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 022F54888B for ; Tue, 18 Jun 2024 20:32:18 +0000 (UTC) Received: from [127.0.1.1] (localhost [127.0.0.1]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id 5F03368D470; Tue, 18 Jun 2024 23:32:15 +0300 (EEST) Received: from mail-lf1-f48.google.com (mail-lf1-f48.google.com [209.85.167.48]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTPS id 90D2268D178 for ; Tue, 18 Jun 2024 23:32:08 +0300 (EEST) Received: by mail-lf1-f48.google.com with SMTP id 2adb3069b0e04-52c8c0d73d3so6124508e87.1 for ; Tue, 18 Jun 2024 13:32:08 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=martin-st.20230601.gappssmtp.com; s=20230601; t=1718742727; x=1719347527; darn=ffmpeg.org; h=mime-version:references:message-id:in-reply-to:subject:cc:to:from :date:from:to:cc:subject:date:message-id:reply-to; bh=35l8fvJRcXOpfF9AMyvVChogt9qPuiRyivff+1VPFWM=; b=BqbFmYpTdrrUfOQuezEfzFb1//BQceYidBSI2saZB9x+MzP4nXDgG6Y71sDTg0J3o1 UCkZeXucPyYkfIw1X8WZGwSsHdQF1ovbHYHbtfWubf2aF0PKW2GlA4olxRct1ZAgV4PY qnG18mjL/FQaNc+6uVw23jpZjCjZRVyfjvGjkpLadj4rZXRYO4SAoEyrqHDbzb5AlrUY F8z7oSPhUzOXTxM8RUlwgZgE1F36m9qBJpqiZ7xGN9wMh0YmKSrW9826z0IUfjDPgy1Y 7Si+Af12rQp3WSoiWLnR0AYsJtPjidHcW0a/33ATpIOpXDbV/nymP5C3P2BEC6WZaa2b OZ+g== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1718742727; x=1719347527; h=mime-version:references:message-id:in-reply-to:subject:cc:to:from :date:x-gm-message-state:from:to:cc:subject:date:message-id:reply-to; bh=35l8fvJRcXOpfF9AMyvVChogt9qPuiRyivff+1VPFWM=; b=QhbsWcCmnhHIuzCk70dVMJQTBBxdpar6czwLUmmOsujl0Gl9jndTc/DE5KBR98E/U/ 2HFsHRgq9xFm+HwsMLSUDFxTX3g4BnR41roA54ucYTa0n8WMCy/5SflQJI2dJWjst/C6 ZkXS8d0YSi0KitNXeTS20aN5hKwgozBPNBfx4z4qG3+fveRHVq5Z0U8O63M3oxpEg8h/ dryIbcDX8A8d5cEEhyolsJ9pBg8uafanQtnG5zV4U+w2OfcG/F+py/62/2qWd5lrV7si /Flx9EyW67kSPr7DP3vMc76Xqq8CfQpGOF6XayxkH1eHk7HDThlAsAgXYTN5mO9nr122 utWw== X-Gm-Message-State: AOJu0YyA3h+a51YOPaWhFXggudE3s8G1I2srdUvTs4hzOfBtMIYA6tSU Sy7A/6pNyRw7CK/w4mKzSftO1f3niK/xnHqGqBdeikzWtMMUvMNlp2oKk0Z8QKgDwSGRlbNpN6J 0BA== X-Google-Smtp-Source: AGHT+IFgSSh+dANIzsMmo3hER1SdAou5yA9rib2vOtgbhaUuYpZ7IFN0L8WdoiRk4Pc+V084o8v4nQ== X-Received: by 2002:a05:6512:b20:b0:52c:86eb:a2e6 with SMTP id 2adb3069b0e04-52ccaa56a21mr470113e87.4.1718742726849; Tue, 18 Jun 2024 13:32:06 -0700 (PDT) Received: from tunnel335574-pt.tunnel.tserv24.sto1.ipv6.he.net (tunnel335574-pt.tunnel.tserv24.sto1.ipv6.he.net. [2001:470:27:11::2]) by smtp.gmail.com with ESMTPSA id 2adb3069b0e04-52ca2887eccsm1641959e87.258.2024.06.18.13.32.06 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Tue, 18 Jun 2024 13:32:06 -0700 (PDT) Date: Tue, 18 Jun 2024 23:32:04 +0300 (EEST) From: =?ISO-8859-15?Q?Martin_Storsj=F6?= To: FFmpeg development discussions and patches In-Reply-To: Message-ID: References: <20240615095718.37319-1-quinkblack@foxmail.com> MIME-Version: 1.0 Subject: Re: [FFmpeg-devel] [PATCH 2/2] swscale/aarch64: Add bgra/rgba to yuv 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 Cc: Zhao Zhili Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset="us-ascii"; Format="flowed" Errors-To: ffmpeg-devel-bounces@ffmpeg.org Sender: "ffmpeg-devel" Archived-At: List-Archive: List-Post: On Sat, 15 Jun 2024, Zhao Zhili wrote: > From: Zhao Zhili > > Test on Apple M1 with kperf > > bgra_to_uv_8_c: 13.4 > bgra_to_uv_8_neon: 37.4 > bgra_to_uv_128_c: 155.9 > bgra_to_uv_128_neon: 91.7 > bgra_to_uv_1080_c: 1173.2 > bgra_to_uv_1080_neon: 822.7 > bgra_to_uv_1920_c: 2078.2 > bgra_to_uv_1920_neon: 1437.7 > bgra_to_uv_half_8_c: 17.9 > bgra_to_uv_half_8_neon: 37.4 > bgra_to_uv_half_128_c: 103.9 > bgra_to_uv_half_128_neon: 73.9 > bgra_to_uv_half_1080_c: 850.2 > bgra_to_uv_half_1080_neon: 484.2 > bgra_to_uv_half_1920_c: 1479.2 > bgra_to_uv_half_1920_neon: 824.2 > bgra_to_y_8_c: 8.2 > bgra_to_y_8_neon: 18.2 > bgra_to_y_128_c: 101.4 > bgra_to_y_128_neon: 74.9 > bgra_to_y_1080_c: 739.4 > bgra_to_y_1080_neon: 613.4 > bgra_to_y_1920_c: 1298.7 > bgra_to_y_1920_neon: 918.7 > --- > libswscale/aarch64/input.S | 81 +++++++++++++++++++++++++++++++----- > libswscale/aarch64/swscale.c | 16 +++++++ > 2 files changed, 86 insertions(+), 11 deletions(-) > > diff --git a/libswscale/aarch64/input.S b/libswscale/aarch64/input.S > index 2b956fe5c2..37f1158504 100644 > --- a/libswscale/aarch64/input.S > +++ b/libswscale/aarch64/input.S > @@ -20,8 +20,12 @@ > > #include "libavutil/aarch64/asm.S" > > -.macro rgb_to_yuv_load_rgb src > +.macro rgb_to_yuv_load_rgb src, element=3 > + .if \element == 3 > ld3 { v16.16b, v17.16b, v18.16b }, [\src] > + .else > + ld4 { v16.16b, v17.16b, v18.16b, v19.16b }, [\src] > + .endif > uxtl v19.8h, v16.8b // v19: r > uxtl v20.8h, v17.8b // v20: g > uxtl v21.8h, v18.8b // v21: b > @@ -43,7 +47,7 @@ > sqshrn2 \dst\().8h, \dst2\().4s, \right_shift // dst_higher_half = dst2 >> right_shift > .endm > > -.macro rgbToY bgr > +.macro rgbToY bgr, element=3 > cmp w4, #0 // check width > 0 > .if \bgr > ldr w12, [x5] // w12: ry > @@ -67,11 +71,15 @@ > dup v2.8h, w12 > b.lt 2f > 1: > - rgb_to_yuv_load_rgb x1 > + rgb_to_yuv_load_rgb x1, \element > rgb_to_yuv_product v19, v20, v21, v25, v26, v16, v0, v1, v2, #9 > rgb_to_yuv_product v22, v23, v24, v27, v28, v17, v0, v1, v2, #9 > sub w4, w4, #16 // width -= 16 > + .if \element == 3 > add x1, x1, #48 // src += 48 > + .else > + add x1, x1, #64 > + .endif I guess this also could be just #(16*\element) > cmp w4, #16 // width >= 16 ? > stp q16, q17, [x0], #32 // store to dst > b.ge 1b > @@ -86,7 +94,7 @@ > smaddl x13, w15, w12, x13 // x13 += by * b > asr w13, w13, #9 // x13 >>= 9 > sub w4, w4, #1 // width-- > - add x1, x1, #3 // src += 3 > + add x1, x1, \element Keep the # for the immediate constant here, i.e. #\element. Perhaps it doen't matter for most assemblers we use, but it's good to stay consistent. > strh w13, [x0], #2 // store to dst > cbnz w4, 2b > 3: > @@ -101,6 +109,14 @@ function ff_bgr24ToY_neon, export=1 > rgbToY bgr=1 > endfunc > > +function ff_rgba32ToY_neon, export=1 > + rgbToY bgr=0, element=4 > +endfunc > + > +function ff_bgra32ToY_neon, export=1 > + rgbToY bgr=1, element=4 > +endfunc > + > .macro rgb_load_uv_coeff half, bgr > .if \bgr > ldr w12, [x6, #12] > @@ -130,7 +146,7 @@ endfunc > dup v6.4s, w9 > .endm > > -.macro rgbToUV_half bgr > +.macro rgbToUV_half bgr, element=3 > cmp w5, #0 // check width > 0 > b.le 3f > > @@ -139,7 +155,11 @@ endfunc > b.lt 2f > // The following comments assume RGB order. The logic for RGB and BGR is the same. > 1: > + .if \element == 3 > ld3 { v16.16b, v17.16b, v18.16b }, [x3] > + .else > + ld4 { v16.16b, v17.16b, v18.16b, v19.16b }, [x3] > + .endif > uaddlp v19.8h, v16.16b // v19: r > uaddlp v20.8h, v17.16b // v20: g > uaddlp v21.8h, v18.16b // v21: b > @@ -147,7 +167,11 @@ endfunc > rgb_to_yuv_product v19, v20, v21, v22, v23, v16, v0, v1, v2, #10 > rgb_to_yuv_product v19, v20, v21, v24, v25, v17, v3, v4, v5, #10 > sub w5, w5, #8 // width -= 8 > - add x3, x3, #48 // src += 48 > + .if \element == 3 > + add x3, x3, #48 > + .else > + add x3, x3, #64 > + .endif > cmp w5, #8 // width >= 8 ? > str q16, [x0], #16 // store dst_u > str q17, [x1], #16 // store dst_v > @@ -155,9 +179,10 @@ endfunc > cbz w5, 3f > 2: > ldrb w2, [x3] // w2: r1 > - ldrb w4, [x3, #3] // w4: r2 > + ldrb w4, [x3, \element] // w4: r2 Ditto about keeping the # > add w2, w2, w4 // w2 = r1 + r2 > > + .if \element == 3 > ldrb w4, [x3, #1] // w4: g1 > ldrb w7, [x3, #4] // w7: g2 > add w4, w4, w7 // w4 = g1 + g2 > @@ -165,6 +190,15 @@ endfunc > ldrb w7, [x3, #2] // w7: b1 > ldrb w8, [x3, #5] // w8: b2 > add w7, w7, w8 // w7 = b1 + b2 > + .else > + ldrb w4, [x3, #1] // w4: g1 > + ldrb w7, [x3, #5] // w7: g2 > + add w4, w4, w7 // w4 = g1 + g2 > + > + ldrb w7, [x3, #2] // w7: b1 > + ldrb w8, [x3, #6] // w8: b2 > + add w7, w7, w8 // w7 = b1 + b2 > + .endif > > smaddl x8, w2, w10, x9 // dst_u = ru * r + const_offset > smaddl x8, w4, w11, x8 // dst_u += gu * g > @@ -177,7 +211,12 @@ endfunc > smaddl x8, w7, w15, x8 // dst_v += bv * b > asr x8, x8, #10 // dst_v >>= 10 > sub w5, w5, #1 > - add x3, x3, #6 // src += 6 > + ldrb w4, [x3, #1] // w4: g1 > + .if \element == 3 > + add x3, x3, #6 > + .else > + add x3, x3, #8 > + .endif And this can be #(2*\element) // Martin _______________________________________________ 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".