From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from ffbox0-bg.mplayerhq.hu (ffbox0-bg.ffmpeg.org [79.124.17.100]) by master.gitmailbox.com (Postfix) with ESMTP id 3717743098 for ; Wed, 20 Jul 2022 04:41:39 +0000 (UTC) Received: from [127.0.1.1] (localhost [127.0.0.1]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id 4227168B813; Wed, 20 Jul 2022 07:41:30 +0300 (EEST) Received: from mail-pg1-f177.google.com (mail-pg1-f177.google.com [209.85.215.177]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTPS id 14CC368B7CC for ; Wed, 20 Jul 2022 07:41:24 +0300 (EEST) Received: by mail-pg1-f177.google.com with SMTP id r186so15362349pgr.2 for ; Tue, 19 Jul 2022 21:41:23 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20210112; h=from:to:cc:subject:date:message-id:in-reply-to:references :mime-version:content-transfer-encoding; bh=B9B0QDrFUjTO8dgges6mFHrfz1CKo0jCmRo91X5zRTk=; b=jKG1pTmq/j54S+tcVikjZ7KTknAgdJdKVUdpp3QXVRpZ/659NiNst0qWvHYuqsovN6 frf+PMiD98V4VOxq6npZgL6JYTp+9l7efNPNkb7Pfxx5W2jpsqXaiAnFJ/ugA9xNeWby gfbNjYTvirGawvLen41ohK2TeGSCeXUAnEvIesuiTvqQi/T5XQ0qK1axYK+Vt3M3Bis8 uFQ7cLa25nnVQVgbTStybfJb7DY2imSo6JOMrn6XfV0KAhc2bn6TkzpnwCWhis6PdqMK g+2zQi3T8phChcpsYiZP3EtbByvSeOZFkX4ZnfgyZtandi/n9d/lLy2T32ji8L17VCuo BLxw== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20210112; h=x-gm-message-state:from:to:cc:subject:date:message-id:in-reply-to :references:mime-version:content-transfer-encoding; bh=B9B0QDrFUjTO8dgges6mFHrfz1CKo0jCmRo91X5zRTk=; b=iOnr2Mfs/GTvrNOLv6VaclCwcMRKSXMD0N9dXMnQ6ZkBrv2X1Udqhrj/N+bKx2apkt SAohcmr59BjJs6xQfUXpkuDHmQ193HOZ6tojoxamOlMVgwZ87qIC5vxz3eFHMnLy/nL9 ah20Vaksv1SowJkYIY7p/4FrlK42OKLckhXBMACyMLaB9imj6KhJIybhu8q32YTYx/fa OcabNQOwJu96MKvO7EHzjrl8lrYx69NO2dR/rDjIfRh5W2HAfTSvOEtkON9LJhscLn1c RygGxzkmbLFDkDefmDO+U8Kj36zsFmkveC3B02FkBDQU6Vejm9N7w9SnsO+39v+4Wwaf dNYg== X-Gm-Message-State: AJIora91iGgX9P/q8nZjOEUj5GtVmFG1qz4smZfbzA1akDu8nTOP4rOv dVbdCbr/HSsr3y3T4Xfrjg/JUEOEXJXMAg== X-Google-Smtp-Source: AGRyM1ttBROCV72fcAoUADTFpJ9R/lFG9Lmv+9VuoEnYS28M/+SaWbmvF7ODmcOd1b+2OW7esfKfFQ== X-Received: by 2002:a63:688a:0:b0:412:6728:4bf3 with SMTP id d132-20020a63688a000000b0041267284bf3mr32569746pgc.339.1658292082108; Tue, 19 Jul 2022 21:41:22 -0700 (PDT) Received: from localhost.localdomain (23-121-159-29.lightspeed.sntcca.sbcglobal.net. [23.121.159.29]) by smtp.googlemail.com with ESMTPSA id f16-20020a635110000000b003fba1a97c49sm10855907pgb.61.2022.07.19.21.41.21 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Tue, 19 Jul 2022 21:41:21 -0700 (PDT) From: Chris Phlipot To: ffmpeg-devel@ffmpeg.org Date: Tue, 19 Jul 2022 21:41:14 -0700 Message-Id: <20220720044117.1282961-2-cphlipot0@gmail.com> X-Mailer: git-send-email 2.25.1 In-Reply-To: <20220720044117.1282961-1-cphlipot0@gmail.com> References: <20220720044117.1282961-1-cphlipot0@gmail.com> MIME-Version: 1.0 Subject: [FFmpeg-devel] [PATCH 2/5] avfilter/vf_yadif: Allow alignment to be configurable 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 Cc: Chris Phlipot Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Errors-To: ffmpeg-devel-bounces@ffmpeg.org Sender: "ffmpeg-devel" Archived-At: List-Archive: List-Post: Allow the alignment to be determined based on what yadif_filter_line implementation is used. Currently this is either 1, or 8 depending on whether the C code or the x86 SSE code is used, but allows for other future implementations that use a larger alignment. Adjusting MAX_ALIGN to 32 in the case of an AVX2 implementation could potentially hurt the performance of the SSE implementation, so we allow yadif to use the smallest needed alignment instead to maintain existing performance if implementations with wider vectors are added. Signed-off-by: Chris Phlipot --- libavfilter/vf_yadif.c | 16 +++++++++------- libavfilter/x86/vf_yadif_init.c | 1 + libavfilter/yadif.h | 4 +++- 3 files changed, 13 insertions(+), 8 deletions(-) diff --git a/libavfilter/vf_yadif.c b/libavfilter/vf_yadif.c index 055327d7a4..42f6246330 100644 --- a/libavfilter/vf_yadif.c +++ b/libavfilter/vf_yadif.c @@ -108,9 +108,9 @@ static void filter_line_c(void *dst1, FILTER(0, w, 1) } -#define MAX_ALIGN 8 static void filter_edges(void *dst1, void *prev1, void *cur1, void *next1, - int w, int prefs, int mrefs, int parity, int mode) + int w, int prefs, int mrefs, int parity, int mode, + int alignment) { uint8_t *dst = dst1; uint8_t *prev = prev1; @@ -120,7 +120,7 @@ static void filter_edges(void *dst1, void *prev1, void *cur1, void *next1, uint8_t *prev2 = parity ? prev : cur ; uint8_t *next2 = parity ? cur : next; - const int edge = FFMAX(MAX_ALIGN - 1, 3); + const int edge = FFMAX(alignment - 1, 3); int offset = FFMAX(w - edge, 3); /* Only edge pixels need to be processed here. A constant value of false @@ -159,7 +159,8 @@ static void filter_line_c_16bit(void *dst1, } static void filter_edges_16bit(void *dst1, void *prev1, void *cur1, void *next1, - int w, int prefs, int mrefs, int parity, int mode) + int w, int prefs, int mrefs, int parity, int mode, + int alignment) { uint16_t *dst = dst1; uint16_t *prev = prev1; @@ -169,7 +170,7 @@ static void filter_edges_16bit(void *dst1, void *prev1, void *cur1, void *next1, uint16_t *prev2 = parity ? prev : cur ; uint16_t *next2 = parity ? cur : next; - const int edge = FFMAX(MAX_ALIGN / 2 - 1, 3); + const int edge = FFMAX(alignment / 2 - 1, 3); int offset = FFMAX(w - edge, 3); mrefs /= 2; @@ -199,7 +200,7 @@ static int filter_slice(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs) int slice_start = (td->h * jobnr ) / nb_jobs; int slice_end = (td->h * (jobnr+1)) / nb_jobs; int y; - int edge = 3 + MAX_ALIGN / df - 1; + int edge = 3 + s->req_align / df - 1; /* filtering reads 3 pixels to the left/right; to avoid invalid reads, * we need to call the c variant which avoids this for border pixels @@ -219,7 +220,7 @@ static int filter_slice(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs) s->filter_edges(dst, prev, cur, next, td->w, y + 1 < td->h ? refs : -refs, y ? -refs : refs, - td->parity ^ td->tff, mode); + td->parity ^ td->tff, mode, s->req_align); } else { memcpy(&td->frame->data[td->plane][y * td->frame->linesize[td->plane]], &s->cur->data[td->plane][y * refs], td->w * df); @@ -303,6 +304,7 @@ static int config_output(AVFilterLink *outlink) s->csp = av_pix_fmt_desc_get(outlink->format); s->filter = filter; + s->req_align = 1; if (s->csp->comp[0].depth > 8) { s->filter_line = filter_line_c_16bit; s->filter_edges = filter_edges_16bit; diff --git a/libavfilter/x86/vf_yadif_init.c b/libavfilter/x86/vf_yadif_init.c index 257c3f9199..9dd73f8e44 100644 --- a/libavfilter/x86/vf_yadif_init.c +++ b/libavfilter/x86/vf_yadif_init.c @@ -53,6 +53,7 @@ av_cold void ff_yadif_init_x86(YADIFContext *yadif) int bit_depth = (!yadif->csp) ? 8 : yadif->csp->comp[0].depth; + yadif->req_align = 8; if (bit_depth >= 15) { if (EXTERNAL_SSE2(cpu_flags)) yadif->filter_line = ff_yadif_filter_line_16bit_sse2; diff --git a/libavfilter/yadif.h b/libavfilter/yadif.h index c928911b35..b81f2fc1d9 100644 --- a/libavfilter/yadif.h +++ b/libavfilter/yadif.h @@ -66,11 +66,13 @@ typedef struct YADIFContext { /** * Required alignment for filter_line */ + int req_align; void (*filter_line)(void *dst, void *prev, void *cur, void *next, int w, int prefs, int mrefs, int parity, int mode); void (*filter_edges)(void *dst, void *prev, void *cur, void *next, - int w, int prefs, int mrefs, int parity, int mode); + int w, int prefs, int mrefs, int parity, int mode, + int alignment); const AVPixFmtDescriptor *csp; int eof; -- 2.25.1 _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".