From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from ffbox0-bg.mplayerhq.hu (ffbox0-bg.ffmpeg.org [79.124.17.100]) by master.gitmailbox.com (Postfix) with ESMTP id 3080C4CC26 for ; Thu, 15 Aug 2024 14:44:17 +0000 (UTC) Received: from [127.0.1.1] (localhost [127.0.0.1]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id A7AB368DBA8; Thu, 15 Aug 2024 17:44:14 +0300 (EEST) Received: from mail-wr1-f50.google.com (mail-wr1-f50.google.com [209.85.221.50]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTPS id 63D9668D79D for ; Thu, 15 Aug 2024 17:44:08 +0300 (EEST) Received: by mail-wr1-f50.google.com with SMTP id ffacd0b85a97d-3687f91af40so608306f8f.0 for ; Thu, 15 Aug 2024 07:44:08 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20230601; t=1723733047; x=1724337847; darn=ffmpeg.org; h=content-transfer-encoding:mime-version:message-id:date:subject:to :from:from:to:cc:subject:date:message-id:reply-to; bh=gX5ZlGFn0oEeyW/jc2sePkB0/AL8XX48C8e8PLYsbZM=; b=NK6mzYyan8fy9aaDwZ4hFED7FgYGShI/QGCLbPztoQyXWmmvO/QkjrNdV5XftGS+Sv xMqkQP5qVELzZA567mlIE3opY+SNoc+VB1fU3i0zo+kvAxtjHlC+/iWeGNahVx6FW3zd x4tilcdIB83HBuVlsVMwPTk0AP3V5/7Ejf3XyZQon0hvQIFDJl5RiewPZWnnM8aHuKbn /+Daz4jO1GotN88B73EEhpp3qzSq0hC9hKcwicU1rvCjNo+VHMK7x6kJjxkIThs9id4F j5JvVUygJ7D0SsLwY1Iw7RIrNSiRzqzOR4yLDUnXNJ6h981cigXRbZki3HN4D1rdnwjw +CGA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1723733047; x=1724337847; h=content-transfer-encoding:mime-version:message-id:date:subject:to :from:x-gm-message-state:from:to:cc:subject:date:message-id:reply-to; bh=gX5ZlGFn0oEeyW/jc2sePkB0/AL8XX48C8e8PLYsbZM=; b=rEZG21B16102mhuxycLpY9MPp2NeDqLklpqO35bT/zYiFtYrNDQEzu3/tJ/afl7KTb PzfJlPeAuK+PUshwt1GkHEQHdo3rKPAOx9MIqp69bQptNZPV37cdnBpmTc/rtuPm83eo IVAiGx46vqCMracnn/gjSDa/SKPh9cEJJhnEdwlObWd5p1FDohelFbzYtw9uZXMyydTg si6QKVjGn/JU9zBL/4OuiClqwYZnI1WXlP+REFOQ30xLn34rYhbaulbV0KH6W+yH0AU+ osfE/6ia0DPMU23Q+c1WqeGQfBoeuM2sV5ZayWmKqb4vbz8BPIN8h1Y5uMYsBCRV8xem oL1w== X-Gm-Message-State: AOJu0YxsmK3PWOHrVceMwefb+B5TgMWeOMGjnYsg9BnbbgcqLNtp4ime 9KSXdw/By2u5+QHxIM1/XG2uPgFUPDSpKmI5kAd3hWx0mid49rSUPS/u5io7 X-Google-Smtp-Source: AGHT+IHJRjU3wy4xrIDuWMfhnXHMws7CDr0k/M4xneuwY1/R0GAUoWMnLq7+T1gHlsnNBV0wc6xu5g== X-Received: by 2002:a05:6000:b44:b0:368:334d:aad4 with SMTP id ffacd0b85a97d-371777696d6mr4510264f8f.4.1723733046937; Thu, 15 Aug 2024 07:44:06 -0700 (PDT) Received: from localhost.localdomain (205.66-130-109.adsl-dyn.isp.belgacom.be. [109.130.66.205]) by smtp.gmail.com with ESMTPSA id ffacd0b85a97d-37189896c5bsm1697189f8f.80.2024.08.15.07.44.05 for (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Thu, 15 Aug 2024 07:44:06 -0700 (PDT) From: Ramiro Polla To: ffmpeg-devel@ffmpeg.org Date: Thu, 15 Aug 2024 16:44:05 +0200 Message-Id: <20240815144405.17971-1-ramiro.polla@gmail.com> X-Mailer: git-send-email 2.30.2 MIME-Version: 1.0 Subject: [FFmpeg-devel] [PATCH] avcodec/aarch64/me_cmp: add dotprod implementations of sse16 and vsse_intra16 X-BeenThere: ffmpeg-devel@ffmpeg.org X-Mailman-Version: 2.1.29 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 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Errors-To: ffmpeg-devel-bounces@ffmpeg.org Sender: "ffmpeg-devel" Archived-At: List-Archive: List-Post: checkasm --bench for Raspberry Pi 5 Model B Rev 1.0: sse_0_c: 241.5 sse_0_neon: 37.2 sse_0_dotprod: 22.2 vsse_4_c: 148.7 vsse_4_neon: 31.0 vsse_4_dotprod: 15.7 --- libavcodec/aarch64/me_cmp_init_aarch64.c | 14 +++ libavcodec/aarch64/me_cmp_neon.S | 114 +++++++++++++++++++++++ 2 files changed, 128 insertions(+) diff --git a/libavcodec/aarch64/me_cmp_init_aarch64.c b/libavcodec/aarch64/me_cmp_init_aarch64.c index 1e0f1cf4f1..fa2724403d 100644 --- a/libavcodec/aarch64/me_cmp_init_aarch64.c +++ b/libavcodec/aarch64/me_cmp_init_aarch64.c @@ -77,6 +77,13 @@ int vsse8_neon(MpegEncContext *c, const uint8_t *s1, const uint8_t *s2, int vsse_intra8_neon(MpegEncContext *c, const uint8_t *s, const uint8_t *dummy, ptrdiff_t stride, int h); +#if HAVE_DOTPROD +int sse16_neon_dotprod(MpegEncContext *v, const uint8_t *pix1, const uint8_t *pix2, + ptrdiff_t stride, int h); +int vsse_intra16_neon_dotprod(MpegEncContext *c, const uint8_t *s1, const uint8_t *s2, + ptrdiff_t stride, int h); +#endif + av_cold void ff_me_cmp_init_aarch64(MECmpContext *c, AVCodecContext *avctx) { int cpu_flags = av_get_cpu_flags(); @@ -113,6 +120,13 @@ av_cold void ff_me_cmp_init_aarch64(MECmpContext *c, AVCodecContext *avctx) c->median_sad[0] = pix_median_abs16_neon; c->median_sad[1] = pix_median_abs8_neon; } + +#if HAVE_DOTPROD + if (have_dotprod(cpu_flags)) { + c->sse[0] = sse16_neon_dotprod; + c->vsse[4] = vsse_intra16_neon_dotprod; + } +#endif } int nsse16_neon_wrapper(MpegEncContext *c, const uint8_t *s1, const uint8_t *s2, diff --git a/libavcodec/aarch64/me_cmp_neon.S b/libavcodec/aarch64/me_cmp_neon.S index 7500c324bd..20e3b33a83 100644 --- a/libavcodec/aarch64/me_cmp_neon.S +++ b/libavcodec/aarch64/me_cmp_neon.S @@ -1559,3 +1559,117 @@ function pix_median_abs8_neon, export=1 ret endfunc + +#if HAVE_DOTPROD +ENABLE_DOTPROD + +function sse16_neon_dotprod, export=1 + // x0 - unused + // x1 - pix1 + // x2 - pix2 + // x3 - stride + // w4 - h + + cmp w4, #4 + movi v17.4s, #0 + b.lt 2f + +// Make 4 iterations at once +1: + + // res = abs(pix1[0] - pix2[0]) + // res * res + + ld1 {v0.16b}, [x1], x3 // Load pix1 vector for first iteration + ld1 {v1.16b}, [x2], x3 // Load pix2 vector for first iteration + ld1 {v2.16b}, [x1], x3 // Load pix1 vector for second iteration + uabd v30.16b, v0.16b, v1.16b // Absolute difference, first iteration + ld1 {v3.16b}, [x2], x3 // Load pix2 vector for second iteration + udot v17.4s, v30.16b, v30.16b + uabd v27.16b, v2.16b, v3.16b // Absolute difference, second iteration + ld1 {v4.16b}, [x1], x3 // Load pix1 for third iteration + udot v17.4s, v27.16b, v27.16b + ld1 {v5.16b}, [x2], x3 // Load pix2 for third iteration + uabd v24.16b, v4.16b, v5.16b // Absolute difference, third iteration + ld1 {v6.16b}, [x1], x3 // Load pix1 for fourth iteration + udot v17.4s, v24.16b, v24.16b + ld1 {v7.16b}, [x2], x3 // Load pix2 for fouth iteration + uabd v21.16b, v6.16b, v7.16b // Absolute difference, fourth iteration + sub w4, w4, #4 // h -= 4 + udot v17.4s, v21.16b, v21.16b + cmp w4, #4 + b.ge 1b + + cbz w4, 3f + +// iterate by one +2: + + ld1 {v0.16b}, [x1], x3 // Load pix1 + ld1 {v1.16b}, [x2], x3 // Load pix2 + + uabd v30.16b, v0.16b, v1.16b + subs w4, w4, #1 + udot v17.4s, v30.16b, v30.16b + + b.ne 2b + +3: + uaddlv d16, v17.4s // add up accumulator vector + + fmov w0, s16 + + ret +endfunc + +function vsse_intra16_neon_dotprod, export=1 + // x0 unused + // x1 uint8_t *pix1 + // x2 uint8_t *dummy + // x3 ptrdiff_t stride + // w4 int h + + ld1 {v0.16b}, [x1], x3 + movi v17.4s, #0 + + sub w4, w4, #1 // we need to make h-1 iterations + cmp w4, #3 + b.lt 2f + +1: + // v = abs( pix1[0] - pix1[0 + stride] ) + // score = sum( v * v ) + ld1 {v1.16b}, [x1], x3 + ld1 {v2.16b}, [x1], x3 + uabd v30.16b, v0.16b, v1.16b + ld1 {v3.16b}, [x1], x3 + udot v17.4s, v30.16b, v30.16b + uabd v27.16b, v1.16b, v2.16b + udot v17.4s, v27.16b, v27.16b + uabd v25.16b, v2.16b, v3.16b + sub w4, w4, #3 + udot v17.4s, v25.16b, v25.16b + cmp w4, #3 + mov v0.16b, v3.16b + + b.ge 1b + cbz w4, 3f + +// iterate by one +2: + ld1 {v1.16b}, [x1], x3 + subs w4, w4, #1 + uabd v30.16b, v0.16b, v1.16b + mov v0.16b, v1.16b + udot v17.4s, v30.16b, v30.16b + cbnz w4, 2b + +3: + uaddlv d17, v17.4s + fmov w0, s17 + + ret +endfunc + +DISABLE_DOTPROD +#endif -- 2.39.2 _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".