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 8B84A4A9E6 for ; Thu, 6 Jun 2024 15:45:35 +0000 (UTC) Received: from [127.0.1.1] (localhost [127.0.0.1]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id 5F72768D6FF; Thu, 6 Jun 2024 18:45:32 +0300 (EEST) Received: from mail-il1-f172.google.com (mail-il1-f172.google.com [209.85.166.172]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTPS id 839BD68D628 for ; Thu, 6 Jun 2024 18:45:26 +0300 (EEST) Received: by mail-il1-f172.google.com with SMTP id e9e14a558f8ab-36dd56cf5f5so4285375ab.3 for ; Thu, 06 Jun 2024 08:45:26 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20230601; t=1717688724; x=1718293524; darn=ffmpeg.org; h=content-transfer-encoding:in-reply-to:from:content-language :references:to:subject:user-agent:mime-version:date:message-id:from :to:cc:subject:date:message-id:reply-to; bh=wfLWGcFg/Nq6g7+dp9Xyt1c9YglzTSDlyKITpq3i/Nw=; b=XVdJ9lsUgiUn3cJPLISFm1BzB+vrBpCr97Jdv320SHcfHjAwN0gSZODane0l7+mHiC cBBD1eYBIr1S205654JCONFVBoE4AEUGbYFNgFpldUto98P5eZTdq6xt+ID9ej4v9fSk UfEBIvoqBpL2d6tV1XVsUXRLQrHe71u4tBopcIIXfmw3zQm6PhtOeVxQpTejMPK4Ew3e O9vtIrfeeF048PVkiUSxzIsFZfdhi8AsewHAQF4r+vMbRyi49eQS16OdK7YeFtbLr0o1 rMjjoS4apb6b+dG9smVNYyma7b3rgDM2U3xwNgdDlGK7nYziczjY/FFTeULftUfdEPrj gyiw== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1717688724; x=1718293524; h=content-transfer-encoding:in-reply-to:from:content-language :references:to:subject:user-agent:mime-version:date:message-id :x-gm-message-state:from:to:cc:subject:date:message-id:reply-to; bh=wfLWGcFg/Nq6g7+dp9Xyt1c9YglzTSDlyKITpq3i/Nw=; b=dcXCvWqnsb0NDohanJwyrWWqpH5XKQRHCOGj7yCUjXz6NGaU9ArzFNx/k+uCLob4LU 2oP2Xve+ZgtNrzQw/N1gXnJJYNcAsXtkBUjDfPIWx58uzrfnjaAyT8zdR+JcWNig3qNc L3vyWA45ICJASvWK/lfEe9STDRO6yINJcHL4c+RkRCkN1hviOfH+cJsaSNtMJV6Pz1/+ w2JsxWsIDx+Um5/W979GVMPRX3ZvsTsT5MTFFbpFTkFW46E2gLwLk7jDpSGjhnV4yern LQHkPW6/U5/MOYDoYlzuNLy8hEr19TDo+9rz/BlOnMCumkNQVhUncIVd2LHE7odQqcIS cPEg== X-Gm-Message-State: AOJu0YzBIDX67HP0MWzyVyRNyelb7eGfvKADs43VKY2ipkAX0ZMwU3RO LlwU3Rjt8nhGzO/uruinhQtcYxBM5YZ7RBtMA6ItbRMM6NBC1qyPcQrK7w== X-Google-Smtp-Source: AGHT+IGF+u1E+uV2b54pn4U3aqkqLLsIk1nMhtf8AouRlqloHxhFeDvx/SFizoQp1pFWWVyMdFesXQ== X-Received: by 2002:a05:6e02:1d07:b0:374:9b42:1724 with SMTP id e9e14a558f8ab-37580388278mr661955ab.29.1717688723630; Thu, 06 Jun 2024 08:45:23 -0700 (PDT) Received: from [192.168.0.16] ([190.194.167.233]) by smtp.gmail.com with ESMTPSA id 41be03b00d2f7-6de2760a672sm1273945a12.78.2024.06.06.08.45.22 for (version=TLS1_3 cipher=TLS_AES_128_GCM_SHA256 bits=128/128); Thu, 06 Jun 2024 08:45:23 -0700 (PDT) Message-ID: <748be8f7-6dde-49c5-bf8a-66b8f7e8f3f0@gmail.com> Date: Thu, 6 Jun 2024 12:45:36 -0300 MIME-Version: 1.0 User-Agent: Mozilla Thunderbird To: ffmpeg-devel@ffmpeg.org References: <20240605205116.3258-1-jamrial@gmail.com> <20240606141505.132-1-jamrial@gmail.com> Content-Language: en-US From: James Almer In-Reply-To: Subject: Re: [FFmpeg-devel] [PATCH] swscale/x86/rgb2rgb: add SSE2 shuffle_bytes functions 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-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 6/6/2024 11:48 AM, Andreas Rheinhardt wrote: > James Almer: >> And remove shuffle_bytes_2103_mmxext. >> >> shuffle_bytes_0321_c: 28.1 >> shuffle_bytes_0321_sse2: 13.6 >> shuffle_bytes_0321_ssse3: 9.6 >> shuffle_bytes_0321_avx2: 7.1 >> shuffle_bytes_1230_c: 52.6 >> shuffle_bytes_1230_sse2: 12.1 >> shuffle_bytes_1230_ssse3: 8.6 >> shuffle_bytes_1230_avx2: 6.6 >> shuffle_bytes_2103_c: 29.1 >> shuffle_bytes_2103_mmxext: 29.3 // removed >> shuffle_bytes_2103_sse2: 12.5 >> shuffle_bytes_2103_ssse3: 8.6 >> shuffle_bytes_2103_avx2: 7.1 >> shuffle_bytes_3012_c: 52.1 >> shuffle_bytes_3012_sse2: 12.1 >> shuffle_bytes_3012_ssse3: 8.6 >> shuffle_bytes_3012_avx2: 7.1 >> shuffle_bytes_3210_c: 50.6 >> shuffle_bytes_3210_sse2: 14.6 >> shuffle_bytes_3210_ssse3: 8.6 >> shuffle_bytes_3210_avx2: 7.1 >> >> Signed-off-by: James Almer >> --- >> libswscale/x86/rgb2rgb.c | 14 ++++-- >> libswscale/x86/rgb_2_rgb.asm | 83 +++++++++++++++++++++++++----------- >> 2 files changed, 69 insertions(+), 28 deletions(-) >> >> diff --git a/libswscale/x86/rgb2rgb.c b/libswscale/x86/rgb2rgb.c >> index 21ccfafe51..9f6c8efc72 100644 >> --- a/libswscale/x86/rgb2rgb.c >> +++ b/libswscale/x86/rgb2rgb.c >> @@ -116,7 +116,11 @@ DECLARE_ALIGNED(8, extern const uint64_t, ff_bgr2UVOffset); >> >> #endif /* HAVE_INLINE_ASM */ >> >> -void ff_shuffle_bytes_2103_mmxext(const uint8_t *src, uint8_t *dst, int src_size); >> +void ff_shuffle_bytes_2103_sse2(const uint8_t *src, uint8_t *dst, int src_size); >> +void ff_shuffle_bytes_0321_sse2(const uint8_t *src, uint8_t *dst, int src_size); >> +void ff_shuffle_bytes_1230_sse2(const uint8_t *src, uint8_t *dst, int src_size); >> +void ff_shuffle_bytes_3012_sse2(const uint8_t *src, uint8_t *dst, int src_size); >> +void ff_shuffle_bytes_3210_sse2(const uint8_t *src, uint8_t *dst, int src_size); >> void ff_shuffle_bytes_2103_ssse3(const uint8_t *src, uint8_t *dst, int src_size); >> void ff_shuffle_bytes_0321_ssse3(const uint8_t *src, uint8_t *dst, int src_size); >> void ff_shuffle_bytes_1230_ssse3(const uint8_t *src, uint8_t *dst, int src_size); >> @@ -154,10 +158,12 @@ av_cold void rgb2rgb_init_x86(void) >> rgb2rgb_init_avx(); >> #endif /* HAVE_INLINE_ASM */ >> >> - if (EXTERNAL_MMXEXT(cpu_flags)) { >> - shuffle_bytes_2103 = ff_shuffle_bytes_2103_mmxext; >> - } >> if (EXTERNAL_SSE2(cpu_flags)) { >> + shuffle_bytes_2103 = ff_shuffle_bytes_2103_sse2; >> + shuffle_bytes_0321 = ff_shuffle_bytes_0321_sse2; >> + shuffle_bytes_1230 = ff_shuffle_bytes_1230_sse2; >> + shuffle_bytes_3012 = ff_shuffle_bytes_3012_sse2; >> + shuffle_bytes_3210 = ff_shuffle_bytes_3210_sse2; >> #if ARCH_X86_64 >> uyvytoyuv422 = ff_uyvytoyuv422_sse2; >> #endif >> diff --git a/libswscale/x86/rgb_2_rgb.asm b/libswscale/x86/rgb_2_rgb.asm >> index 0bf1278718..9fc1974389 100644 >> --- a/libswscale/x86/rgb_2_rgb.asm >> +++ b/libswscale/x86/rgb_2_rgb.asm >> @@ -25,7 +25,6 @@ >> >> SECTION_RODATA >> >> -pb_mask_shuffle2103_mmx times 8 dw 255 >> pb_shuffle2103: db 2, 1, 0, 3, 6, 5, 4, 7, 10, 9, 8, 11, 14, 13, 12, 15 >> pb_shuffle0321: db 0, 3, 2, 1, 4, 7, 6, 5, 8, 11, 10, 9, 12, 15, 14, 13 >> pb_shuffle1230: db 1, 2, 3, 0, 5, 6, 7, 4, 9, 10, 11, 8, 13, 14, 15, 12 >> @@ -50,11 +49,50 @@ SECTION .text >> ;------------------------------------------------------------------------------ >> ; shuffle_bytes_2103_mmext (const uint8_t *src, uint8_t *dst, int src_size) >> ;------------------------------------------------------------------------------ >> -INIT_MMX mmxext >> -cglobal shuffle_bytes_2103, 3, 5, 8, src, dst, w, tmp, x >> - mova m6, [pb_mask_shuffle2103_mmx] >> - mova m7, m6 >> - psllq m7, 8 >> + >> +%macro SHUFFLE2103_SSE2 0 >> + pshuflw m1, m0, 0xb1 >> + pshufhw m1, m1, 0xb1 >> + >> + pand m0, m3 >> + pand m1, m2 >> +%endmacro >> + >> +%macro SHUFFLE0321_SSE2 0 >> + pshuflw m1, m0, 0xb1 >> + pshufhw m1, m1, 0xb1 >> + >> + pand m0, m2 >> + pand m1, m3 >> +%endmacro >> + >> +%macro SHUFFLE1230_SSE2 0 >> + pslld m1, m0, 24 >> + psrld m0, 8 >> +%endmacro >> + >> +%macro SHUFFLE3012_SSE2 0 >> + pslld m1, m0, 8 >> + psrld m0, 24 >> +%endmacro >> + >> +%macro SHUFFLE3210_SSE2 0 >> + pshuflw m1, m0, 0xb1 >> + pshufhw m1, m1, 0xb1 >> + >> + psrlw m0, m1, 8 >> + psllw m1, 8 >> +%endmacro >> + >> +; %1-4 index shuffle >> +; %5 load mask >> +%macro SHUFFLE_BYTES_SSE2 5 >> +cglobal shuffle_bytes_%1%2%3%4, 3, 5, 4, src, dst, w, tmp, x >> +%if %5 >> + pcmpeqw m2, m2 >> + psllw m3, m2, 8 ; (word) { 0xff00 } x4 >> + psrlw m2, 8 ; (word) { 0x00ff } x4 >> +%endif >> >> movsxdifnidn wq, wd >> mov xq, wq >> @@ -68,13 +106,13 @@ cglobal shuffle_bytes_2103, 3, 5, 8, src, dst, w, tmp, x >> je .loop_simd >> >> .loop_scalar: >> - mov tmpb, [srcq + wq + 2] >> + mov tmpb, [srcq + wq + %1] >> mov [dstq+wq + 0], tmpb >> - mov tmpb, [srcq + wq + 1] >> + mov tmpb, [srcq + wq + %2] >> mov [dstq+wq + 1], tmpb >> - mov tmpb, [srcq + wq + 0] >> + mov tmpb, [srcq + wq + %3] >> mov [dstq+wq + 2], tmpb >> - mov tmpb, [srcq + wq + 3] >> + mov tmpb, [srcq + wq + %4] >> mov [dstq+wq + 3], tmpb >> add wq, 4 >> sub xq, 4 >> @@ -86,29 +124,26 @@ jge .end >> >> .loop_simd: >> movu m0, [srcq+wq] >> - movu m1, [srcq+wq+8] >> - >> - pshufw m3, m0, 177 >> - pshufw m5, m1, 177 >> - >> - pand m0, m7 >> - pand m3, m6 >> >> - pand m1, m7 >> - pand m5, m6 >> + SHUFFLE%1%2%3%4_SSE2 >> >> - por m0, m3 >> - por m1, m5 >> + por m0, m1 >> >> movu [dstq+wq], m0 >> - movu [dstq+wq + 8], m1 >> >> - add wq, mmsize*2 >> + add wq, mmsize >> jl .loop_simd >> >> .end: >> - emms >> RET >> +%endmacro >> + >> +INIT_XMM sse2 >> +SHUFFLE_BYTES_SSE2 2, 1, 0, 3, 1 >> +SHUFFLE_BYTES_SSE2 0, 3, 2, 1, 1 >> +SHUFFLE_BYTES_SSE2 1, 2, 3, 0, 0 >> +SHUFFLE_BYTES_SSE2 3, 0, 1, 2, 0 >> +SHUFFLE_BYTES_SSE2 3, 2, 1, 0, 0 >> >> ;------------------------------------------------------------------------------ >> ; shuffle_bytes_## (const uint8_t *src, uint8_t *dst, int src_size) > > How old are the youngest processors with SSE2, but without SSSE3? AMD Phenom/K10. > According to Wikipedia, nearly 15 years. Which makes me believe that the > SSE2 versions are not worth it (how many of these CPUs will use a new > FFmpeg anyway?). Simply by using the latest version of a video player that uses ffmpeg is enough to be able to run the newest code. It was easy to write and i don't feel particularly interested enough to argue, so if you think it's not worth adding, i can just remove the mmxext version and skip adding anything. _______________________________________________ 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".