From patchwork Wed Mar 8 10:00:58 2017 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: =?utf-8?q?Martin_Storsj=C3=B6?= X-Patchwork-Id: 2825 Delivered-To: ffmpegpatchwork@gmail.com Received: by 10.103.50.79 with SMTP id y76csp955249vsy; Wed, 8 Mar 2017 02:08:46 -0800 (PST) X-Received: by 10.223.163.7 with SMTP id c7mr4435190wrb.17.1488967726294; Wed, 08 Mar 2017 02:08:46 -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 g10si3689710wrc.189.2017.03.08.02.08.45; Wed, 08 Mar 2017 02:08:46 -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=@martin-st.20150623.gappssmtp.com; spf=pass (google.com: domain of ffmpeg-devel-bounces@ffmpeg.org designates 79.124.17.100 as permitted sender) smtp.mailfrom=ffmpeg-devel-bounces@ffmpeg.org Received: from [127.0.1.1] (localhost [127.0.0.1]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id B0F866882D2; Wed, 8 Mar 2017 12:08:30 +0200 (EET) X-Original-To: ffmpeg-devel@ffmpeg.org Delivered-To: ffmpeg-devel@ffmpeg.org Received: from mail-lf0-f67.google.com (mail-lf0-f67.google.com [209.85.215.67]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTPS id 258E5688252 for ; Wed, 8 Mar 2017 12:08:24 +0200 (EET) Received: by mail-lf0-f67.google.com with SMTP id r36so2020245lfi.0 for ; Wed, 08 Mar 2017 02:08:37 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=martin-st.20150623.gappssmtp.com; s=20150623; h=from:to:subject:date:message-id:in-reply-to:references; bh=dTpYrIaBlJkWEw1Na9o341TZGlmr3cRVtIvaDb2fC6U=; b=GWXUOJcgUFFax8oEG5ZYJSMmJi7khihIkQNoOJnw/Ju4zAW/s/UKHMVYL8GDBWMV/f fw8212pMv92aB/4x0T8PZzMDkqSk7czjC8Y9ImZXppYPSZAUIWACTRucYBArnaNpyAvV CaXWnkeXdujDiVVqn1w2dSFml+UUr/i+E9eOndjM3ikRO/8hn/i/cHBTV1tbRQFHXg4u iD/55xz5QxZOZHEPqOGWRh4b2WFK9zOX91GUX5fWnw047AaKgslaSX4QkuvCh/A6fIAj nG75RxgQTZVIfzQQ8eFqIeUV7Gek+VdCqnl8lrXjDUEcDSwOORqwKpYU1NoPt2SCqkbY 55Ow== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:from:to:subject:date:message-id:in-reply-to :references; bh=dTpYrIaBlJkWEw1Na9o341TZGlmr3cRVtIvaDb2fC6U=; b=ADRg8CDxYLlnBzED20690qd1Y3kOuew0xbMhBoVtr2b3Ny0BrH8x1QaV07tekRmStt 2KATcvu4acrggKP318/byK8Ds/B9eMdQzb6X4ywyJ+6T+fXJLGw0BH1fGEizifzMVuEG EhP6Fi/YcQccnZgHRVkgBAB22fXDKNE8LXVd6mZZGnXbMUXToa9bOEukn/+BCniDc81A GgGXOm72cRPb4AuNRj+IYeBb+CjHE4e91VVc63Kiit6NtmFJiMnzeVAi9A8cEH9id07Q IqfcMUw/7Y4oIbYy/wAAYClPmcQLqlnKNM9TYj6zV90oD824GzdJ3iiZnubnuCGMOPyF CZ+w== X-Gm-Message-State: AMke39nHDxVKvm3kxdBBFO+53Kamzj9i+rvoV1lrQ5jlIEuK+5h/PpGNv75AH0kOXp0LJw== X-Received: by 10.46.8.65 with SMTP id g1mr1839418ljd.128.1488967290845; Wed, 08 Mar 2017 02:01:30 -0800 (PST) Received: from localhost.localdomain ([2001:470:28:852:7d47:68e:13e8:4933]) by smtp.gmail.com with ESMTPSA id m127sm513064lfg.58.2017.03.08.02.01.30 for (version=TLS1_2 cipher=ECDHE-RSA-AES128-SHA bits=128/128); Wed, 08 Mar 2017 02:01:30 -0800 (PST) From: =?UTF-8?q?Martin=20Storsj=C3=B6?= To: ffmpeg-devel@ffmpeg.org Date: Wed, 8 Mar 2017 12:00:58 +0200 Message-Id: <1488967274-8143-18-git-send-email-martin@martin.st> X-Mailer: git-send-email 2.7.4 In-Reply-To: <1488967274-8143-1-git-send-email-martin@martin.st> References: <1488967274-8143-1-git-send-email-martin@martin.st> Subject: [FFmpeg-devel] [PATCH 18/34] arm: vp9itxfm: Optimize 16x16 and 32x32 idct dc by unrolling X-BeenThere: ffmpeg-devel@ffmpeg.org X-Mailman-Version: 2.1.20 Precedence: list List-Id: FFmpeg development discussions and patches List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Reply-To: FFmpeg development discussions and patches MIME-Version: 1.0 Errors-To: ffmpeg-devel-bounces@ffmpeg.org Sender: "ffmpeg-devel" This work is sponsored by, and copyright, Google. Before: Cortex A7 A8 A9 A53 vp9_inv_dct_dct_16x16_sub1_add_neon: 273.0 189.5 211.7 235.8 vp9_inv_dct_dct_32x32_sub1_add_neon: 752.0 459.2 862.2 553.9 After: vp9_inv_dct_dct_16x16_sub1_add_neon: 226.5 145.0 225.1 171.8 vp9_inv_dct_dct_32x32_sub1_add_neon: 721.2 415.7 727.6 475.0 This is cherrypicked from libav commit a76bf8cf1277ef6feb1580b578f5e6ca327e713c. --- libavcodec/arm/vp9itxfm_neon.S | 54 ++++++++++++++++++++++++++++-------------- 1 file changed, 36 insertions(+), 18 deletions(-) diff --git a/libavcodec/arm/vp9itxfm_neon.S b/libavcodec/arm/vp9itxfm_neon.S index 78fdae6..dee2f05 100644 --- a/libavcodec/arm/vp9itxfm_neon.S +++ b/libavcodec/arm/vp9itxfm_neon.S @@ -542,16 +542,23 @@ function idct16x16_dc_add_neon vrshr.s16 q8, q8, #6 + mov r3, r0 mov r12, #16 1: @ Loop to add the constant from q8 into all 16x16 outputs - vld1.8 {q3}, [r0,:128] - vaddw.u8 q10, q8, d6 - vaddw.u8 q11, q8, d7 - vqmovun.s16 d6, q10 - vqmovun.s16 d7, q11 - vst1.8 {q3}, [r0,:128], r1 - subs r12, r12, #1 + subs r12, r12, #2 + vld1.8 {q2}, [r0,:128], r1 + vaddw.u8 q10, q8, d4 + vld1.8 {q3}, [r0,:128], r1 + vaddw.u8 q11, q8, d5 + vaddw.u8 q12, q8, d6 + vaddw.u8 q13, q8, d7 + vqmovun.s16 d4, q10 + vqmovun.s16 d5, q11 + vqmovun.s16 d6, q12 + vst1.8 {q2}, [r3,:128], r1 + vqmovun.s16 d7, q13 + vst1.8 {q3}, [r3,:128], r1 bne 1b bx lr @@ -1147,20 +1154,31 @@ function idct32x32_dc_add_neon vrshr.s16 q8, q8, #6 + mov r3, r0 mov r12, #32 1: @ Loop to add the constant from q8 into all 32x32 outputs - vld1.8 {q2-q3}, [r0,:128] - vaddw.u8 q10, q8, d4 - vaddw.u8 q11, q8, d5 - vaddw.u8 q12, q8, d6 - vaddw.u8 q13, q8, d7 - vqmovun.s16 d4, q10 - vqmovun.s16 d5, q11 - vqmovun.s16 d6, q12 - vqmovun.s16 d7, q13 - vst1.8 {q2-q3}, [r0,:128], r1 - subs r12, r12, #1 + subs r12, r12, #2 + vld1.8 {q0-q1}, [r0,:128], r1 + vaddw.u8 q9, q8, d0 + vaddw.u8 q10, q8, d1 + vld1.8 {q2-q3}, [r0,:128], r1 + vaddw.u8 q11, q8, d2 + vaddw.u8 q12, q8, d3 + vaddw.u8 q13, q8, d4 + vaddw.u8 q14, q8, d5 + vaddw.u8 q15, q8, d6 + vqmovun.s16 d0, q9 + vaddw.u8 q9, q8, d7 + vqmovun.s16 d1, q10 + vqmovun.s16 d2, q11 + vqmovun.s16 d3, q12 + vqmovun.s16 d4, q13 + vqmovun.s16 d5, q14 + vst1.8 {q0-q1}, [r3,:128], r1 + vqmovun.s16 d6, q15 + vqmovun.s16 d7, q9 + vst1.8 {q2-q3}, [r3,:128], r1 bne 1b bx lr