From patchwork Wed Mar 8 10:00:59 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: 2828 Delivered-To: ffmpegpatchwork@gmail.com Received: by 10.103.50.79 with SMTP id y76csp955400vsy; Wed, 8 Mar 2017 02:09:14 -0800 (PST) X-Received: by 10.28.26.196 with SMTP id a187mr21907934wma.33.1488967754043; Wed, 08 Mar 2017 02:09:14 -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 e5si3695759wrd.173.2017.03.08.02.09.13; Wed, 08 Mar 2017 02:09:14 -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 9A0FF688310; Wed, 8 Mar 2017 12:08:57 +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 D6DBB688252 for ; Wed, 8 Mar 2017 12:08:56 +0200 (EET) Received: by mail-lf0-f67.google.com with SMTP id r36so2021097lfi.0 for ; Wed, 08 Mar 2017 02:09:10 -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=7raFOipaH0ylX1biN2Js4xjR21bgg6N08fVJdGaU4DY=; b=x3SG3AkRezQiPjaSkm3JPlyMa0Yx/7YpIL4k2kxjHpvU4j3jqyEPcuTle5BQeO/LtK t8lhvkEDmzsr6Z1QTXkcmlJpNZYuxikD/4N5SgyxCN6e5Q6H6KNaV7hDEWINSECw/Sam Bs31pYKeYJQ1iZ4qQTRiSFs7ej71VdEFxYWwN9C/Zs5iiG/VX3bpUhA+L8Pu0HsEa4SK aSbheo7BgakNWpucK3MyUNKm5IbjUSvTZTNqi2mfYO0tytFVn2RL0JXaPV7lrKpEA2p/ D/ZJmwx7+DfL1Z7QV7ixZ55WUC7gVNHYSEsjzfNxiaBkGVm+ChhDIEH7YOi2nGnZoJ+M r61w== 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=7raFOipaH0ylX1biN2Js4xjR21bgg6N08fVJdGaU4DY=; b=iXBSlpm8QYKwlbDMTGRGVkYVyh/x/ZkRUVmNkkNcU1rExjBIfdp8jO2h0ZP1RrSY+L y33amrw0BAVpty+2pBOrrL0OfzMVKk2AUJI5Lra5xfjgntekpJflqVqKV9paJxBoH7E4 Nzp1fBQIHWqi8b4NI0sItRh2MLNc8hxw6naxjACagNBFKDURU7vpmrXa3NoZBKqoZRCs TNOHIG21ZqQld8qxNFUA8F+NBum0xVfMNaDixgwxJKtlv5NHwffHTpOR0LuN5sFjJCUs GHVlhcMjVy7xr+MP8Z0sLDwMx0OiJM361P8/EMgp2GPG6hOKr9JZshYC0vSNMHK5fHxP A/rw== X-Gm-Message-State: AMke39nVRWhC8Sbad5l/F9+jdp7qvw6qop5JUNRr/am9lqkcH4Q+Pl202kP/BxZfOT9bvg== X-Received: by 10.46.7.80 with SMTP id i16mr1770004ljd.129.1488967291635; Wed, 08 Mar 2017 02:01:31 -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 02:01:31 -0800 (PST) From: =?UTF-8?q?Martin=20Storsj=C3=B6?= To: ffmpeg-devel@ffmpeg.org Date: Wed, 8 Mar 2017 12:00:59 +0200 Message-Id: <1488967274-8143-19-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 19/34] aarch64: 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 A53 vp9_inv_dct_dct_16x16_sub1_add_neon: 235.3 vp9_inv_dct_dct_32x32_sub1_add_neon: 555.1 After: vp9_inv_dct_dct_16x16_sub1_add_neon: 180.2 vp9_inv_dct_dct_32x32_sub1_add_neon: 475.3 This is cherrypicked from libav commit 3fcf788fbbccc4130868e7abe58a88990290f7c1. --- libavcodec/aarch64/vp9itxfm_neon.S | 54 +++++++++++++++++++++++++------------- 1 file changed, 36 insertions(+), 18 deletions(-) diff --git a/libavcodec/aarch64/vp9itxfm_neon.S b/libavcodec/aarch64/vp9itxfm_neon.S index 6bb097b..be65eb7 100644 --- a/libavcodec/aarch64/vp9itxfm_neon.S +++ b/libavcodec/aarch64/vp9itxfm_neon.S @@ -495,16 +495,23 @@ function idct16x16_dc_add_neon srshr v2.8h, v2.8h, #6 + mov x3, x0 mov x4, #16 1: // Loop to add the constant from v2 into all 16x16 outputs - ld1 {v3.16b}, [x0] - uaddw v4.8h, v2.8h, v3.8b - uaddw2 v5.8h, v2.8h, v3.16b - sqxtun v4.8b, v4.8h - sqxtun2 v4.16b, v5.8h - st1 {v4.16b}, [x0], x1 - subs x4, x4, #1 + subs x4, x4, #2 + ld1 {v3.16b}, [x0], x1 + ld1 {v4.16b}, [x0], x1 + uaddw v16.8h, v2.8h, v3.8b + uaddw2 v17.8h, v2.8h, v3.16b + uaddw v18.8h, v2.8h, v4.8b + uaddw2 v19.8h, v2.8h, v4.16b + sqxtun v3.8b, v16.8h + sqxtun2 v3.16b, v17.8h + sqxtun v4.8b, v18.8h + sqxtun2 v4.16b, v19.8h + st1 {v3.16b}, [x3], x1 + st1 {v4.16b}, [x3], x1 b.ne 1b ret @@ -1054,20 +1061,31 @@ function idct32x32_dc_add_neon srshr v0.8h, v2.8h, #6 + mov x3, x0 mov x4, #32 1: // Loop to add the constant v0 into all 32x32 outputs - ld1 {v1.16b,v2.16b}, [x0] - uaddw v3.8h, v0.8h, v1.8b - uaddw2 v4.8h, v0.8h, v1.16b - uaddw v5.8h, v0.8h, v2.8b - uaddw2 v6.8h, v0.8h, v2.16b - sqxtun v3.8b, v3.8h - sqxtun2 v3.16b, v4.8h - sqxtun v4.8b, v5.8h - sqxtun2 v4.16b, v6.8h - st1 {v3.16b,v4.16b}, [x0], x1 - subs x4, x4, #1 + subs x4, x4, #2 + ld1 {v1.16b,v2.16b}, [x0], x1 + uaddw v16.8h, v0.8h, v1.8b + uaddw2 v17.8h, v0.8h, v1.16b + ld1 {v3.16b,v4.16b}, [x0], x1 + uaddw v18.8h, v0.8h, v2.8b + uaddw2 v19.8h, v0.8h, v2.16b + uaddw v20.8h, v0.8h, v3.8b + uaddw2 v21.8h, v0.8h, v3.16b + uaddw v22.8h, v0.8h, v4.8b + uaddw2 v23.8h, v0.8h, v4.16b + sqxtun v1.8b, v16.8h + sqxtun2 v1.16b, v17.8h + sqxtun v2.8b, v18.8h + sqxtun2 v2.16b, v19.8h + sqxtun v3.8b, v20.8h + sqxtun2 v3.16b, v21.8h + st1 {v1.16b,v2.16b}, [x3], x1 + sqxtun v4.8b, v22.8h + sqxtun2 v4.16b, v23.8h + st1 {v3.16b,v4.16b}, [x3], x1 b.ne 1b ret