From patchwork Thu Jan 18 00:03:32 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Marton Balint X-Patchwork-Id: 7315 Delivered-To: ffmpegpatchwork@gmail.com Received: by 10.2.78.2 with SMTP id r2csp6224352jaa; Wed, 17 Jan 2018 16:04:05 -0800 (PST) X-Google-Smtp-Source: ACJfBoukb+PdHCmGcAX8b9CBmwR96dyymnXX+66yj4uvEcjyZIHPs7+c6d1001dJn6uVTP60OwCs X-Received: by 10.28.175.18 with SMTP id y18mr3346803wme.111.1516233845269; Wed, 17 Jan 2018 16:04:05 -0800 (PST) ARC-Seal: i=1; a=rsa-sha256; t=1516233845; cv=none; d=google.com; s=arc-20160816; b=UysAgmQSvwRjifk9xlE1Qf9o90CsoWIYKq4j20RlsE7qnQDMnZYYbFzknkYk1QX6hj YlGujdUo/Pvq/TGq2ur8XMW/uJ44YOnVe6ro7rQuC3AWU5leM9/hzfMWdwf0KMu1zk8p gckrh3tgdiy3SGXk1aj54vVkhAVuX2X63vFFHbYXzQPG/GuEF6+TFKuQd9aarp3c45/G ZRpJ06uEH+x6r1RL94bSJM2i1hjtoUyOMHsNRqxIQLyjX7dqA8e5nAWhUnqi9QmN6TQk 7ircufnxO0H2gB1t+UOLEZOMJaeAxzyd2sOhrOOPnegDHeLNnhq1FYdPWAZPB34Y2yrz rqAQ== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=sender:errors-to:content-transfer-encoding:mime-version:cc:reply-to :list-subscribe:list-help:list-post:list-archive:list-unsubscribe :list-id:precedence:subject:references:in-reply-to:message-id:date :to:from:delivered-to:arc-authentication-results; bh=PwbAw+cmaVn4SHeIPUpaK7geI88so4wRwmzAZiOxOjo=; b=PcY0gqyPke/SltMX87+rjbEarJwzzYNx4789m0c3+Est1E/J/FXNDevnuIJA2KFe2y YJz/886hJMhJHOESjuC/Ak1Brd6iWzEROtKYx068e4iL0z5Tvx91rhDu7n/N6Iygffar OH/OBmGeW2IIEh5FLqyH8LZc5CtRu9QtImeQtN5GaLl2kK73WY0gWL98wx/wsijqEyO5 SDmNpK02KNipxRZIIQ/jkzVxH/fg9VsF7M5W2h2FnmZL1Gi/0yV/2QSlmWwTiqjq/0Gf qCHWAuK8pIIOi4orTHIj11GOD1f1v4Rx8ZW5HXOKskhhW/YYm+H4MeOzTdH4EQ9JOp9v tcXw== ARC-Authentication-Results: i=1; mx.google.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 Return-Path: Received: from ffbox0-bg.mplayerhq.hu (ffbox0-bg.ffmpeg.org. [79.124.17.100]) by mx.google.com with ESMTP id x15si5149861wrg.146.2018.01.17.16.04.04; Wed, 17 Jan 2018 16:04:05 -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; 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 6EB8568A161; Thu, 18 Jan 2018 02:03:53 +0200 (EET) X-Original-To: ffmpeg-devel@ffmpeg.org Delivered-To: ffmpeg-devel@ffmpeg.org Received: from iq.passwd.hu (iq.passwd.hu [217.27.212.140]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id 967E268A12B for ; Thu, 18 Jan 2018 02:03:46 +0200 (EET) Received: from localhost (localhost [127.0.0.1]) by iq.passwd.hu (Postfix) with ESMTP id 8B7EAE1BD7; Thu, 18 Jan 2018 01:03:46 +0100 (CET) X-Virus-Scanned: amavisd-new at passwd.hu Received: from iq.passwd.hu ([127.0.0.1]) by localhost (iq.passwd.hu [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id LTMcswMVEvxI; Thu, 18 Jan 2018 01:03:45 +0100 (CET) Received: from bluegene.passwd.hu (localhost [127.0.0.1]) by iq.passwd.hu (Postfix) with ESMTP id E0450E1B66; Thu, 18 Jan 2018 01:03:44 +0100 (CET) From: Marton Balint To: ffmpeg-devel@ffmpeg.org Date: Thu, 18 Jan 2018 01:03:32 +0100 Message-Id: <20180118000332.31119-4-cus@passwd.hu> X-Mailer: git-send-email 2.13.6 In-Reply-To: <20180118000332.31119-1-cus@passwd.hu> References: <20180118000332.31119-1-cus@passwd.hu> Subject: [FFmpeg-devel] [PATCH 4/4] avfilter/vf_framerate: add SIMD functions for frame blending 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 Cc: Marton Balint MIME-Version: 1.0 Errors-To: ffmpeg-devel-bounces@ffmpeg.org Sender: "ffmpeg-devel" Blend function speedups on x86_64 Core i5 4460: ffmpeg -f lavfi -i allyuv -vf framerate=60:threads=1 -f null none C: 447548411 decicycles in Blend, 2048 runs, 0 skips SSSE3: 130020087 decicycles in Blend, 2048 runs, 0 skips AVX2: 128508221 decicycles in Blend, 2048 runs, 0 skips ffmpeg -f lavfi -i allyuv -vf format=yuv420p12,framerate=60:threads=1 -f null none C: 228932745 decicycles in Blend, 2048 runs, 0 skips SSE4: 123357781 decicycles in Blend, 2048 runs, 0 skips AVX2: 121215353 decicycles in Blend, 2048 runs, 0 skips Signed-off-by: Marton Balint --- libavfilter/vf_framerate.c | 24 ++++++- libavfilter/x86/Makefile | 1 + libavfilter/x86/vf_framerate.asm | 131 +++++++++++++++++++++++++++++++++++++++ 3 files changed, 153 insertions(+), 3 deletions(-) create mode 100644 libavfilter/x86/vf_framerate.asm diff --git a/libavfilter/vf_framerate.c b/libavfilter/vf_framerate.c index d315ef5d09..1c10699c3a 100644 --- a/libavfilter/vf_framerate.c +++ b/libavfilter/vf_framerate.c @@ -29,11 +29,13 @@ #define DEBUG #include "libavutil/avassert.h" +#include "libavutil/cpu.h" #include "libavutil/imgutils.h" #include "libavutil/internal.h" #include "libavutil/opt.h" #include "libavutil/pixdesc.h" #include "libavutil/pixelutils.h" +#include "libavutil/x86/cpu.h" #include "avfilter.h" #include "internal.h" @@ -246,7 +248,7 @@ static int blend_frames(AVFilterContext *ctx, int interpolate) av_frame_copy_props(s->work, s->f0); ff_dlog(ctx, "blend_frames() INTERPOLATE to create work frame\n"); - ctx->internal->execute(ctx, filter_slice, &td, NULL, FFMIN(outlink->h, ff_filter_get_nb_threads(ctx))); + ctx->internal->execute(ctx, filter_slice, &td, NULL, FFMIN(FFMAX(1, outlink->h >> 2), ff_filter_get_nb_threads(ctx))); return 1; } return 0; @@ -347,6 +349,11 @@ static void blend_frames_c(BLEND_FUNC_PARAMS) } } +void ff_blend_frames_ssse3(BLEND_FUNC_PARAMS); +void ff_blend_frames_avx2(BLEND_FUNC_PARAMS); +void ff_blend_frames16_sse4(BLEND_FUNC_PARAMS); +void ff_blend_frames16_avx2(BLEND_FUNC_PARAMS); + static void blend_frames16_c(BLEND_FUNC_PARAMS) { int line, pixel; @@ -371,6 +378,7 @@ static int config_input(AVFilterLink *inlink) AVFilterContext *ctx = inlink->dst; FrameRateContext *s = ctx->priv; const AVPixFmtDescriptor *pix_desc = av_pix_fmt_desc_get(inlink->format); + int cpu_flags = av_get_cpu_flags(); int plane; for (plane = 0; plane < 4; plane++) { @@ -389,10 +397,20 @@ static int config_input(AVFilterLink *inlink) if (s->bitdepth == 8) { s->blend_factor_max = 1 << BLEND_FACTOR_DEPTH8; - s->blend = blend_frames_c; + if (ARCH_X86 && EXTERNAL_AVX2(cpu_flags)) + s->blend = ff_blend_frames_avx2; + else if (ARCH_X86 && EXTERNAL_SSSE3(cpu_flags)) + s->blend = ff_blend_frames_ssse3; + else + s->blend = blend_frames_c; } else { s->blend_factor_max = 1 << BLEND_FACTOR_DEPTH16; - s->blend = blend_frames16_c; + if (ARCH_X86 && EXTERNAL_AVX2(cpu_flags)) + s->blend = ff_blend_frames16_avx2; + else if (ARCH_X86 && EXTERNAL_SSE4(cpu_flags)) + s->blend = ff_blend_frames16_sse4; + else + s->blend = blend_frames16_c; } return 0; diff --git a/libavfilter/x86/Makefile b/libavfilter/x86/Makefile index 2fc5c62644..07fc0f9486 100644 --- a/libavfilter/x86/Makefile +++ b/libavfilter/x86/Makefile @@ -31,6 +31,7 @@ X86ASM-OBJS-$(CONFIG_AFIR_FILTER) += x86/af_afir.o X86ASM-OBJS-$(CONFIG_BLEND_FILTER) += x86/vf_blend.o X86ASM-OBJS-$(CONFIG_BWDIF_FILTER) += x86/vf_bwdif.o X86ASM-OBJS-$(CONFIG_COLORSPACE_FILTER) += x86/colorspacedsp.o +X86ASM-OBJS-$(CONFIG_FRAMERATE_FILTER) += x86/vf_framerate.o X86ASM-OBJS-$(CONFIG_FSPP_FILTER) += x86/vf_fspp.o X86ASM-OBJS-$(CONFIG_GRADFUN_FILTER) += x86/vf_gradfun.o X86ASM-OBJS-$(CONFIG_HFLIP_FILTER) += x86/vf_hflip.o diff --git a/libavfilter/x86/vf_framerate.asm b/libavfilter/x86/vf_framerate.asm new file mode 100644 index 0000000000..0b1bed821f --- /dev/null +++ b/libavfilter/x86/vf_framerate.asm @@ -0,0 +1,131 @@ +;***************************************************************************** +;* x86-optimized functions for framerate filter +;* +;* Copyright (C) 2018 Marton Balint +;* +;* Based on vf_blend.asm, Copyright (C) 2015 Paul B Mahol +;* +;* This file is part of FFmpeg. +;* +;* FFmpeg is free software; you can redistribute it and/or +;* modify it under the terms of the GNU Lesser General Public +;* License as published by the Free Software Foundation; either +;* version 2.1 of the License, or (at your option) any later version. +;* +;* FFmpeg is distributed in the hope that it will be useful, +;* but WITHOUT ANY WARRANTY; without even the implied warranty of +;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +;* Lesser General Public License for more details. +;* +;* You should have received a copy of the GNU Lesser General Public +;* License along with FFmpeg; if not, write to the Free Software +;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA +;****************************************************************************** + +%include "libavutil/x86/x86util.asm" + +SECTION .text + + +%macro XSPLAT 3 +%if cpuflag(avx2) + vpbroadcast%3 %1, %2 +%else + movd %1, %2 +%ifidn %3, d + SPLATD %1 +%else + SPLATW %1, %1 +%endif +%endif +%endmacro + + +%macro BLEND_INIT 0-1 +%if ARCH_X86_64 +cglobal blend_frames%1, 6, 9, 5, src1, src1_linesize, src2, src2_linesize, dst, dst_linesize, width, end, x + mov widthd, dword widthm +%else +cglobal blend_frames%1, 5, 7, 5, src1, src1_linesize, src2, src2_linesize, dst, end, x +%define dst_linesizeq r5mp +%define widthq r6mp +%endif + mov endd, dword r7m + add src1q, widthq + add src2q, widthq + add dstq, widthq + neg widthq +%endmacro + + +%macro BLEND_LOOP 4 +.nextrow: + mov xq, widthq + + .loop: + movu m0, [src1q + xq] + movu m1, [src2q + xq] + SBUTTERFLY %1%2, 0, 1, 4 ; aAbBcCdD + ; eEfFgGhH + pmadd%3 m0, m2 + pmadd%3 m1, m2 + + padd%2 m0, m3 + padd%2 m1, m3 + psrl%2 m0, %4 ; 0A0B0C0D + psrl%2 m1, %4 ; 0E0F0G0H + + packus%2%1 m0, m1 ; ABCDEFGH + movu [dstq + xq], m0 + add xq, mmsize + jl .loop + add src1q, src1_linesizeq + add src2q, src2_linesizeq + add dstq, dst_linesizeq + sub endd, 1 + jg .nextrow +REP_RET +%endmacro + + +%macro BLEND_FRAMES 0 + BLEND_INIT + + XSPLAT m2, r8m, w ; factor1 + XSPLAT m3, r9m, w ; factor2 + + psllw m3, 8 + por m2, m3 ; interleaved factors + + XSPLAT m3, r10m, w ; half + + BLEND_LOOP b, w, ubsw, 7 +%endmacro + + +%macro BLEND_FRAMES16 0 + BLEND_INIT 16 + + XSPLAT m2, r8m, d ; factor1 + XSPLAT m3, r9m, d ; factor2 + + pslld m3, 16 + por m2, m3 ; interleaved factors + + XSPLAT m3, r10m, d ; half + + BLEND_LOOP w, d, wd, 15 +%endmacro + + +INIT_XMM ssse3 +BLEND_FRAMES + +INIT_YMM avx2 +BLEND_FRAMES + +INIT_XMM sse4 +BLEND_FRAMES16 + +INIT_YMM avx2 +BLEND_FRAMES16