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 938A04A9E2 for ; Thu, 6 Jun 2024 14:15:21 +0000 (UTC) Received: from [127.0.1.1] (localhost [127.0.0.1]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id 6FE0C68D661; Thu, 6 Jun 2024 17:15:18 +0300 (EEST) Received: from mail-pl1-f180.google.com (mail-pl1-f180.google.com [209.85.214.180]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTPS id B277468A62B for ; Thu, 6 Jun 2024 17:15:12 +0300 (EEST) Received: by mail-pl1-f180.google.com with SMTP id d9443c01a7336-1f44b5b9de6so9118645ad.3 for ; Thu, 06 Jun 2024 07:15:12 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20230601; t=1717683310; x=1718288110; darn=ffmpeg.org; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:date:subject:to:from:from:to:cc:subject:date:message-id :reply-to; bh=VESMBXyiitYUu1dRiY25XPwHIHXQVGvI3NEuuEZeGXA=; b=LR1U4b3tBV3mVLlnj6CSzhksKhoAjilmkxR9eudklkKDeNBkWhkfo/rfQP1y6R7dsk 25dSEMcKrzarc9GDD4pF1G8PqEDGYFY7OHQSVX2DNx8RRU4nvSB4w8XxRhaWrYxLBMuD BHouwt+qvH+Sl0ascM9mB/a5y2Ri1JHE6DIl82mOl1d29A+qEl8x4hgy/ob0UUxGAPxl 0A9Rs74IL5+PUubG8J0YxXoXWSz1RRoDviKtVyHZ2fWXKJ3zUCECqL6bOqfPGNJYkPBQ 3s6AdeW3rOJXLo6fn/gl34YUG15wyznSNXDP8zbJKlZWk4zvGJ5Rold715DaXUjXZxgm g9Qg== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1717683310; x=1718288110; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:date:subject:to:from:x-gm-message-state:from:to:cc :subject:date:message-id:reply-to; bh=VESMBXyiitYUu1dRiY25XPwHIHXQVGvI3NEuuEZeGXA=; b=uvmB7Z4XypFVZ53dKvjy13d/TGSnlmrZh7OCTyOvPZAAhxZ7whDukNxsyjxEhhbWlI aPDyGtqVu7wufcy3SehJr/fG/d1sCyTxC/n67nzXpNP9JIG2fCfUt9l6AFawCxf+VgwJ U+SFrUaxlcfPptZiWGtyihK8SUHEwZjjLAmH42NRqmD7UW5OUm2/Z1QyQztpVVpNubO/ dxH/no4DGt5KENHxK788PnTxQDv20GDie/cgsC6k+H4UDVRYxBkdr57M2HctE7YwqIj8 6dD0yDOtI+G7tYXd0wQn/UI62RYmsZ15tdrEEGJy1zJwlCQgwMR6yoNhcx/VOxec0ZmR X4oQ== X-Gm-Message-State: AOJu0Yw9eNRPvlvfb1cSXkJBfSCYJenw8v6N1z/SW4/m4GM7S0NPWI3l JUPldqnzoVbHjHG+wBlidPdqRpG7jnu8eY82XVkZ+3VKPHmS8klGOlsU5Q== X-Google-Smtp-Source: AGHT+IFTGCFpG4tTCJ07nlWe8icrOG9UCao6ivPzA6uAv3UyDq8JDne49NUwdhJO7AeO58HeoFHbYw== X-Received: by 2002:a17:902:fb85:b0:1f3:903:5c9a with SMTP id d9443c01a7336-1f6a5a83651mr51941205ad.58.1717683309839; Thu, 06 Jun 2024 07:15:09 -0700 (PDT) Received: from localhost.localdomain ([190.194.167.233]) by smtp.gmail.com with ESMTPSA id d9443c01a7336-1f6bd7ccee0sm15368925ad.148.2024.06.06.07.15.08 for (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Thu, 06 Jun 2024 07:15:09 -0700 (PDT) From: James Almer To: ffmpeg-devel@ffmpeg.org Date: Thu, 6 Jun 2024 11:15:05 -0300 Message-ID: <20240606141505.132-1-jamrial@gmail.com> X-Mailer: git-send-email 2.45.1 In-Reply-To: <20240605205116.3258-1-jamrial@gmail.com> References: <20240605205116.3258-1-jamrial@gmail.com> MIME-Version: 1.0 Subject: [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-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: 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) -- 2.45.1 _______________________________________________ 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".