From patchwork Sun Mar 24 13:10:35 2019 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Lauri Kasanen X-Patchwork-Id: 12415 Return-Path: X-Original-To: patchwork@ffaux-bg.ffmpeg.org Delivered-To: patchwork@ffaux-bg.ffmpeg.org Received: from ffbox0-bg.mplayerhq.hu (ffbox0-bg.ffmpeg.org [79.124.17.100]) by ffaux.localdomain (Postfix) with ESMTP id 5A9D6449559 for ; Sun, 24 Mar 2019 15:07:55 +0200 (EET) Received: from [127.0.1.1] (localhost [127.0.0.1]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id 3222D680B69; Sun, 24 Mar 2019 15:07:55 +0200 (EET) X-Original-To: ffmpeg-devel@ffmpeg.org Delivered-To: ffmpeg-devel@ffmpeg.org Received: from mout.gmx.net (mout.gmx.net [212.227.17.21]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTPS id 5AD256805D0 for ; Sun, 24 Mar 2019 15:07:49 +0200 (EET) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=gmx.net; s=badeba3b8450; t=1553432868; bh=VqF7RKwtLhiu07J3lY56oZ2OC/x8VzK9zm/LH4IUZ7w=; h=X-UI-Sender-Class:Date:From:To:Subject; b=e26ZuXJQYYTtVz9NACFZSXSyNWJuHhLK+S0G/G3/fnNNinW3pXR4NKJ7F81nezaNF T0mm5AwpOLUv9uT8X6XNL2LdoCtJR6I2H7Vrz+QOw1Y+mEVFBTmYeLx2RJ6QPlmoU7 dI6HHMWmv7DRLclgqhTHxKZv+0djUlGnc2gkP7D4= X-UI-Sender-Class: 01bb95c1-4bf8-414a-932a-4f6e2808ef9c Received: from Valinor ([84.250.81.169]) by mail.gmx.com (mrgmx103 [212.227.17.174]) with ESMTPSA (Nemesis) id 0LpbfG-1gSgUc2uz0-00fSkM for ; Sun, 24 Mar 2019 14:07:48 +0100 Date: Sun, 24 Mar 2019 15:10:35 +0200 From: Lauri Kasanen To: ffmpeg-devel@ffmpeg.org Message-Id: <20190324151035.fe2360d570202b4c34ded513@gmx.com> X-Mailer: Sylpheed 3.5.0 (GTK+ 2.18.6; x86_64-unknown-linux-gnu) Mime-Version: 1.0 X-Provags-ID: V03:K1:6GNNKj0/4TllYgIjZ4s6rVQYfLtZdbC/VXDgUTW5IlYrg9rqerw +0xoglVS1MFg5+PUSLQfM0M1/JV+RwsQbiCOof7SL6qSltbzoUawNJZlKkjIt9t4MzIQjP5 S13bGHsQP0Z4/wZ85icxS5Z+k8ViSyIhYr/+sjIV4dwJcVrJ4oFXO6gb3otApN6AuienqOO RFhBB/HEVZ0xJ9TSJgo/w== X-Spam-Flag: NO X-UI-Out-Filterresults: notjunk:1; V03:K0:xohMZ5vmC4w=:p7MOmz3dmuOH+zYAzR+XX6 lPqmV4NQXq3O+DL9nhTE1702oBQd7BsJUiIFsMa959ZtUZspBjDoWKlBp4FtVDvPEzLpV/209 ZPQ3UlXEREAgBWRDSNIuKGXnkk/UXdP/E7Mx1CooVmvq5UeTQw91F5S7a3KGG5Ixm/zxx87sh yHaHPM3Mkcuxwl2dMNR+jr6v5Sdb6WQoXpVZk2266N2lZPaczqQ08dAi/eAG1iB9e7uNfCjSA 7My4coHo0CdOSLQybs6Qe9JssqdLw+Ux0z0cBAnFbSqTfjqK4hTe65Fd5gxEkPjKLAxoqrjhP 1NevKfIedHh2F7FLAukFRHGq+krs2iBPPxJwWJL3hUwlMqXTS+PUlThEu2W9vE6JW6ZvpUY2M 94qsmsTsYHgufun3w+6/PXh+Kc0Kfy2JJEFe++h1nHlLxhDKUKwSK/viJye9gsHMEWpi4RT/E 9kLhs7wEQuQV4GKtIhBEHUBzDk9lz46Yp29AADjJfLNJpVFdDgmKLYg/uKJUChcfKnyOVVnsP EUVMHPIwEzsdwrveyvRdJTE009NbqUq23MpY6Kc7vaIIt1HKRHejssXepqGdrQksapx4VQuc+ IVK4pllWmfg47MToksUcpCWyIz1DMO0/JcJzaUjGwrRlFIa8STLyq8oZvAnKNsozyOJxb8hyn +48RzpnRmn8qfF2aV69ES4TdP0r5OXSbBIUL5QlPpov7OPbtQYxmjg3x5yHDg4VbSogN8JBS7 5meq/QmGjeVKyaeUlhHSSGlinzcJV3q6O85qIYGGqoDjlG0giTBtPR0CwVT2GnMVqec5KHcaN xNxRWxcVkt28L+pOj02cbmkrk5AdvHU3PLVnTdSEa9R0IQW8VROkIDON0JtoQ1A6BnJwmqk4o swqvDwmCQu3uLEsYgdcwVqDCzEThiw1yjqawb0sEofMh2+MKgexTSrV9ns275V Subject: [FFmpeg-devel] [PATCH 1/3] swscale/ppc: VSX-optimize yuv2422_1 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 Errors-To: ffmpeg-devel-bounces@ffmpeg.org Sender: "ffmpeg-devel" ./ffmpeg -f lavfi -i yuvtestsrc=duration=1:size=1200x1440 \ -s 1200x1440 -f null -vframes 100 -pix_fmt $i -nostats \ -cpuflags 0 -v error - 15.3x speedup: yuyv422 14513 UNITS in yuv2packed1, 32768 runs, 0 skips 949 UNITS in yuv2packed1, 32767 runs, 1 skips yvyu422 14516 UNITS in yuv2packed1, 32767 runs, 1 skips 943 UNITS in yuv2packed1, 32767 runs, 1 skips uyvy422 14530 UNITS in yuv2packed1, 32767 runs, 1 skips 941 UNITS in yuv2packed1, 32766 runs, 2 skips Signed-off-by: Lauri Kasanen --- libswscale/ppc/swscale_vsx.c | 149 +++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 149 insertions(+) Series on top of "swscale/ppc: VSX-optimize yuv2rgb_full". -- 2.6.2 diff --git a/libswscale/ppc/swscale_vsx.c b/libswscale/ppc/swscale_vsx.c index 062ab0d..0bb82ac 100644 --- a/libswscale/ppc/swscale_vsx.c +++ b/libswscale/ppc/swscale_vsx.c @@ -664,6 +664,143 @@ YUV2RGBWRAPPER(yuv2, rgb_full, xbgr32_full, AV_PIX_FMT_ABGR, 0) YUV2RGBWRAPPER(yuv2, rgb_full, rgb24_full, AV_PIX_FMT_RGB24, 0) YUV2RGBWRAPPER(yuv2, rgb_full, bgr24_full, AV_PIX_FMT_BGR24, 0) +static av_always_inline void +write422(const vector int16_t vy1, const vector int16_t vy2, + const vector int16_t vu, const vector int16_t vv, + uint8_t *dest, const enum AVPixelFormat target) +{ + vector uint8_t vd1, vd2, tmp; + const vector uint8_t yuyv1 = (vector uint8_t) { + 0x0, 0x10, 0x1, 0x18, + 0x2, 0x11, 0x3, 0x19, + 0x4, 0x12, 0x5, 0x1a, + 0x6, 0x13, 0x7, 0x1b }; + const vector uint8_t yuyv2 = (vector uint8_t) { + 0x8, 0x14, 0x9, 0x1c, + 0xa, 0x15, 0xb, 0x1d, + 0xc, 0x16, 0xd, 0x1e, + 0xe, 0x17, 0xf, 0x1f }; + const vector uint8_t yvyu1 = (vector uint8_t) { + 0x0, 0x18, 0x1, 0x10, + 0x2, 0x19, 0x3, 0x11, + 0x4, 0x1a, 0x5, 0x12, + 0x6, 0x1b, 0x7, 0x13 }; + const vector uint8_t yvyu2 = (vector uint8_t) { + 0x8, 0x1c, 0x9, 0x14, + 0xa, 0x1d, 0xb, 0x15, + 0xc, 0x1e, 0xd, 0x16, + 0xe, 0x1f, 0xf, 0x17 }; + const vector uint8_t uyvy1 = (vector uint8_t) { + 0x10, 0x0, 0x18, 0x1, + 0x11, 0x2, 0x19, 0x3, + 0x12, 0x4, 0x1a, 0x5, + 0x13, 0x6, 0x1b, 0x7 }; + const vector uint8_t uyvy2 = (vector uint8_t) { + 0x14, 0x8, 0x1c, 0x9, + 0x15, 0xa, 0x1d, 0xb, + 0x16, 0xc, 0x1e, 0xd, + 0x17, 0xe, 0x1f, 0xf }; + + vd1 = vec_packsu(vy1, vy2); + vd2 = vec_packsu(vu, vv); + + switch (target) { + case AV_PIX_FMT_YUYV422: + tmp = vec_perm(vd1, vd2, yuyv1); + vec_st(tmp, 0, dest); + tmp = vec_perm(vd1, vd2, yuyv2); + vec_st(tmp, 16, dest); + break; + case AV_PIX_FMT_YVYU422: + tmp = vec_perm(vd1, vd2, yvyu1); + vec_st(tmp, 0, dest); + tmp = vec_perm(vd1, vd2, yvyu2); + vec_st(tmp, 16, dest); + break; + case AV_PIX_FMT_UYVY422: + tmp = vec_perm(vd1, vd2, uyvy1); + vec_st(tmp, 0, dest); + tmp = vec_perm(vd1, vd2, uyvy2); + vec_st(tmp, 16, dest); + break; + } +} + +static av_always_inline void +yuv2422_1_vsx_template(SwsContext *c, const int16_t *buf0, + const int16_t *ubuf[2], const int16_t *vbuf[2], + const int16_t *abuf0, uint8_t *dest, int dstW, + int uvalpha, int y, enum AVPixelFormat target) +{ + const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0]; + vector int16_t vy1, vy2, vu, vv, tmp; + const vector int16_t add64 = vec_splats((int16_t) 64); + const vector int16_t add128 = vec_splats((int16_t) 128); + const vector uint16_t shift7 = vec_splat_u16(7); + const vector uint16_t shift8 = vec_splat_u16(8); + int i; + + if (uvalpha < 2048) { + for (i = 0; i < ((dstW + 1) >> 1); i += 8) { + vy1 = vec_ld(0, &buf0[i * 2]); + vy2 = vec_ld(0, &buf0[(i + 4) * 2]); + vu = vec_ld(0, &ubuf0[i]); + vv = vec_ld(0, &vbuf0[i]); + + vy1 = vec_add(vy1, add64); + vy2 = vec_add(vy2, add64); + vu = vec_add(vu, add64); + vv = vec_add(vv, add64); + + vy1 = vec_sra(vy1, shift7); + vy2 = vec_sra(vy2, shift7); + vu = vec_sra(vu, shift7); + vv = vec_sra(vv, shift7); + + write422(vy1, vy2, vu, vv, &dest[i * 4], target); + } + } else { + const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1]; + for (i = 0; i < ((dstW + 1) >> 1); i += 8) { + vy1 = vec_ld(0, &buf0[i * 2]); + vy2 = vec_ld(0, &buf0[(i + 4) * 2]); + vu = vec_ld(0, &ubuf0[i]); + tmp = vec_ld(0, &ubuf1[i]); + vu = vec_adds(vu, tmp); + vv = vec_ld(0, &vbuf0[i]); + tmp = vec_ld(0, &vbuf1[i]); + vv = vec_adds(vv, tmp); + + vy1 = vec_add(vy1, add64); + vy2 = vec_add(vy2, add64); + vu = vec_adds(vu, add128); + vv = vec_adds(vv, add128); + + vy1 = vec_sra(vy1, shift7); + vy2 = vec_sra(vy2, shift7); + vu = vec_sra(vu, shift8); + vv = vec_sra(vv, shift8); + + write422(vy1, vy2, vu, vv, &dest[i * 4], target); + } + } +} + +#define YUV2PACKEDWRAPPER(name, base, ext, fmt) \ +static void name ## ext ## _1_vsx(SwsContext *c, const int16_t *buf0, \ + const int16_t *ubuf[2], const int16_t *vbuf[2], \ + const int16_t *abuf0, uint8_t *dest, int dstW, \ + int uvalpha, int y) \ +{ \ + name ## base ## _1_vsx_template(c, buf0, ubuf, vbuf, \ + abuf0, dest, dstW, uvalpha, \ + y, fmt); \ +} + +YUV2PACKEDWRAPPER(yuv2, 422, yuyv422, AV_PIX_FMT_YUYV422) +YUV2PACKEDWRAPPER(yuv2, 422, yvyu422, AV_PIX_FMT_YVYU422) +YUV2PACKEDWRAPPER(yuv2, 422, uyvy422, AV_PIX_FMT_UYVY422) + #endif /* !HAVE_BIGENDIAN */ #endif /* HAVE_VSX */ @@ -768,6 +905,18 @@ av_cold void ff_sws_init_swscale_vsx(SwsContext *c) } break; } + } else { /* !SWS_FULL_CHR_H_INT */ + switch (dstFormat) { + case AV_PIX_FMT_YUYV422: + c->yuv2packed1 = yuv2yuyv422_1_vsx; + break; + case AV_PIX_FMT_YVYU422: + c->yuv2packed1 = yuv2yvyu422_1_vsx; + break; + case AV_PIX_FMT_UYVY422: + c->yuv2packed1 = yuv2uyvy422_1_vsx; + break; + } } #endif /* !HAVE_BIGENDIAN */