From patchwork Mon Apr 1 10:37:32 2019 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Lauri Kasanen X-Patchwork-Id: 12548 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 58A6044901F for ; Mon, 1 Apr 2019 13:37:49 +0300 (EEST) Received: from [127.0.1.1] (localhost [127.0.0.1]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id 3F95D68AC55; Mon, 1 Apr 2019 13:37:49 +0300 (EEST) X-Original-To: ffmpeg-devel@ffmpeg.org Delivered-To: ffmpeg-devel@ffmpeg.org Received: from mout.gmx.net (mout.gmx.net [212.227.17.20]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTPS id AADCD68AC22 for ; Mon, 1 Apr 2019 13:37:42 +0300 (EEST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=gmx.net; s=badeba3b8450; t=1554115062; bh=rWUXxP0YhCWJvgg8CnNB1hwoaNw31TWDB71Yvq8TrGk=; h=X-UI-Sender-Class:Date:From:To:Subject; b=jdesLJVZPaRs5RmClqvzaE/chm5oKwDBH28ZgXDm3WZFRaW4G6zEqobi85X6t6/1d fBwncvHm5HNN/iAUtszqodrDmVLu7yA4NScos2ID9lIF/BF9lxmyvXm1NFfjZkrTem jjKMWXBb3fS9hPcnpDG2FKvqfaBC/De0jRS7wSLM= X-UI-Sender-Class: 01bb95c1-4bf8-414a-932a-4f6e2808ef9c Received: from Valinor ([84.250.81.169]) by mail.gmx.com (mrgmx102 [212.227.17.174]) with ESMTPSA (Nemesis) id 0MMHaL-1hILN23vDh-007zQs for ; Mon, 01 Apr 2019 12:37:42 +0200 Date: Mon, 1 Apr 2019 13:37:32 +0300 From: Lauri Kasanen To: ffmpeg-devel@ffmpeg.org Message-Id: <20190401133732.7168c7e3c28b7d12866a40ed@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:XOZtSlkSBkUKwPKvgEH6ipsDBKwg389VePxvPD7XH4mI5RPiLBf 1pipPRvJNzJhTEwAZGusTcji5FCB5zbu1WcThHTRrInMkj2Psoq4JRbPN2dWRXBzKHk2izy the3zs73fYllO0MvlCV1MNr6wCrfQEJ5UMao9/a4cxuOWZActVyymCEudful3azEyGP4jrS kj9wCzxUb9qw+dnTCBDvA== X-Spam-Flag: NO X-UI-Out-Filterresults: notjunk:1; V03:K0:XBjPz/7yIxI=:Y9bHXXAf++/nwHEp3bqao9 PXuVkZyi++TF+RurvJysQNHAEDYXJWKHovXsxM4UT46+fg+8rPZFmq0bqI4mxrAq0A1bExra9 lRrOlhS8BsRdJKCdouTNQBcO6hO4N1VGAGVf3Kl3H6BOi3yBQQ/7tBSE8GgKBQxVZtHoZAXj/ LAkmr5+hY9vgQ/Uz+teblXlp9Y5Fg2xjZdylC4KF/1Pr0skluf7QKaCZrM84RyEVneRUz1+jR xF1cl18TV9RWJpim/M2xDFRnagAo3axlaHkjRnhLO6epcppAC05sSjrdrhXDkgyXWvHMcnyw4 3q678XbH02DB/NtIQOshyAW907JnJERZUFDG/hxNIs4fXkPZ85Cnn0iCG9VtWmaIXFQ+Xm/5H C67TjO/siQETzYIEF4DKnxD/9y0P7MhEaov1x+Ym3oqhETZb8M0GGPx1DT24tGe1NW9NjVvE5 qUeNmNLQp4cftUOfX3/jz30SRf3dDmXJafcEfFX59okuX4RVVMYdVCumIBH090ybDXD3LCIvU xAOfaMLFd6ElkJNYdd4A/niv1aR+FWr5V6toMFBXjHqGy4lM/3behxNGJIkDERzDxmb9PYLpC X8awEes7jpHfBcdzrP2G1CiqhZk+K2k5ohOIKAj50LOZXaL+TbYsmYaqF2Z2NH89XuOtfMVAS /C7uOZNItr2CI4E8l6fsm77bOUn5myau04NcXzYJ3i/3eMcoG9HLzFCGQIG5dCmf7OoVZOEqd Vh0lNj6vkYpgbsPEaoftR03ApImlDXv7ky8f9N2R803RVBUUzv4uXPUpcM05FiB6nANoE1Z98 w7OibTqhMu7qC6SezS5n3bOMCtSlEvigttQuz20whK49/e6fguKtkUYhvFBqKl0TywBqZIMNP 7juDyh4QNCAT9BUQjtXUUuSjeaToO7v6w8N6FnGCR/U5KEiAPOGTyxz/A1I0Kl Subject: [FFmpeg-devel] [PATCH] swscale/ppc: VSX-optimize yuv2rgb_full_X 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 1200x720 -f null -vframes 100 -pix_fmt $i -nostats \ -cpuflags 0 -v error - 32-bit mul, power8 only. ~6.4x speedup: rgb24 214278 UNITS in yuv2packedX, 16384 runs, 0 skips 33249 UNITS in yuv2packedX, 16384 runs, 0 skips bgr24 214616 UNITS in yuv2packedX, 16384 runs, 0 skips 33233 UNITS in yuv2packedX, 16384 runs, 0 skips rgba 214517 UNITS in yuv2packedX, 16384 runs, 0 skips 33271 UNITS in yuv2packedX, 16384 runs, 0 skips bgra 214973 UNITS in yuv2packedX, 16384 runs, 0 skips 33397 UNITS in yuv2packedX, 16384 runs, 0 skips argb 214613 UNITS in yuv2packedX, 16384 runs, 0 skips 33310 UNITS in yuv2packedX, 16384 runs, 0 skips bgra 214637 UNITS in yuv2packedX, 16384 runs, 0 skips 33330 UNITS in yuv2packedX, 16384 runs, 0 skips Signed-off-by: Lauri Kasanen --- libswscale/ppc/swscale_vsx.c | 160 +++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 160 insertions(+) -- 2.6.2 diff --git a/libswscale/ppc/swscale_vsx.c b/libswscale/ppc/swscale_vsx.c index 6ff8b62..e05f9ec 100644 --- a/libswscale/ppc/swscale_vsx.c +++ b/libswscale/ppc/swscale_vsx.c @@ -520,6 +520,139 @@ yuv2NBPSX(16, LE, 0, 16, int32_t) break; \ } +static av_always_inline void +yuv2rgb_full_X_vsx_template(SwsContext *c, const int16_t *lumFilter, + const int16_t **lumSrc, int lumFilterSize, + const int16_t *chrFilter, const int16_t **chrUSrc, + const int16_t **chrVSrc, int chrFilterSize, + const int16_t **alpSrc, uint8_t *dest, + int dstW, int y, enum AVPixelFormat target, int hasAlpha) +{ + vector int16_t vv; + vector int32_t vy32_l, vy32_r, vu32_l, vu32_r, vv32_l, vv32_r, tmp32; + vector int32_t R_l, R_r, G_l, G_r, B_l, B_r; + vector int32_t tmp, tmp2, tmp3, tmp4; + vector uint16_t rd16, gd16, bd16; + vector uint8_t rd, bd, gd, ad, out0, out1, tmp8; + vector int16_t vlumFilter[MAX_FILTER_SIZE], vchrFilter[MAX_FILTER_SIZE]; + const vector int32_t ystart = vec_splats(1 << 9); + const vector int32_t uvstart = vec_splats((1 << 9) - (128 << 19)); + const vector uint16_t zero16 = vec_splat_u16(0); + const vector int32_t y_offset = vec_splats(c->yuv2rgb_y_offset); + const vector int32_t y_coeff = vec_splats(c->yuv2rgb_y_coeff); + const vector int32_t y_add = vec_splats(1 << 21); + const vector int32_t v2r_coeff = vec_splats(c->yuv2rgb_v2r_coeff); + const vector int32_t v2g_coeff = vec_splats(c->yuv2rgb_v2g_coeff); + const vector int32_t u2g_coeff = vec_splats(c->yuv2rgb_u2g_coeff); + const vector int32_t u2b_coeff = vec_splats(c->yuv2rgb_u2b_coeff); + const vector int32_t rgbclip = vec_splats(1 << 30); + const vector int32_t zero32 = vec_splat_s32(0); + const vector uint32_t shift22 = vec_splats(22U); + const vector uint32_t shift10 = vec_splat_u32(10); + int i, j; + + // Various permutations + const vector uint8_t perm3rg0 = (vector uint8_t) {0x0, 0x10, 0, + 0x1, 0x11, 0, + 0x2, 0x12, 0, + 0x3, 0x13, 0, + 0x4, 0x14, 0, + 0x5 }; + const vector uint8_t perm3rg1 = (vector uint8_t) { 0x15, 0, + 0x6, 0x16, 0, + 0x7, 0x17, 0 }; + const vector uint8_t perm3tb0 = (vector uint8_t) {0x0, 0x1, 0x10, + 0x3, 0x4, 0x11, + 0x6, 0x7, 0x12, + 0x9, 0xa, 0x13, + 0xc, 0xd, 0x14, + 0xf }; + const vector uint8_t perm3tb1 = (vector uint8_t) { 0x0, 0x15, + 0x2, 0x3, 0x16, + 0x5, 0x6, 0x17 }; + + ad = vec_splats((uint8_t) 255); + + for (i = 0; i < lumFilterSize; i++) + vlumFilter[i] = vec_splats(lumFilter[i]); + for (i = 0; i < chrFilterSize; i++) + vchrFilter[i] = vec_splats(chrFilter[i]); + + for (i = 0; i < dstW; i += 8) { + vy32_l = + vy32_r = ystart; + vu32_l = + vu32_r = + vv32_l = + vv32_r = uvstart; + + for (j = 0; j < lumFilterSize; j++) { + vv = vec_ld(0, &lumSrc[j][i]); + tmp = vec_mule(vv, vlumFilter[j]); + tmp2 = vec_mulo(vv, vlumFilter[j]); + tmp3 = vec_mergeh(tmp, tmp2); + tmp4 = vec_mergel(tmp, tmp2); + + vy32_l = vec_adds(vy32_l, tmp3); + vy32_r = vec_adds(vy32_r, tmp4); + } + + for (j = 0; j < chrFilterSize; j++) { + vv = vec_ld(0, &chrUSrc[j][i]); + tmp = vec_mule(vv, vchrFilter[j]); + tmp2 = vec_mulo(vv, vchrFilter[j]); + tmp3 = vec_mergeh(tmp, tmp2); + tmp4 = vec_mergel(tmp, tmp2); + + vu32_l = vec_adds(vu32_l, tmp3); + vu32_r = vec_adds(vu32_r, tmp4); + + vv = vec_ld(0, &chrVSrc[j][i]); + tmp = vec_mule(vv, vchrFilter[j]); + tmp2 = vec_mulo(vv, vchrFilter[j]); + tmp3 = vec_mergeh(tmp, tmp2); + tmp4 = vec_mergel(tmp, tmp2); + + vv32_l = vec_adds(vv32_l, tmp3); + vv32_r = vec_adds(vv32_r, tmp4); + } + + vy32_l = vec_sra(vy32_l, shift10); + vy32_r = vec_sra(vy32_r, shift10); + vu32_l = vec_sra(vu32_l, shift10); + vu32_r = vec_sra(vu32_r, shift10); + vv32_l = vec_sra(vv32_l, shift10); + vv32_r = vec_sra(vv32_r, shift10); + + vy32_l = vec_sub(vy32_l, y_offset); + vy32_r = vec_sub(vy32_r, y_offset); + vy32_l = vec_mul(vy32_l, y_coeff); + vy32_r = vec_mul(vy32_r, y_coeff); + vy32_l = vec_add(vy32_l, y_add); + vy32_r = vec_add(vy32_r, y_add); + + R_l = vec_mul(vv32_l, v2r_coeff); + R_l = vec_add(R_l, vy32_l); + R_r = vec_mul(vv32_r, v2r_coeff); + R_r = vec_add(R_r, vy32_r); + G_l = vec_mul(vv32_l, v2g_coeff); + tmp32 = vec_mul(vu32_l, u2g_coeff); + G_l = vec_add(G_l, vy32_l); + G_l = vec_add(G_l, tmp32); + G_r = vec_mul(vv32_r, v2g_coeff); + tmp32 = vec_mul(vu32_r, u2g_coeff); + G_r = vec_add(G_r, vy32_r); + G_r = vec_add(G_r, tmp32); + + B_l = vec_mul(vu32_l, u2b_coeff); + B_l = vec_add(B_l, vy32_l); + B_r = vec_mul(vu32_r, u2b_coeff); + B_r = vec_add(B_r, vy32_r); + + WRITERGB + } +} + #define SETUP(x, buf0, alpha1, buf1, alpha) { \ x = vec_ld(0, buf0); \ tmp = vec_mule(x, alpha1); \ @@ -977,6 +1110,19 @@ yuv2rgb_1_vsx_template(SwsContext *c, const int16_t *buf0, #undef WRITERGB +#define YUV2RGBWRAPPERX(name, base, ext, fmt, hasAlpha) \ +static void name ## ext ## _X_vsx(SwsContext *c, const int16_t *lumFilter, \ + const int16_t **lumSrc, int lumFilterSize, \ + const int16_t *chrFilter, const int16_t **chrUSrc, \ + const int16_t **chrVSrc, int chrFilterSize, \ + const int16_t **alpSrc, uint8_t *dest, int dstW, \ + int y) \ +{ \ + name ## base ## _X_vsx_template(c, lumFilter, lumSrc, lumFilterSize, \ + chrFilter, chrUSrc, chrVSrc, chrFilterSize, \ + alpSrc, dest, dstW, y, fmt, hasAlpha); \ +} + #define YUV2RGBWRAPPERX2(name, base, ext, fmt, hasAlpha) \ static void name ## ext ## _2_vsx(SwsContext *c, const int16_t *buf[2], \ const int16_t *ubuf[2], const int16_t *vbuf[2], \ @@ -1021,6 +1167,14 @@ YUV2RGBWRAPPERX2(yuv2, rgb_full, xbgr32_full, AV_PIX_FMT_ABGR, 0) YUV2RGBWRAPPERX2(yuv2, rgb_full, rgb24_full, AV_PIX_FMT_RGB24, 0) YUV2RGBWRAPPERX2(yuv2, rgb_full, bgr24_full, AV_PIX_FMT_BGR24, 0) +YUV2RGBWRAPPERX(yuv2, rgb_full, bgrx32_full, AV_PIX_FMT_BGRA, 0) +YUV2RGBWRAPPERX(yuv2, rgb_full, rgbx32_full, AV_PIX_FMT_RGBA, 0) +YUV2RGBWRAPPERX(yuv2, rgb_full, xrgb32_full, AV_PIX_FMT_ARGB, 0) +YUV2RGBWRAPPERX(yuv2, rgb_full, xbgr32_full, AV_PIX_FMT_ABGR, 0) + +YUV2RGBWRAPPERX(yuv2, rgb_full, rgb24_full, AV_PIX_FMT_RGB24, 0) +YUV2RGBWRAPPERX(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, @@ -1394,12 +1548,14 @@ av_cold void ff_sws_init_swscale_vsx(SwsContext *c) if (HAVE_POWER8 && cpu_flags & AV_CPU_FLAG_POWER8) { c->yuv2packed1 = yuv2rgb24_full_1_vsx; c->yuv2packed2 = yuv2rgb24_full_2_vsx; + c->yuv2packedX = yuv2rgb24_full_X_vsx; } break; case AV_PIX_FMT_BGR24: if (HAVE_POWER8 && cpu_flags & AV_CPU_FLAG_POWER8) { c->yuv2packed1 = yuv2bgr24_full_1_vsx; c->yuv2packed2 = yuv2bgr24_full_2_vsx; + c->yuv2packedX = yuv2bgr24_full_X_vsx; } break; case AV_PIX_FMT_BGRA: @@ -1407,6 +1563,7 @@ av_cold void ff_sws_init_swscale_vsx(SwsContext *c) if (!c->needAlpha) { c->yuv2packed1 = yuv2bgrx32_full_1_vsx; c->yuv2packed2 = yuv2bgrx32_full_2_vsx; + c->yuv2packedX = yuv2bgrx32_full_X_vsx; } } break; @@ -1415,6 +1572,7 @@ av_cold void ff_sws_init_swscale_vsx(SwsContext *c) if (!c->needAlpha) { c->yuv2packed1 = yuv2rgbx32_full_1_vsx; c->yuv2packed2 = yuv2rgbx32_full_2_vsx; + c->yuv2packedX = yuv2rgbx32_full_X_vsx; } } break; @@ -1423,6 +1581,7 @@ av_cold void ff_sws_init_swscale_vsx(SwsContext *c) if (!c->needAlpha) { c->yuv2packed1 = yuv2xrgb32_full_1_vsx; c->yuv2packed2 = yuv2xrgb32_full_2_vsx; + c->yuv2packedX = yuv2xrgb32_full_X_vsx; } } break; @@ -1431,6 +1590,7 @@ av_cold void ff_sws_init_swscale_vsx(SwsContext *c) if (!c->needAlpha) { c->yuv2packed1 = yuv2xbgr32_full_1_vsx; c->yuv2packed2 = yuv2xbgr32_full_2_vsx; + c->yuv2packedX = yuv2xbgr32_full_X_vsx; } } break;