From patchwork Fri Jan 19 13:38:14 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Wu Jianhua X-Patchwork-Id: 45660 Delivered-To: ffmpegpatchwork2@gmail.com Received: by 2002:a05:6a20:968f:b0:199:de12:6fa6 with SMTP id hp15csp942697pzc; Fri, 19 Jan 2024 05:39:29 -0800 (PST) X-Google-Smtp-Source: AGHT+IGAdCMpImdntNt0RvnbBhGXV1XbqgVO8FCYq8qydDdGDVfrjMcvmoeTVPJCX48dO2K7zmgJ X-Received: by 2002:a05:6402:3414:b0:55a:4d55:364f with SMTP id k20-20020a056402341400b0055a4d55364fmr745891edc.20.1705671569640; Fri, 19 Jan 2024 05:39:29 -0800 (PST) Return-Path: Received: from ffbox0-bg.mplayerhq.hu (ffbox0-bg.ffmpeg.org. [79.124.17.100]) by mx.google.com with ESMTP id f12-20020a056402194c00b00559e219478fsi2655904edz.143.2024.01.19.05.39.29; Fri, 19 Jan 2024 05:39:29 -0800 (PST) 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=@outlook.com header.s=selector1 header.b=A2VFplDn; arc=fail (body hash mismatch); 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=QUARANTINE dis=NONE) header.from=outlook.com Received: from [127.0.1.1] (localhost [127.0.0.1]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id 19CC268D063; Fri, 19 Jan 2024 15:39:23 +0200 (EET) X-Original-To: ffmpeg-devel@ffmpeg.org Delivered-To: ffmpeg-devel@ffmpeg.org Received: from JPN01-OS0-obe.outbound.protection.outlook.com (mail-os0jpn01olkn2102.outbound.protection.outlook.com [40.92.98.102]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTPS id C255E68D054 for ; Fri, 19 Jan 2024 15:39:15 +0200 (EET) ARC-Seal: i=1; a=rsa-sha256; s=arcselector9901; d=microsoft.com; cv=none; b=JgHxj1cwYSFGg6pggVjyw4iFqih2WvzwL/OtzUCiA3MqfRlii4z+oJLQ5iOpBwT0747Jf9hfekzIlE1uQvprZrgoRKdmAiWTpnWMsxlYhA408wJ+LjL8TupljL8kPOtLMpb6xZIJerJujgvhEBndtiVKFIzR16rft0CmHXYLtV/cdLvQMF+cVXd85XDvFah0c/D8MraxsUfS+I/xTVLfFnZnfv5zSkV2JiQ/uKgnGXAA8XlA810LqiU67ZTOWy4CQrx3h9U6IPA8IwNf4LvvRmrvQp/GwL1wb7zCt63aptbyLLINPSlDT0FPvwNTfj2exw6sIshJYluyE6ppH3gjCA== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=microsoft.com; s=arcselector9901; h=From:Date:Subject:Message-ID:Content-Type:MIME-Version:X-MS-Exchange-AntiSpam-MessageData-ChunkCount:X-MS-Exchange-AntiSpam-MessageData-0:X-MS-Exchange-AntiSpam-MessageData-1; bh=o9p+JWkf2/gi0FVu/oZcjQ/tbdr260lXxz8jg8iDHco=; b=NzVMZ9wgA2sWAVkEYTbRrUPZXEFLR0tfWUPEcTLlKBmUxp7ePUaY2Qb+0mhOCLCCe3q9LtBad5qXCb0oGLhT3hMETRSk/IbNMfCj4xZElCk9OOGEziPA2cdXJR7Zy1imWW58mh4ex8m7nrOmIIYMi+6j809cV4Su9uHY/F87LnNnKFJ8SBFXjn+RR2eEakgYkoW5HqyQPTWgPFwJTSCDfrshsLeRFzk/OkjgfzVR/TwBRLmvKQZwSvbpLxZv/whtvlS9N1EjSGSfc/GIaY39uwSulKUMi4ld8JoIzNyj7jOLEf22W+2fiuw3SEeL36LX1Vy3ZrxqhOBqAl5T3q1qPQ== ARC-Authentication-Results: i=1; mx.microsoft.com 1; spf=none; dmarc=none; dkim=none; arc=none DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=outlook.com; s=selector1; h=From:Date:Subject:Message-ID:Content-Type:MIME-Version:X-MS-Exchange-SenderADCheck; bh=o9p+JWkf2/gi0FVu/oZcjQ/tbdr260lXxz8jg8iDHco=; b=A2VFplDniVrDvLvXxNghi1p5CU9MeR1iorv3yiDGBQ1zoe+lyqhCi6qCEpyQQ8OrO9su8phLpq2QZFAo9aZj2CCdrd9up4nmt7zchGUlGxfWnbHYz1gLi7NT94Cb4pJM7ahLSpK3ZvcQTe07mcO93t4H8WkAGzdkakTnVSYIuZUP/woqlZ4vZXelbWnVyIy973AdtAMKotGIOFQC4BSMulESV2i8V+p9TyDxruSSNEOOxlqAGb9MF86w9jso2aMKoLS0HFCJmWfyFctN6wR8m99ndTJh39uQnWflvGTT1EhG77Iz5lnJ9nhwKK3vc+WQiAofPioRc0UuL/T0XGMh2w== Received: from OSZP286MB2173.JPNP286.PROD.OUTLOOK.COM (2603:1096:604:186::5) by OS3P286MB2523.JPNP286.PROD.OUTLOOK.COM (2603:1096:604:1ef::9) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384) id 15.20.7202.24; Fri, 19 Jan 2024 13:39:12 +0000 Received: from OSZP286MB2173.JPNP286.PROD.OUTLOOK.COM ([fe80::1bbf:406b:216:f56e]) by OSZP286MB2173.JPNP286.PROD.OUTLOOK.COM ([fe80::1bbf:406b:216:f56e%7]) with mapi id 15.20.7202.024; Fri, 19 Jan 2024 13:39:12 +0000 From: toqsxw@outlook.com To: ffmpeg-devel@ffmpeg.org Date: Fri, 19 Jan 2024 21:38:14 +0800 Message-ID: X-Mailer: git-send-email 2.34.1 In-Reply-To: <20240119133820.1048730-1-toqsxw@outlook.com> References: <20240119133820.1048730-1-toqsxw@outlook.com> X-TMN: [FDGIROQoBszuIEdS1AAcP9PtCYkx8WB5] X-ClientProxiedBy: SG2PR02CA0137.apcprd02.prod.outlook.com (2603:1096:4:188::17) To OSZP286MB2173.JPNP286.PROD.OUTLOOK.COM (2603:1096:604:186::5) X-Microsoft-Original-Message-ID: <20240119133820.1048730-2-toqsxw@outlook.com> MIME-Version: 1.0 X-MS-Exchange-MessageSentRepresentingType: 1 X-MS-PublicTrafficType: Email X-MS-TrafficTypeDiagnostic: OSZP286MB2173:EE_|OS3P286MB2523:EE_ X-MS-Office365-Filtering-Correlation-Id: 124aa786-b880-4429-5ba2-08dc18f40470 X-Microsoft-Antispam: BCL:0; X-Microsoft-Antispam-Message-Info: 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 X-MS-Exchange-AntiSpam-MessageData-ChunkCount: 1 X-MS-Exchange-AntiSpam-MessageData-0: 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 X-OriginatorOrg: outlook.com X-MS-Exchange-CrossTenant-Network-Message-Id: 124aa786-b880-4429-5ba2-08dc18f40470 X-MS-Exchange-CrossTenant-AuthSource: OSZP286MB2173.JPNP286.PROD.OUTLOOK.COM X-MS-Exchange-CrossTenant-AuthAs: Internal X-MS-Exchange-CrossTenant-OriginalArrivalTime: 19 Jan 2024 13:39:12.2928 (UTC) X-MS-Exchange-CrossTenant-FromEntityHeader: Hosted X-MS-Exchange-CrossTenant-Id: 84df9e7f-e9f6-40af-b435-aaaaaaaaaaaa X-MS-Exchange-CrossTenant-RMS-PersistedConsumerOrg: 00000000-0000-0000-0000-000000000000 X-MS-Exchange-Transport-CrossTenantHeadersStamped: OS3P286MB2523 Subject: [FFmpeg-devel] [PATCH v2 2/8] avcodec/hevcdsp_template: reuse put/put_luma/put_chroma from h2656_inter_template 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: Wu Jianhua Errors-To: ffmpeg-devel-bounces@ffmpeg.org Sender: "ffmpeg-devel" X-TUID: NhZpk/ffocUt From: Wu Jianhua Signed-off-by: Wu Jianhua --- libavcodec/hevcdsp_template.c | 594 +++------------------------------- 1 file changed, 46 insertions(+), 548 deletions(-) diff --git a/libavcodec/hevcdsp_template.c b/libavcodec/hevcdsp_template.c index 0de14e9dcf..9b48bdf08e 100644 --- a/libavcodec/hevcdsp_template.c +++ b/libavcodec/hevcdsp_template.c @@ -26,6 +26,7 @@ #include "bit_depth_template.c" #include "hevcdsp.h" #include "h26x/h2656_sao_template.c" +#include "h26x/h2656_inter_template.c" static void FUNC(put_pcm)(uint8_t *_dst, ptrdiff_t stride, int width, int height, GetBitContext *gb, int pcm_bit_depth) @@ -299,37 +300,51 @@ IDCT_DC(32) //////////////////////////////////////////////////////////////////////////////// // //////////////////////////////////////////////////////////////////////////////// -static void FUNC(put_hevc_pel_pixels)(int16_t *dst, - const uint8_t *_src, ptrdiff_t _srcstride, - int height, intptr_t mx, intptr_t my, int width) -{ - int x, y; - const pixel *src = (const pixel *)_src; - ptrdiff_t srcstride = _srcstride / sizeof(pixel); - - for (y = 0; y < height; y++) { - for (x = 0; x < width; x++) - dst[x] = src[x] << (14 - BIT_DEPTH); - src += srcstride; - dst += MAX_PB_SIZE; - } -} - -static void FUNC(put_hevc_pel_uni_pixels)(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, - int height, intptr_t mx, intptr_t my, int width) -{ - int y; - const pixel *src = (const pixel *)_src; - ptrdiff_t srcstride = _srcstride / sizeof(pixel); - pixel *dst = (pixel *)_dst; - ptrdiff_t dststride = _dststride / sizeof(pixel); - - for (y = 0; y < height; y++) { - memcpy(dst, src, width * sizeof(pixel)); - src += srcstride; - dst += dststride; - } -} +#define ff_hevc_pel_filters ff_hevc_qpel_filters +#define DECL_HV_FILTER(f) \ + const uint8_t *hf = ff_hevc_ ## f ## _filters[mx - 1]; \ + const uint8_t *vf = ff_hevc_ ## f ## _filters[my - 1]; + +#define FW_PUT(p, f, t) \ +static void FUNC(put_hevc_## f)(int16_t *dst, const uint8_t *src, ptrdiff_t srcstride, int height, \ + intptr_t mx, intptr_t my, int width) \ +{ \ + DECL_HV_FILTER(p) \ + FUNC(put_ ## t)(dst, src, srcstride, height, hf, vf, width); \ +} + +#define FW_PUT_UNI(p, f, t) \ +static void FUNC(put_hevc_ ## f)(uint8_t *dst, ptrdiff_t dststride, const uint8_t *src, \ + ptrdiff_t srcstride, int height, intptr_t mx, intptr_t my, int width) \ +{ \ + DECL_HV_FILTER(p) \ + FUNC(put_ ## t)(dst, dststride, src, srcstride, height, hf, vf, width); \ +} + +#define FW_PUT_UNI_W(p, f, t) \ +static void FUNC(put_hevc_ ## f)(uint8_t *dst, ptrdiff_t dststride, const uint8_t *src, \ + ptrdiff_t srcstride,int height, int denom, int wx, int ox, \ + intptr_t mx, intptr_t my, int width) \ +{ \ + DECL_HV_FILTER(p) \ + FUNC(put_ ## t)(dst, dststride, src, srcstride, height, denom, wx, ox, hf, vf, width); \ +} + +#define FW_PUT_FUNCS(f, t, dir) \ + FW_PUT(f, f ## _ ## dir, t ## _ ## dir) \ + FW_PUT_UNI(f, f ## _uni_ ## dir, uni_ ## t ## _ ## dir) \ + FW_PUT_UNI_W(f, f ## _uni_w_ ## dir, uni_## t ## _w_ ## dir) + +FW_PUT(pel, pel_pixels, pixels) +FW_PUT_UNI(pel, pel_uni_pixels, uni_pixels) +FW_PUT_UNI_W(pel, pel_uni_w_pixels, uni_w_pixels) + +FW_PUT_FUNCS(qpel, luma, h ) +FW_PUT_FUNCS(qpel, luma, v ) +FW_PUT_FUNCS(qpel, luma, hv ) +FW_PUT_FUNCS(epel, chroma, h ) +FW_PUT_FUNCS(epel, chroma, v ) +FW_PUT_FUNCS(epel, chroma, hv ) static void FUNC(put_hevc_pel_bi_pixels)(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, const int16_t *src2, @@ -357,30 +372,6 @@ static void FUNC(put_hevc_pel_bi_pixels)(uint8_t *_dst, ptrdiff_t _dststride, co } } -static void FUNC(put_hevc_pel_uni_w_pixels)(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, - int height, int denom, int wx, int ox, intptr_t mx, intptr_t my, int width) -{ - int x, y; - const pixel *src = (const pixel *)_src; - ptrdiff_t srcstride = _srcstride / sizeof(pixel); - pixel *dst = (pixel *)_dst; - ptrdiff_t dststride = _dststride / sizeof(pixel); - int shift = denom + 14 - BIT_DEPTH; -#if BIT_DEPTH < 14 - int offset = 1 << (shift - 1); -#else - int offset = 0; -#endif - - ox = ox * (1 << (BIT_DEPTH - 8)); - for (y = 0; y < height; y++) { - for (x = 0; x < width; x++) - dst[x] = av_clip_pixel((((src[x] << (14 - BIT_DEPTH)) * wx + offset) >> shift) + ox); - src += srcstride; - dst += dststride; - } -} - static void FUNC(put_hevc_pel_bi_w_pixels)(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, const int16_t *src2, int height, int denom, int wx0, int wx1, @@ -420,96 +411,6 @@ static void FUNC(put_hevc_pel_bi_w_pixels)(uint8_t *_dst, ptrdiff_t _dststride, filter[6] * src[x + 3 * stride] + \ filter[7] * src[x + 4 * stride]) -static void FUNC(put_hevc_qpel_h)(int16_t *dst, - const uint8_t *_src, ptrdiff_t _srcstride, - int height, intptr_t mx, intptr_t my, int width) -{ - int x, y; - const pixel *src = (const pixel*)_src; - ptrdiff_t srcstride = _srcstride / sizeof(pixel); - const int8_t *filter = ff_hevc_qpel_filters[mx - 1]; - for (y = 0; y < height; y++) { - for (x = 0; x < width; x++) - dst[x] = QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8); - src += srcstride; - dst += MAX_PB_SIZE; - } -} - -static void FUNC(put_hevc_qpel_v)(int16_t *dst, - const uint8_t *_src, ptrdiff_t _srcstride, - int height, intptr_t mx, intptr_t my, int width) -{ - int x, y; - const pixel *src = (const pixel*)_src; - ptrdiff_t srcstride = _srcstride / sizeof(pixel); - const int8_t *filter = ff_hevc_qpel_filters[my - 1]; - for (y = 0; y < height; y++) { - for (x = 0; x < width; x++) - dst[x] = QPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8); - src += srcstride; - dst += MAX_PB_SIZE; - } -} - -static void FUNC(put_hevc_qpel_hv)(int16_t *dst, - const uint8_t *_src, - ptrdiff_t _srcstride, - int height, intptr_t mx, - intptr_t my, int width) -{ - int x, y; - const int8_t *filter; - const pixel *src = (const pixel*)_src; - ptrdiff_t srcstride = _srcstride / sizeof(pixel); - int16_t tmp_array[(MAX_PB_SIZE + QPEL_EXTRA) * MAX_PB_SIZE]; - int16_t *tmp = tmp_array; - - src -= QPEL_EXTRA_BEFORE * srcstride; - filter = ff_hevc_qpel_filters[mx - 1]; - for (y = 0; y < height + QPEL_EXTRA; y++) { - for (x = 0; x < width; x++) - tmp[x] = QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8); - src += srcstride; - tmp += MAX_PB_SIZE; - } - - tmp = tmp_array + QPEL_EXTRA_BEFORE * MAX_PB_SIZE; - filter = ff_hevc_qpel_filters[my - 1]; - for (y = 0; y < height; y++) { - for (x = 0; x < width; x++) - dst[x] = QPEL_FILTER(tmp, MAX_PB_SIZE) >> 6; - tmp += MAX_PB_SIZE; - dst += MAX_PB_SIZE; - } -} - -static void FUNC(put_hevc_qpel_uni_h)(uint8_t *_dst, ptrdiff_t _dststride, - const uint8_t *_src, ptrdiff_t _srcstride, - int height, intptr_t mx, intptr_t my, int width) -{ - int x, y; - const pixel *src = (const pixel*)_src; - ptrdiff_t srcstride = _srcstride / sizeof(pixel); - pixel *dst = (pixel *)_dst; - ptrdiff_t dststride = _dststride / sizeof(pixel); - const int8_t *filter = ff_hevc_qpel_filters[mx - 1]; - int shift = 14 - BIT_DEPTH; - -#if BIT_DEPTH < 14 - int offset = 1 << (shift - 1); -#else - int offset = 0; -#endif - - for (y = 0; y < height; y++) { - for (x = 0; x < width; x++) - dst[x] = av_clip_pixel(((QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) + offset) >> shift); - src += srcstride; - dst += dststride; - } -} - static void FUNC(put_hevc_qpel_bi_h)(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, const int16_t *src2, int height, intptr_t mx, intptr_t my, int width) @@ -538,33 +439,6 @@ static void FUNC(put_hevc_qpel_bi_h)(uint8_t *_dst, ptrdiff_t _dststride, const } } -static void FUNC(put_hevc_qpel_uni_v)(uint8_t *_dst, ptrdiff_t _dststride, - const uint8_t *_src, ptrdiff_t _srcstride, - int height, intptr_t mx, intptr_t my, int width) -{ - int x, y; - const pixel *src = (const pixel*)_src; - ptrdiff_t srcstride = _srcstride / sizeof(pixel); - pixel *dst = (pixel *)_dst; - ptrdiff_t dststride = _dststride / sizeof(pixel); - const int8_t *filter = ff_hevc_qpel_filters[my - 1]; - int shift = 14 - BIT_DEPTH; - -#if BIT_DEPTH < 14 - int offset = 1 << (shift - 1); -#else - int offset = 0; -#endif - - for (y = 0; y < height; y++) { - for (x = 0; x < width; x++) - dst[x] = av_clip_pixel(((QPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) + offset) >> shift); - src += srcstride; - dst += dststride; - } -} - - static void FUNC(put_hevc_qpel_bi_v)(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, const int16_t *src2, int height, intptr_t mx, intptr_t my, int width) @@ -593,46 +467,6 @@ static void FUNC(put_hevc_qpel_bi_v)(uint8_t *_dst, ptrdiff_t _dststride, } } -static void FUNC(put_hevc_qpel_uni_hv)(uint8_t *_dst, ptrdiff_t _dststride, - const uint8_t *_src, ptrdiff_t _srcstride, - int height, intptr_t mx, intptr_t my, int width) -{ - int x, y; - const int8_t *filter; - const pixel *src = (const pixel*)_src; - ptrdiff_t srcstride = _srcstride / sizeof(pixel); - pixel *dst = (pixel *)_dst; - ptrdiff_t dststride = _dststride / sizeof(pixel); - int16_t tmp_array[(MAX_PB_SIZE + QPEL_EXTRA) * MAX_PB_SIZE]; - int16_t *tmp = tmp_array; - int shift = 14 - BIT_DEPTH; - -#if BIT_DEPTH < 14 - int offset = 1 << (shift - 1); -#else - int offset = 0; -#endif - - src -= QPEL_EXTRA_BEFORE * srcstride; - filter = ff_hevc_qpel_filters[mx - 1]; - for (y = 0; y < height + QPEL_EXTRA; y++) { - for (x = 0; x < width; x++) - tmp[x] = QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8); - src += srcstride; - tmp += MAX_PB_SIZE; - } - - tmp = tmp_array + QPEL_EXTRA_BEFORE * MAX_PB_SIZE; - filter = ff_hevc_qpel_filters[my - 1]; - - for (y = 0; y < height; y++) { - for (x = 0; x < width; x++) - dst[x] = av_clip_pixel(((QPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) + offset) >> shift); - tmp += MAX_PB_SIZE; - dst += dststride; - } -} - static void FUNC(put_hevc_qpel_bi_hv)(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, const int16_t *src2, int height, intptr_t mx, intptr_t my, int width) @@ -673,33 +507,6 @@ static void FUNC(put_hevc_qpel_bi_hv)(uint8_t *_dst, ptrdiff_t _dststride, } } -static void FUNC(put_hevc_qpel_uni_w_h)(uint8_t *_dst, ptrdiff_t _dststride, - const uint8_t *_src, ptrdiff_t _srcstride, - int height, int denom, int wx, int ox, - intptr_t mx, intptr_t my, int width) -{ - int x, y; - const pixel *src = (const pixel*)_src; - ptrdiff_t srcstride = _srcstride / sizeof(pixel); - pixel *dst = (pixel *)_dst; - ptrdiff_t dststride = _dststride / sizeof(pixel); - const int8_t *filter = ff_hevc_qpel_filters[mx - 1]; - int shift = denom + 14 - BIT_DEPTH; -#if BIT_DEPTH < 14 - int offset = 1 << (shift - 1); -#else - int offset = 0; -#endif - - ox = ox * (1 << (BIT_DEPTH - 8)); - for (y = 0; y < height; y++) { - for (x = 0; x < width; x++) - dst[x] = av_clip_pixel((((QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) * wx + offset) >> shift) + ox); - src += srcstride; - dst += dststride; - } -} - static void FUNC(put_hevc_qpel_bi_w_h)(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, const int16_t *src2, int height, int denom, int wx0, int wx1, @@ -728,33 +535,6 @@ static void FUNC(put_hevc_qpel_bi_w_h)(uint8_t *_dst, ptrdiff_t _dststride, } } -static void FUNC(put_hevc_qpel_uni_w_v)(uint8_t *_dst, ptrdiff_t _dststride, - const uint8_t *_src, ptrdiff_t _srcstride, - int height, int denom, int wx, int ox, - intptr_t mx, intptr_t my, int width) -{ - int x, y; - const pixel *src = (const pixel*)_src; - ptrdiff_t srcstride = _srcstride / sizeof(pixel); - pixel *dst = (pixel *)_dst; - ptrdiff_t dststride = _dststride / sizeof(pixel); - const int8_t *filter = ff_hevc_qpel_filters[my - 1]; - int shift = denom + 14 - BIT_DEPTH; -#if BIT_DEPTH < 14 - int offset = 1 << (shift - 1); -#else - int offset = 0; -#endif - - ox = ox * (1 << (BIT_DEPTH - 8)); - for (y = 0; y < height; y++) { - for (x = 0; x < width; x++) - dst[x] = av_clip_pixel((((QPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) * wx + offset) >> shift) + ox); - src += srcstride; - dst += dststride; - } -} - static void FUNC(put_hevc_qpel_bi_w_v)(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, const int16_t *src2, int height, int denom, int wx0, int wx1, @@ -783,47 +563,6 @@ static void FUNC(put_hevc_qpel_bi_w_v)(uint8_t *_dst, ptrdiff_t _dststride, } } -static void FUNC(put_hevc_qpel_uni_w_hv)(uint8_t *_dst, ptrdiff_t _dststride, - const uint8_t *_src, ptrdiff_t _srcstride, - int height, int denom, int wx, int ox, - intptr_t mx, intptr_t my, int width) -{ - int x, y; - const int8_t *filter; - const pixel *src = (const pixel*)_src; - ptrdiff_t srcstride = _srcstride / sizeof(pixel); - pixel *dst = (pixel *)_dst; - ptrdiff_t dststride = _dststride / sizeof(pixel); - int16_t tmp_array[(MAX_PB_SIZE + QPEL_EXTRA) * MAX_PB_SIZE]; - int16_t *tmp = tmp_array; - int shift = denom + 14 - BIT_DEPTH; -#if BIT_DEPTH < 14 - int offset = 1 << (shift - 1); -#else - int offset = 0; -#endif - - src -= QPEL_EXTRA_BEFORE * srcstride; - filter = ff_hevc_qpel_filters[mx - 1]; - for (y = 0; y < height + QPEL_EXTRA; y++) { - for (x = 0; x < width; x++) - tmp[x] = QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8); - src += srcstride; - tmp += MAX_PB_SIZE; - } - - tmp = tmp_array + QPEL_EXTRA_BEFORE * MAX_PB_SIZE; - filter = ff_hevc_qpel_filters[my - 1]; - - ox = ox * (1 << (BIT_DEPTH - 8)); - for (y = 0; y < height; y++) { - for (x = 0; x < width; x++) - dst[x] = av_clip_pixel((((QPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) * wx + offset) >> shift) + ox); - tmp += MAX_PB_SIZE; - dst += dststride; - } -} - static void FUNC(put_hevc_qpel_bi_w_hv)(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, const int16_t *src2, int height, int denom, int wx0, int wx1, @@ -873,94 +612,6 @@ static void FUNC(put_hevc_qpel_bi_w_hv)(uint8_t *_dst, ptrdiff_t _dststride, filter[2] * src[x + stride] + \ filter[3] * src[x + 2 * stride]) -static void FUNC(put_hevc_epel_h)(int16_t *dst, - const uint8_t *_src, ptrdiff_t _srcstride, - int height, intptr_t mx, intptr_t my, int width) -{ - int x, y; - const pixel *src = (const pixel *)_src; - ptrdiff_t srcstride = _srcstride / sizeof(pixel); - const int8_t *filter = ff_hevc_epel_filters[mx - 1]; - for (y = 0; y < height; y++) { - for (x = 0; x < width; x++) - dst[x] = EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8); - src += srcstride; - dst += MAX_PB_SIZE; - } -} - -static void FUNC(put_hevc_epel_v)(int16_t *dst, - const uint8_t *_src, ptrdiff_t _srcstride, - int height, intptr_t mx, intptr_t my, int width) -{ - int x, y; - const pixel *src = (const pixel *)_src; - ptrdiff_t srcstride = _srcstride / sizeof(pixel); - const int8_t *filter = ff_hevc_epel_filters[my - 1]; - - for (y = 0; y < height; y++) { - for (x = 0; x < width; x++) - dst[x] = EPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8); - src += srcstride; - dst += MAX_PB_SIZE; - } -} - -static void FUNC(put_hevc_epel_hv)(int16_t *dst, - const uint8_t *_src, ptrdiff_t _srcstride, - int height, intptr_t mx, intptr_t my, int width) -{ - int x, y; - const pixel *src = (const pixel *)_src; - ptrdiff_t srcstride = _srcstride / sizeof(pixel); - const int8_t *filter = ff_hevc_epel_filters[mx - 1]; - int16_t tmp_array[(MAX_PB_SIZE + EPEL_EXTRA) * MAX_PB_SIZE]; - int16_t *tmp = tmp_array; - - src -= EPEL_EXTRA_BEFORE * srcstride; - - for (y = 0; y < height + EPEL_EXTRA; y++) { - for (x = 0; x < width; x++) - tmp[x] = EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8); - src += srcstride; - tmp += MAX_PB_SIZE; - } - - tmp = tmp_array + EPEL_EXTRA_BEFORE * MAX_PB_SIZE; - filter = ff_hevc_epel_filters[my - 1]; - - for (y = 0; y < height; y++) { - for (x = 0; x < width; x++) - dst[x] = EPEL_FILTER(tmp, MAX_PB_SIZE) >> 6; - tmp += MAX_PB_SIZE; - dst += MAX_PB_SIZE; - } -} - -static void FUNC(put_hevc_epel_uni_h)(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, - int height, intptr_t mx, intptr_t my, int width) -{ - int x, y; - const pixel *src = (const pixel *)_src; - ptrdiff_t srcstride = _srcstride / sizeof(pixel); - pixel *dst = (pixel *)_dst; - ptrdiff_t dststride = _dststride / sizeof(pixel); - const int8_t *filter = ff_hevc_epel_filters[mx - 1]; - int shift = 14 - BIT_DEPTH; -#if BIT_DEPTH < 14 - int offset = 1 << (shift - 1); -#else - int offset = 0; -#endif - - for (y = 0; y < height; y++) { - for (x = 0; x < width; x++) - dst[x] = av_clip_pixel(((EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) + offset) >> shift); - src += srcstride; - dst += dststride; - } -} - static void FUNC(put_hevc_epel_bi_h)(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, const int16_t *src2, int height, intptr_t mx, intptr_t my, int width) @@ -988,30 +639,6 @@ static void FUNC(put_hevc_epel_bi_h)(uint8_t *_dst, ptrdiff_t _dststride, } } -static void FUNC(put_hevc_epel_uni_v)(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, - int height, intptr_t mx, intptr_t my, int width) -{ - int x, y; - const pixel *src = (const pixel *)_src; - ptrdiff_t srcstride = _srcstride / sizeof(pixel); - pixel *dst = (pixel *)_dst; - ptrdiff_t dststride = _dststride / sizeof(pixel); - const int8_t *filter = ff_hevc_epel_filters[my - 1]; - int shift = 14 - BIT_DEPTH; -#if BIT_DEPTH < 14 - int offset = 1 << (shift - 1); -#else - int offset = 0; -#endif - - for (y = 0; y < height; y++) { - for (x = 0; x < width; x++) - dst[x] = av_clip_pixel(((EPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) + offset) >> shift); - src += srcstride; - dst += dststride; - } -} - static void FUNC(put_hevc_epel_bi_v)(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, const int16_t *src2, int height, intptr_t mx, intptr_t my, int width) @@ -1038,44 +665,6 @@ static void FUNC(put_hevc_epel_bi_v)(uint8_t *_dst, ptrdiff_t _dststride, } } -static void FUNC(put_hevc_epel_uni_hv)(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, - int height, intptr_t mx, intptr_t my, int width) -{ - int x, y; - const pixel *src = (const pixel *)_src; - ptrdiff_t srcstride = _srcstride / sizeof(pixel); - pixel *dst = (pixel *)_dst; - ptrdiff_t dststride = _dststride / sizeof(pixel); - const int8_t *filter = ff_hevc_epel_filters[mx - 1]; - int16_t tmp_array[(MAX_PB_SIZE + EPEL_EXTRA) * MAX_PB_SIZE]; - int16_t *tmp = tmp_array; - int shift = 14 - BIT_DEPTH; -#if BIT_DEPTH < 14 - int offset = 1 << (shift - 1); -#else - int offset = 0; -#endif - - src -= EPEL_EXTRA_BEFORE * srcstride; - - for (y = 0; y < height + EPEL_EXTRA; y++) { - for (x = 0; x < width; x++) - tmp[x] = EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8); - src += srcstride; - tmp += MAX_PB_SIZE; - } - - tmp = tmp_array + EPEL_EXTRA_BEFORE * MAX_PB_SIZE; - filter = ff_hevc_epel_filters[my - 1]; - - for (y = 0; y < height; y++) { - for (x = 0; x < width; x++) - dst[x] = av_clip_pixel(((EPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) + offset) >> shift); - tmp += MAX_PB_SIZE; - dst += dststride; - } -} - static void FUNC(put_hevc_epel_bi_hv)(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, const int16_t *src2, int height, intptr_t mx, intptr_t my, int width) @@ -1116,32 +705,6 @@ static void FUNC(put_hevc_epel_bi_hv)(uint8_t *_dst, ptrdiff_t _dststride, } } -static void FUNC(put_hevc_epel_uni_w_h)(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, - int height, int denom, int wx, int ox, intptr_t mx, intptr_t my, int width) -{ - int x, y; - const pixel *src = (const pixel *)_src; - ptrdiff_t srcstride = _srcstride / sizeof(pixel); - pixel *dst = (pixel *)_dst; - ptrdiff_t dststride = _dststride / sizeof(pixel); - const int8_t *filter = ff_hevc_epel_filters[mx - 1]; - int shift = denom + 14 - BIT_DEPTH; -#if BIT_DEPTH < 14 - int offset = 1 << (shift - 1); -#else - int offset = 0; -#endif - - ox = ox * (1 << (BIT_DEPTH - 8)); - for (y = 0; y < height; y++) { - for (x = 0; x < width; x++) { - dst[x] = av_clip_pixel((((EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) * wx + offset) >> shift) + ox); - } - dst += dststride; - src += srcstride; - } -} - static void FUNC(put_hevc_epel_bi_w_h)(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, const int16_t *src2, int height, int denom, int wx0, int wx1, @@ -1168,32 +731,6 @@ static void FUNC(put_hevc_epel_bi_w_h)(uint8_t *_dst, ptrdiff_t _dststride, } } -static void FUNC(put_hevc_epel_uni_w_v)(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, - int height, int denom, int wx, int ox, intptr_t mx, intptr_t my, int width) -{ - int x, y; - const pixel *src = (const pixel *)_src; - ptrdiff_t srcstride = _srcstride / sizeof(pixel); - pixel *dst = (pixel *)_dst; - ptrdiff_t dststride = _dststride / sizeof(pixel); - const int8_t *filter = ff_hevc_epel_filters[my - 1]; - int shift = denom + 14 - BIT_DEPTH; -#if BIT_DEPTH < 14 - int offset = 1 << (shift - 1); -#else - int offset = 0; -#endif - - ox = ox * (1 << (BIT_DEPTH - 8)); - for (y = 0; y < height; y++) { - for (x = 0; x < width; x++) { - dst[x] = av_clip_pixel((((EPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) * wx + offset) >> shift) + ox); - } - dst += dststride; - src += srcstride; - } -} - static void FUNC(put_hevc_epel_bi_w_v)(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, const int16_t *src2, int height, int denom, int wx0, int wx1, @@ -1220,45 +757,6 @@ static void FUNC(put_hevc_epel_bi_w_v)(uint8_t *_dst, ptrdiff_t _dststride, } } -static void FUNC(put_hevc_epel_uni_w_hv)(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, - int height, int denom, int wx, int ox, intptr_t mx, intptr_t my, int width) -{ - int x, y; - const pixel *src = (const pixel *)_src; - ptrdiff_t srcstride = _srcstride / sizeof(pixel); - pixel *dst = (pixel *)_dst; - ptrdiff_t dststride = _dststride / sizeof(pixel); - const int8_t *filter = ff_hevc_epel_filters[mx - 1]; - int16_t tmp_array[(MAX_PB_SIZE + EPEL_EXTRA) * MAX_PB_SIZE]; - int16_t *tmp = tmp_array; - int shift = denom + 14 - BIT_DEPTH; -#if BIT_DEPTH < 14 - int offset = 1 << (shift - 1); -#else - int offset = 0; -#endif - - src -= EPEL_EXTRA_BEFORE * srcstride; - - for (y = 0; y < height + EPEL_EXTRA; y++) { - for (x = 0; x < width; x++) - tmp[x] = EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8); - src += srcstride; - tmp += MAX_PB_SIZE; - } - - tmp = tmp_array + EPEL_EXTRA_BEFORE * MAX_PB_SIZE; - filter = ff_hevc_epel_filters[my - 1]; - - ox = ox * (1 << (BIT_DEPTH - 8)); - for (y = 0; y < height; y++) { - for (x = 0; x < width; x++) - dst[x] = av_clip_pixel((((EPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) * wx + offset) >> shift) + ox); - tmp += MAX_PB_SIZE; - dst += dststride; - } -} - static void FUNC(put_hevc_epel_bi_w_hv)(uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, const int16_t *src2, int height, int denom, int wx0, int wx1,