + (bsize - wsize) / 2, buf1 + bsize * j, @@ -543,8 +544,9 @@ static av_cold int init_mdct_win(TwinVQContext *tctx) for (i = 0; i < 3; i++) { int bsize = tctx->mtab->size / tctx->mtab->fmode[i].sub; - if ((ret = ff_mdct_init(&tctx->mdct_ctx[i], av_log2(bsize) + 1, 1, - -sqrt(norm / bsize) / (1 << 15)))) + const float scale = -sqrt(norm / bsize) / (1 << 15); + if ((ret = av_tx_init(&tctx->tx[i], &tctx->tx_fn[i], AV_TX_FLOAT_MDCT, + 1, bsize, &scale, 0))) return ret; } @@ -745,7 +747,7 @@ av_cold int ff_twinvq_decode_close(AVCodecContext *avctx) int i; for (i = 0; i < 3; i++) { - ff_mdct_end(&tctx->mdct_ctx[i]); + av_tx_uninit(&tctx->tx[i]); av_freep(&tctx->cos_tabs[i]); } diff --git a/libavcodec/twinvq.h b/libavcodec/twinvq.h index b3c881cfac..72b9ba8198 100644 --- a/libavcodec/twinvq.h +++ b/libavcodec/twinvq.h @@ -25,10 +25,11 @@ #include #include +#include "libavutil/tx.h" #include "libavutil/common.h" #include "libavutil/float_dsp.h" #include "avcodec.h" -#include "fft.h" +#include "internal.h" enum TwinVQCodec { TWINVQ_CODEC_VQF, @@ -136,7 +137,8 @@ typedef struct TwinVQModeTab { typedef struct TwinVQContext { AVCodecContext *avctx; AVFloatDSPContext *fdsp; - FFTContext mdct_ctx[3]; + AVTXContext *tx[3]; + av_tx_fn tx_fn[3]; const TwinVQModeTab *mtab; -- 2.37.2.609.g9ff673ca1a From patchwork Fri Sep 23 23:20:36 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Lynne X-Patchwork-Id: 38203 Delivered-To: ffmpegpatchwork2@gmail.com Received: by 2002:a05:6a20:3b1c:b0:96:9ee8:5cfd with SMTP id c28csp792501pzh; Fri, 23 Sep 2022 16:20:45 -0700 (PDT) X-Google-Smtp-Source: AMsMyM5Ua1uw+WhrMBnboRgK+8cqI6nwHsN0qthhQT8YfX31cy4skO0cJbM74oS4mH/QK1aec5I6 X-Received: by 2002:a05:6402:5ca:b0:43b:6e01:482c with SMTP id n10-20020a05640205ca00b0043b6e01482cmr11099134edx.189.1663975245482; Fri, 23 Sep 2022 16:20:45 -0700 (PDT) ARC-Seal: i=1; a=rsa-sha256; t=1663975245; cv=none; d=google.com; s=arc-20160816; b=Kd8IpM4dEyxkX6AtO6zRT7sXlTqVLUkHBk46V0OFjLDBXXPRY7XgYCHQDB1+cyD6tP KtNc9LTAX38zAl3XJNDrAZ56ZOiKkis0UFruxbf9hAS6WJY/oKL6wflpAKPCrQTEVGWX hvMxHGSDkuW5Ia8+AFkCgjx1GsLGAoB365dTS5cDwzqdIfFU7vyZYYe4+culoyh9wuv2 rtkPXlVIIxp4j/ygspW21t0xQaaKM8/PToq4PSyQ6tMzNpmrlZcIdbHK6clPieQABdX8 utytWrsMxY5tAHzGAWCTFZjnQh45yeG3QwHaOiBDjPXZwDUWQDq0n4gKydMGlyb42kTr sXAg== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=sender:errors-to:reply-to:list-subscribe:list-help:list-post :list-archive:list-unsubscribe:list-id:precedence:subject :mime-version:references:in-reply-to:message-id:to:from:date :dkim-signature:delivered-to; bh=1G/j3ILkvsznGv1odq5fYcKqqYDAjtdXYTijiMKfAVQ=; b=MZOSMwfz0yndW7myrXE09QSyBzhux83A/ujPhiUdJtSahNNGNur9ArHzKKGH6aSLyR 9/R3HvUFzRpi033avkxNAWX5jGoe9Z4JwJIkzuXcf3zNkr9AJAs0amFHBvbq9CSaXIvK VklruYTSGy8+sVGuchr0GagBKD9X5zOsTb41Tx0P0UhPhN/gJRcTeDo6HCfUXndP9adt X5G5EIxpa7g+etIrPDk67Lw7x6fqR/fS331xYaQkgBi83jx+6V3PLBqjwvmRKw6U2x+8 2yp74y7ezJJwpNYCX5vPtJrIArR2nLVPrtt1eICZ5fl/Ef2ZCaCvCeFUla41KgEm0Tt7 Il4Q== ARC-Authentication-Results: i=1; mx.google.com; dkim=neutral (body hash did not verify) header.i=@lynne.ee header.s=s1 header.b=SrHR8+PQ; 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; dmarc=fail (p=NONE sp=NONE dis=NONE) header.from=lynne.ee Return-Path: Received: from ffbox0-bg.mplayerhq.hu (ffbox0-bg.ffmpeg.org. [79.124.17.100]) by mx.google.com with ESMTP id r8-20020a50aac8000000b0043d37773dbasi9369550edc.90.2022.09.23.16.20.45; Fri, 23 Sep 2022 16:20:45 -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; dkim=neutral (body hash did not verify) header.i=@lynne.ee header.s=s1 header.b=SrHR8+PQ; 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; dmarc=fail (p=NONE sp=NONE dis=NONE) header.from=lynne.ee Received: from [127.0.1.1] (localhost [127.0.0.1]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id 3B7C168BBB3; Sat, 24 Sep 2022 02:20:43 +0300 (EEST) X-Original-To: ffmpeg-devel@ffmpeg.org Delivered-To: ffmpeg-devel@ffmpeg.org Received: from w4.tutanota.de (w4.tutanota.de [81.3.6.165]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTPS id 8642268BB88 for ; Sat, 24 Sep 2022 02:20:36 +0300 (EEST) Received: from w3.tutanota.de (unknown [192.168.1.164]) by w4.tutanota.de (Postfix) with ESMTP id 3EB3A106015B for ; Fri, 23 Sep 2022 23:20:36 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; t=1663975236; s=s1; d=lynne.ee; h=From:From:To:To:Subject:Subject:Content-Description:Content-ID:Content-Type:Content-Type:Content-Transfer-Encoding:Cc:Date:Date:In-Reply-To:In-Reply-To:MIME-Version:MIME-Version:Message-ID:Message-ID:Reply-To:References:References:Sender; bh=EbJk1hbAnYrpt0+VbSW4NzRM2+aQepny5LHUMw+uk/w=; b=SrHR8+PQUma9abjCroxaYPFoctbC/9aT/KAqZ2eHC5z1gLjlQy9HVGwNOZXK1eup qmxs2kWwMja6TKpvdIGlgPC45BU+6NYHQ/A03/JtcRMg6QCdCKIaC1HhDfO2yhM7M93 IUuDNJS4isJgChpfxFVIAkAbhWO0RjHDriU1VN3vplA8L0E5e4bP9/t8GTZm75MQ0s0 /tN0bKtm+TwxrC2mugxBgB7RCk71QzLnAIZJnodZUMK5ZLGCUXj06265mWcfTxeUp6p b6T3RUlGDQ2yWhpXfwa35Oz8Oy1uviQ+JqUG0KR9f0Zfw4HLDl5wwhOq8u6FbBAuUu9 e1Y7nCIDyw== Date: Sat, 24 Sep 2022 01:20:36 +0200 (CEST) From: Lynne To: FFmpeg development discussions and patches Message-ID: In-Reply-To: References: MIME-Version: 1.0 Subject: [FFmpeg-devel] [PATCH 6/6] wmaprodec: convert to lavu/tx 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 Errors-To: ffmpeg-devel-bounces@ffmpeg.org Sender: "ffmpeg-devel" X-TUID: nhP4rPbdo53S Patch attached. From 4ad73f29065051c68991eb96aeae7f771039209a Mon Sep 17 00:00:00 2001 From: Lynne Date: Sat, 24 Sep 2022 01:08:00 +0200 Subject: [PATCH 6/6] wmaprodec: convert to lavu/tx --- libavcodec/wmaprodec.c | 24 ++++++++++++++---------- 1 file changed, 14 insertions(+), 10 deletions(-) diff --git a/libavcodec/wmaprodec.c b/libavcodec/wmaprodec.c index 701dfa955c..68e17e0743 100644 --- a/libavcodec/wmaprodec.c +++ b/libavcodec/wmaprodec.c @@ -89,6 +89,7 @@ #include #include "libavutil/audio_fifo.h" +#include "libavutil/tx.h" #include "libavutil/ffmath.h" #include "libavutil/float_dsp.h" #include "libavutil/intfloat.h" @@ -185,7 +186,8 @@ typedef struct WMAProDecodeCtx { uint8_t frame_data[MAX_FRAMESIZE + AV_INPUT_BUFFER_PADDING_SIZE];///< compressed frame data PutBitContext pb; ///< context for filling the frame_data buffer - FFTContext mdct_ctx[WMAPRO_BLOCK_SIZES]; ///< MDCT context per block size + AVTXContext *tx[WMAPRO_BLOCK_SIZES]; ///< MDCT context per block size + av_tx_fn tx_fn[WMAPRO_BLOCK_SIZES]; DECLARE_ALIGNED(32, float, tmp)[WMAPRO_BLOCK_MAX_SIZE]; ///< IMDCT output buffer const float* windows[WMAPRO_BLOCK_SIZES]; ///< windows for the different block sizes @@ -287,7 +289,7 @@ static av_cold int decode_end(WMAProDecodeCtx *s) av_freep(&s->fdsp); for (i = 0; i < WMAPRO_BLOCK_SIZES; i++) - ff_mdct_end(&s->mdct_ctx[i]); + av_tx_uninit(&s->tx[i]); return 0; } @@ -552,12 +554,13 @@ static av_cold int decode_init(WMAProDecodeCtx *s, AVCodecContext *avctx, int nu return AVERROR(ENOMEM); /** init MDCT, FIXME: only init needed sizes */ - for (int i = 0; i < WMAPRO_BLOCK_SIZES; i++) { - ret = ff_mdct_init(&s->mdct_ctx[i], WMAPRO_BLOCK_MIN_BITS + 1 + i, 1, - 1.0 / (1 << (WMAPRO_BLOCK_MIN_BITS + i - 1)) - / (1ll << (s->bits_per_sample - 1))); - if (ret < 0) - return ret; + for (i = 0; i < WMAPRO_BLOCK_SIZES; i++) { + const float scale = 1.0 / (1 << (WMAPRO_BLOCK_MIN_BITS + i - 1)) + / (1ll << (s->bits_per_sample - 1)); + int err = av_tx_init(&s->tx[i], &s->tx_fn[i], AV_TX_FLOAT_MDCT, 1, + 1 << (WMAPRO_BLOCK_MIN_BITS + i), &scale, 0); + if (err < 0) + return err; } /** init MDCT windows: simple sine window */ @@ -1386,7 +1389,8 @@ static int decode_subframe(WMAProDecodeCtx *s) get_bits_count(&s->gb) - s->subframe_offset); if (transmit_coeffs) { - FFTContext *mdct = &s->mdct_ctx[av_log2(subframe_len) - WMAPRO_BLOCK_MIN_BITS]; + AVTXContext *tx = s->tx[av_log2(subframe_len) - WMAPRO_BLOCK_MIN_BITS]; + av_tx_fn tx_fn = s->tx_fn[av_log2(subframe_len) - WMAPRO_BLOCK_MIN_BITS]; /** reconstruct the per channel data */ inverse_channel_transform(s); for (i = 0; i < s->channels_for_cur_subframe; i++) { @@ -1412,7 +1416,7 @@ static int decode_subframe(WMAProDecodeCtx *s) } /** apply imdct (imdct_half == DCTIV with reverse) */ - mdct->imdct_half(mdct, s->channel[c].coeffs, s->tmp); + tx_fn(tx, s->channel[c].coeffs, s->tmp, sizeof(float)); } } -- 2.37.2.609.g9ff673ca1a