From patchwork Sun Jul 2 12:32:38 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: John Cox X-Patchwork-Id: 42367 Delivered-To: ffmpegpatchwork2@gmail.com Received: by 2002:a05:6a20:3b1e:b0:12b:9ae3:586d with SMTP id c30csp3715182pzh; Sun, 2 Jul 2023 05:34:50 -0700 (PDT) X-Google-Smtp-Source: APBJJlGdzqke94zS30KjsAPc9CT59H9JLgO/TE2CphNQB6Wm3aGNxWoq4fywSdQM5J+e+kiuBVOz X-Received: by 2002:a17:906:3515:b0:992:aab0:5340 with SMTP id r21-20020a170906351500b00992aab05340mr5509317eja.46.1688301290264; Sun, 02 Jul 2023 05:34:50 -0700 (PDT) ARC-Seal: i=1; a=rsa-sha256; t=1688301290; cv=none; d=google.com; s=arc-20160816; b=0LN5Z56U1nsf5z+Vov8O+vVND20ifYkmhZoqxIS0VctOr2CdGWj7GvrTbM48Aj9n71 KL0QJrtbgbp1zKfeNf6ezqZMXJi6/IWNLWzDmtzPhso+idghTpClCyjN37ME6IH+N2G7 qjg/jgr9J/Ay97hw7h9qhEJ12klw6MA7DwU7eVmQomhZzaXG5ovxgnD+436HDxtVDxwZ vdWzho8dW1Bv9qq+FZCKnrkPvqknqhPvIqTGKP/nx0jAVd2BQ9xjyu7asseaaSbjh6nw 9GqP7FuJvju8PLwKJHj8P0tnUczhxxs4pgs2OUrB7AR2RGTUqw8eLbJqFNzbQDA2lPYq AQeQ== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=sender:errors-to:content-transfer-encoding:cc:reply-to :list-subscribe:list-help:list-post:list-archive:list-unsubscribe :list-id:precedence:subject:mime-version:references:in-reply-to :message-id:date:to:from:dkim-signature:delivered-to; bh=KLisMAfcvnRNwB0RlKj/j7GLawB9jCvPMdeKOBLt/Mk=; fh=2QQVLAqz5Dgp0O7PTQ7hb1i3rOEvtuxkp5BnHStC38U=; b=T04r+75eBvVJbVyUPWmV/og3nJ1zwP5A2IATP4plW4/KNKXMlDn+T+lFZ6iSxYBZZQ quSSWZiZcNKCUb9fGfcBQ2DihFcE/U6iB0UJyzDfmPuHH07+FSd91Vtic6xG6FA+Awiu cilO55QB9BH9IiPRBfg9y5kpb5iQunBKWbiaDrO5M1G97zc0Cg6KpO42T88LRu16EEkA 0eXmyqTYqWdJhIyolK+DH5jlhuSnRWaeRN9wrFSMnNWeXseotRVc9DUrRC/UMIrMq9cM sT8QCo16aeP1E3SZoYq3xBY2zNRdTSXXGsSddR2AC9NACq6PyweIY6eX+KnA26SWpWwh 7DIw== ARC-Authentication-Results: i=1; mx.google.com; dkim=neutral (body hash did not verify) header.i=@kynesim-co-uk.20221208.gappssmtp.com header.s=20221208 header.b=Wv96Ew6w; spf=pass (google.com: domain of ffmpeg-devel-bounces@ffmpeg.org designates 79.124.17.100 as permitted sender) smtp.mailfrom=ffmpeg-devel-bounces@ffmpeg.org Return-Path: Received: from ffbox0-bg.mplayerhq.hu (ffbox0-bg.ffmpeg.org. [79.124.17.100]) by mx.google.com with ESMTP id c15-20020a170906528f00b00988c64dd2e7si10501099ejm.850.2023.07.02.05.34.49; Sun, 02 Jul 2023 05:34:50 -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=@kynesim-co-uk.20221208.gappssmtp.com header.s=20221208 header.b=Wv96Ew6w; spf=pass (google.com: domain of ffmpeg-devel-bounces@ffmpeg.org designates 79.124.17.100 as permitted sender) smtp.mailfrom=ffmpeg-devel-bounces@ffmpeg.org Received: from [127.0.1.1] (localhost [127.0.0.1]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id 19B1E68C39A; Sun, 2 Jul 2023 15:33:24 +0300 (EEST) X-Original-To: ffmpeg-devel@ffmpeg.org Delivered-To: ffmpeg-devel@ffmpeg.org Received: from mail-wm1-f49.google.com (mail-wm1-f49.google.com [209.85.128.49]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTPS id C48AF68C33B for ; Sun, 2 Jul 2023 15:33:19 +0300 (EEST) Received: by mail-wm1-f49.google.com with SMTP id 5b1f17b1804b1-3fbc244d384so30535435e9.0 for ; Sun, 02 Jul 2023 05:33:19 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=kynesim-co-uk.20221208.gappssmtp.com; s=20221208; t=1688301199; x=1690893199; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:date:subject:cc:to:from:from:to:cc:subject:date :message-id:reply-to; bh=fXeTvjIX+HG/6iUBcrLfIlM9aJ2d/SMZztMv3qbJvnc=; b=Wv96Ew6wsTqmsmVtw1VBOcRN6L17AiReBdPDcbiGxB3nBitrjr5a0EQ/xSX/Poneb/ KG8Kv+6fzQT417jj8hpSI5Iz9RhDy2Jm5nULYK+Y803CUhRJx0yvBfWUyudqEDV5cW8+ oGSYcC9QvMGDQOG6Pvvr5p52FoqbH3NNoaVIn0SXW3JWVCBzrzcTRZoSUgjETSI+k/Ye zeHhhdIhhyqtTcBaVKOREq+OI5ZQxOV13sFBYTJ6gZYAepzOpWkxNKExDD2+bHEkiEP+ VDlNtj0sO10rwqkCEXy84IrpXUMYX4poltpcuQncvC05n9o6Xde3TWOenlyp/QmZH7BX o6zg== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20221208; t=1688301199; x=1690893199; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:date:subject:cc:to:from:x-gm-message-state:from:to:cc :subject:date:message-id:reply-to; bh=fXeTvjIX+HG/6iUBcrLfIlM9aJ2d/SMZztMv3qbJvnc=; b=SOyl5Ub8UxMPjHREJ3yJY3z5qb3UdCeUuDwUKdbuc1rvvpRmaHw+UjACg9XX69PKuJ S+eUaaS0bBn1zjpFRWaDMjKHhWo8HEL0sAn7gRc63sddw624awuYaDLFpXQrfnP/RaTo v/uXUyO+LqAlRePPoX7w16jnM8DEbTSrzPiVJjbhYaPuM8jPEJ44zQQEw7vD0xi4giU+ G7BAGX0aRwcyAwcI03dg93volFbxVUgvOqf8av61LjDf1viCCfqGzY5GYT36m1ZICrZU t2W6woqZbhtJIqzr7H+jKBQTvh7Fpj/+7SOn7HatxwJhvtLOSmD1CMwvbpAfA4zwxuDT j5Jw== X-Gm-Message-State: AC+VfDwn0Pl1J+X0z7CVQgE0JeuZb78Sf0Sf1O/IKDFwViD7zYno81ML Fcm/7iwdoIAXCMK+EPFbJeCyYdH0ytzkbzvyF1E= X-Received: by 2002:a7b:c012:0:b0:3fa:e92e:7a8b with SMTP id c18-20020a7bc012000000b003fae92e7a8bmr5774414wmb.13.1688301198708; Sun, 02 Jul 2023 05:33:18 -0700 (PDT) Received: from sucnaath.outer.uphall.net (cpc1-cmbg20-2-0-cust759.5-4.cable.virginm.net. [86.21.218.248]) by smtp.gmail.com with ESMTPSA id f12-20020a7bc8cc000000b003fbbe41fd78sm8816167wml.10.2023.07.02.05.33.17 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Sun, 02 Jul 2023 05:33:18 -0700 (PDT) From: John Cox To: ffmpeg-devel@ffmpeg.org Date: Sun, 2 Jul 2023 12:32:38 +0000 Message-Id: <20230702123242.232484-12-jc@kynesim.co.uk> X-Mailer: git-send-email 2.39.2 In-Reply-To: <20230702123242.232484-1-jc@kynesim.co.uk> References: <20230702123242.232484-1-jc@kynesim.co.uk> MIME-Version: 1.0 Subject: [FFmpeg-devel] [PATCH v2 11/15] avfilter/vf_bwdif: Add neon for filter_line 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: thomas.mundt@hr.de, John Cox , martin@martin.st Errors-To: ffmpeg-devel-bounces@ffmpeg.org Sender: "ffmpeg-devel" X-TUID: kD3jb7PR8uW4 Signed-off-by: John Cox --- libavfilter/aarch64/vf_bwdif_init_aarch64.c | 21 ++ libavfilter/aarch64/vf_bwdif_neon.S | 208 ++++++++++++++++++++ 2 files changed, 229 insertions(+) diff --git a/libavfilter/aarch64/vf_bwdif_init_aarch64.c b/libavfilter/aarch64/vf_bwdif_init_aarch64.c index e75cf2f204..21e67884ab 100644 --- a/libavfilter/aarch64/vf_bwdif_init_aarch64.c +++ b/libavfilter/aarch64/vf_bwdif_init_aarch64.c @@ -31,6 +31,26 @@ void ff_bwdif_filter_edge_neon(void *dst1, void *prev1, void *cur1, void *next1, void ff_bwdif_filter_intra_neon(void *dst1, void *cur1, int w, int prefs, int mrefs, int prefs3, int mrefs3, int parity, int clip_max); +void ff_bwdif_filter_line_neon(void *dst1, void *prev1, void *cur1, void *next1, + int w, int prefs, int mrefs, int prefs2, int mrefs2, + int prefs3, int mrefs3, int prefs4, int mrefs4, + int parity, int clip_max); + + +static void filter_line_helper(void *dst1, void *prev1, void *cur1, void *next1, + int w, int prefs, int mrefs, int prefs2, int mrefs2, + int prefs3, int mrefs3, int prefs4, int mrefs4, + int parity, int clip_max) +{ + const int w0 = clip_max != 255 ? 0 : w & ~15; + + ff_bwdif_filter_line_neon(dst1, prev1, cur1, next1, + w0, prefs, mrefs, prefs2, mrefs2, prefs3, mrefs3, prefs4, mrefs4, parity, clip_max); + + if (w0 < w) + ff_bwdif_filter_line_c((char *)dst1 + w0, (char *)prev1 + w0, (char *)cur1 + w0, (char *)next1 + w0, + w - w0, prefs, mrefs, prefs2, mrefs2, prefs3, mrefs3, prefs4, mrefs4, parity, clip_max); +} static void filter_edge_helper(void *dst1, void *prev1, void *cur1, void *next1, int w, int prefs, int mrefs, int prefs2, int mrefs2, @@ -71,6 +91,7 @@ ff_bwdif_init_aarch64(BWDIFContext *s, int bit_depth) return; s->filter_intra = filter_intra_helper; + s->filter_line = filter_line_helper; s->filter_edge = filter_edge_helper; } diff --git a/libavfilter/aarch64/vf_bwdif_neon.S b/libavfilter/aarch64/vf_bwdif_neon.S index d6e7d109f5..abc050565c 100644 --- a/libavfilter/aarch64/vf_bwdif_neon.S +++ b/libavfilter/aarch64/vf_bwdif_neon.S @@ -149,6 +149,214 @@ coeffs: .hword 5570, 3801, 1016, -3801 // hf[0] = v0.h[2], -hf[1] = v0.h[5] .hword 5077, 981 // sp[0] = v0.h[6] +// =========================================================================== +// +// void filter_line( +// void *dst1, // x0 +// void *prev1, // x1 +// void *cur1, // x2 +// void *next1, // x3 +// int w, // w4 +// int prefs, // w5 +// int mrefs, // w6 +// int prefs2, // w7 +// int mrefs2, // [sp, #0] +// int prefs3, // [sp, #SP_INT] +// int mrefs3, // [sp, #SP_INT*2] +// int prefs4, // [sp, #SP_INT*3] +// int mrefs4, // [sp, #SP_INT*4] +// int parity, // [sp, #SP_INT*5] +// int clip_max) // [sp, #SP_INT*6] + +function ff_bwdif_filter_line_neon, export=1 + // Sanity check w + cmp w4, #0 + ble 99f + + // Rearrange regs to be the same as line3 for ease of debug! + mov w10, w4 // w10 = loop count + mov w9, w6 // w9 = mref + mov w12, w7 // w12 = pref2 + mov w11, w5 // w11 = pref + ldr w8, [sp, #0] // w8 = mref2 + ldr w7, [sp, #SP_INT*2] // w7 = mref3 + ldr w6, [sp, #SP_INT*4] // w6 = mref4 + ldr w13, [sp, #SP_INT] // w13 = pref3 + ldr w14, [sp, #SP_INT*3] // w14 = pref4 + + mov x4, x3 + mov x3, x2 + mov x2, x1 + +// #define prev2 cur +// const uint8_t * restrict next2 = parity ? prev : next; + ldr w17, [sp, #SP_INT*5] // parity + cmp w17, #0 + csel x17, x2, x4, ne + + PUSH_VREGS + + ldr q0, coeffs + +// for (x = 0; x < w; x++) { +// int diff0, diff2; +// int d0, d2; +// int temporal_diff0, temporal_diff2; +// +// int i1, i2; +// int j1, j2; +// int p6, p5, p4, p3, p2, p1, c0, m1, m2, m3, m4; + +10: +// c0 = prev2[0] + next2[0]; // c0 = v20, v21 +// d0 = c0 >> 1; // d0 = v10 +// temporal_diff0 = FFABS(prev2[0] - next2[0]); // td0 = v11 + ldr q31, [x3] + ldr q21, [x17] + uhadd v10.16b, v31.16b, v21.16b + uabd v11.16b, v31.16b, v21.16b + uaddl v20.8h, v21.8b, v31.8b + uaddl2 v21.8h, v21.16b, v31.16b + + ldr q31, [x3, w6, sxtw] + ldr q23, [x17, w6, sxtw] + +// i1 = coef_hf[0] * c0; // i1 = v2-v5 + UMULL4K v2, v3, v4, v5, v20, v21, v0.h[2] + + ldr q30, [x3, w14, sxtw] + ldr q25, [x17, w14, sxtw] + +// m4 = prev2[mrefs4] + next2[mrefs4]; // m4 = v22,v23 + uaddl v22.8h, v23.8b, v31.8b + uaddl2 v23.8h, v23.16b, v31.16b + +// p4 = prev2[prefs4] + next2[prefs4]; // p4 = v24,v25, (p4 >> 1) = v12 + uhadd v12.16b, v25.16b, v30.16b + uaddl v24.8h, v25.8b, v30.8b + uaddl2 v25.8h, v25.16b, v30.16b + +// m3 = cur[mrefs3]; // m3 = v20 + ldr q20, [x3, w7, sxtw] + +// p3 = cur[prefs3]; // p3 = v21 + ldr q21, [x3, w13, sxtw] + +// i1 += coef_hf[2] * (m4 + p4); // (-m4:v22,v23) (-p4:v24,v25) + add v22.8h, v22.8h, v24.8h + add v23.8h, v23.8h, v25.8h + UMLAL4K v2, v3, v4, v5, v22, v23, v0.h[4] + + ldr q29, [x3, w8, sxtw] + ldr q23, [x17, w8, sxtw] + +// i1 -= coef_lf[1] * 4 * (m3 + p3); // - + uaddl v30.8h, v20.8b, v21.8b + uaddl2 v31.8h, v20.16b, v21.16b + + UMLSL4K v2, v3, v4, v5, v30, v31, v0.h[1] + +// m2 = prev2[mrefs2] + next2[mrefs2]; // m2 = v22,v23, (m2 >> 1) = v13 + uhadd v13.16b, v23.16b, v29.16b + uaddl v22.8h, v23.8b, v29.8b + uaddl2 v23.8h, v23.16b, v29.16b + + ldr q31, [x3, w12, sxtw] + ldr q27, [x17, w12, sxtw] + +// j1 += coef_hf[2] * (m2 + p6); // (-p6:v24,v25) + add v24.8h, v24.8h, v22.8h + add v25.8h, v25.8h, v23.8h + UMLAL4K v6, v7, v8, v9, v24, v25, v0.h[4] + +// m1 = cur[mrefs]; // m1 = v24 + ldr q24, [x3, w9, sxtw] + +// p2 = prev2[prefs2] + next2[prefs2]; // p2 = v26, v27 +// temporal_diff2 = FFABS(prev2[prefs2] - next2[prefs2]); // td2 = v14 +// d2 = p2 >> 1; // d2 = v15 + uabd v14.16b, v31.16b, v27.16b + uhadd v15.16b, v31.16b, v27.16b + uaddl v26.8h, v27.8b, v31.8b + uaddl2 v27.8h, v27.16b, v31.16b + +// i1 -= coef_hf[1] * (m2 + p2); // (-m2:v22,v23*) (-p2:v26*,v27*) + add v22.8h, v22.8h, v26.8h + add v23.8h, v23.8h, v27.8h + UMLSL4K v2, v3, v4, v5, v22, v23, v0.h[3] + +// p1 = cur[prefs]; // p1 = v22 + ldr q22, [x3, w11, sxtw] + +// i2 = (coef_sp[0] * (m1 + p1) - coef_sp[1] * (m3 + p3)) >> 13; // (-m3:v20*) i2=v17 + uaddl v18.8h, v22.8b, v24.8b + uaddl2 v19.8h, v22.16b, v24.16b + UMULL4K v28, v29, v30, v31, v18, v19, v0.h[6] + + uaddl v18.8h, v20.8b, v21.8b + uaddl2 v19.8h, v20.16b, v21.16b + UMLSL4K v28, v29, v30, v31, v18, v19, v0.h[7] + + SQSHRUNN v17, v28, v29, v30, v31, 13 + +// i1 += coef_lf[0] * 4 * (m1 + p1); // p1 = v22, m1 = v24 + uaddl v26.8h, v24.8b, v22.8b + uaddl2 v27.8h, v24.16b, v22.16b + UMLAL4K v2, v3, v4, v5, v26, v27, v0.h[0] + + ldr q31, [x2, w9, sxtw] + ldr q29, [x4, w9, sxtw] + + ldr q30, [x2, w11, sxtw] + ldr q28, [x4, w11, sxtw] + +// i1 >>= 15; // i1 = v2, -v3, -v4*, -v5* + SQSHRUNN v2, v2, v3, v4, v5, 15 + +// { +// int t1 =(FFABS(prev[mrefs] - m1) + FFABS(prev[prefs] - p1)) >> 1; +// int t2 =(FFABS(next[mrefs] - m1) + FFABS(next[prefs] - p1)) >> 1; + uabd v30.16b, v22.16b, v30.16b + uabd v31.16b, v24.16b, v31.16b + uabd v28.16b, v22.16b, v28.16b + uabd v29.16b, v24.16b, v29.16b + uhadd v31.16b, v31.16b, v30.16b + uhadd v29.16b, v29.16b, v28.16b + +// diff0 = FFMAX3(temporal_diff0 >> 1, t1, t2); // diff0=v18 + ushr v18.16b, v11.16b, #1 + umax v18.16b, v18.16b, v31.16b + umax v18.16b, v18.16b, v29.16b + + // diff0 = v18, (m2 >> 1) = v13, m1 = v24, d0 = v10, p1 = v22, d2 = v15 + SPAT_CHECK v18, v13, v24, v10, v22, v15, v31, v30, v29, v28 + + // i1 = v2, i2 = v17, m1 = v24, d0 = v10, p1 = v22, td2 = v11, diff2 = v18 + INTERPOL v2, v2, v17, v24, v10, v22, v11, v18, v31, v30, v29 + +// dst[0] = av_clip_uint8(interpol); + str q2, [x0], #16 +// } +// +// dst++; +// cur++; +// prev++; +// prev2++; +// next++; +// } + + subs w10, w10, #16 + add x2, x2, #16 + add x3, x3, #16 + add x4, x4, #16 + add x17, x17, #16 + bgt 10b + + POP_VREGS +99: + ret +endfunc + // ============================================================================ // // void ff_bwdif_filter_edge_neon(