DSUtil now use new VirtualDub libraries (SSE2 deinterlacing for MPEG2 decoder)

AudioSwitcher rare memory corruption fixed

git-svn-id: https://mpc-hc.svn.sourceforge.net/svnroot/mpc-hc/trunk@1907 10f7b99b-c216-0410-bff0-8a66a9350fd8

6 files changed, 777 insertions, 1170 deletions

diff --git a/src/DSUtil/deinterlace.cpp b/src/DSUtil/deinterlace.cpp
new file mode 100644
index 000000000..a66915dfd
--- /dev/null
+++ b/src/DSUtil/deinterlace.cpp
@@ -0,0 +1,526 @@
+//	VirtualDub - Video processing and capture application

+//	Copyright (C) 1998-2001 Avery Lee

+//

+//	This program is free software; you can redistribute it and/or modify

+//	it under the terms of the GNU General Public License as published by

+//	the Free Software Foundation; either version 2 of the License, or

+//	(at your option) any later version.

+//

+//	This program is distributed in the hope that it will be useful,

+//	but WITHOUT ANY WARRANTY; without even the implied warranty of

+//	MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the

+//	GNU General Public License for more details.

+//

+//	You should have received a copy of the GNU General Public License

+//	along with this program; if not, write to the Free Software

+//	Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.

+

+#include "stdafx.h"

+#include <emmintrin.h>

+#include <vd2/system/cpuaccel.h>

+

+#define uint8	unsigned char

+#define uint32	unsigned int

+#define uint64	unsigned __int64

+

+#ifdef _M_IX86

+#define VD_CPU_X86

+#endif

+

+#ifdef _M_X64

+#define VD_CPU_AMD64

+#endif

+

+///////////////////////////////////////////////////////////////////////////

+

+#pragma warning(disable: 4799)		// warning C4799: function has no EMMS instruction

+

+///////////////////////////////////////////////////////////////////////////

+

+#ifdef _M_IX86

+static void __declspec(naked) asm_blend_row_clipped(void *dst, const void *src, uint32 w, ptrdiff_t srcpitch) {

+	__asm {

+		push	ebp

+		push	edi

+		push	esi

+		push	ebx

+

+		mov		edi,[esp+20]

+		mov		esi,[esp+24]

+		sub		edi,esi

+		mov		ebp,[esp+28]

+		mov		edx,[esp+32]

+

+xloop:

+		mov		ecx,[esi]

+		mov		eax,0fefefefeh

+

+		mov		ebx,[esi+edx]

+		and		eax,ecx

+

+		shr		eax,1

+		and		ebx,0fefefefeh

+

+		shr		ebx,1

+		add		esi,4

+

+		add		eax,ebx

+		dec		ebp

+

+		mov		[edi+esi-4],eax

+		jnz		xloop

+

+		pop		ebx

+		pop		esi

+		pop		edi

+		pop		ebp

+		ret

+	};

+}

+

+static void __declspec(naked) asm_blend_row(void *dst, const void *src, uint32 w, ptrdiff_t srcpitch) {

+	__asm {

+		push	ebp

+		push	edi

+		push	esi

+		push	ebx

+

+		mov		edi,[esp+20]

+		mov		esi,[esp+24]

+		sub		edi,esi

+		mov		ebp,[esp+28]

+		mov		edx,[esp+32]

+

+xloop:

+		mov		ecx,[esi]

+		mov		eax,0fcfcfcfch

+

+		mov		ebx,[esi+edx]

+		and		eax,ecx

+

+		shr		ebx,1

+		mov		ecx,[esi+edx*2]

+

+		shr		ecx,2

+		and		ebx,07f7f7f7fh

+

+		shr		eax,2

+		and		ecx,03f3f3f3fh

+

+		add		eax,ebx

+		add		esi,4

+

+		add		eax,ecx

+		dec		ebp

+

+		mov		[edi+esi-4],eax

+		jnz		xloop

+

+		pop		ebx

+		pop		esi

+		pop		edi

+		pop		ebp

+		ret

+	};

+}

+

+static void __declspec(naked) asm_blend_row_MMX(void *dst, const void *src, uint32 w, ptrdiff_t srcpitch) {

+	static const __declspec(align(8)) __int64 mask0 = 0xfcfcfcfcfcfcfcfci64;

+	static const __declspec(align(8)) __int64 mask1 = 0x7f7f7f7f7f7f7f7fi64;

+	static const __declspec(align(8)) __int64 mask2 = 0x3f3f3f3f3f3f3f3fi64;

+	__asm {

+		push	ebp

+		push	edi

+		push	esi

+		push	ebx

+

+		mov		edi,[esp+20]

+		mov		esi,[esp+24]

+		sub		edi,esi

+		mov		ebp,[esp+28]

+		mov		edx,[esp+32]

+

+		movq	mm5,mask0

+		movq	mm6,mask1

+		movq	mm7,mask2

+		inc		ebp

+		shr		ebp,1

+xloop:

+		movq	mm2,[esi]

+		movq	mm0,mm5

+

+		movq	mm1,[esi+edx]

+		pand	mm0,mm2

+

+		psrlq	mm1,1

+		movq	mm2,[esi+edx*2]

+

+		psrlq	mm2,2

+		pand	mm1,mm6

+

+		psrlq	mm0,2

+		pand	mm2,mm7

+

+		paddb	mm0,mm1

+		add		esi,8

+

+		paddb	mm0,mm2

+		dec		ebp

+

+		movq	[edi+esi-8],mm0

+		jne		xloop

+

+		pop		ebx

+		pop		esi

+		pop		edi

+		pop		ebp

+		ret

+	};

+}

+

+static void __declspec(naked) asm_blend_row_ISSE(void *dst, const void *src, uint32 w, ptrdiff_t srcpitch) {

+	__asm {

+		push	ebp

+		push	edi

+		push	esi

+		push	ebx

+

+		mov		edi,[esp+20]

+		mov		esi,[esp+24]

+		sub		edi,esi

+		mov		ebp,[esp+28]

+		mov		edx,[esp+32]

+

+		inc		ebp

+		shr		ebp,1

+		pcmpeqb	mm7, mm7

+

+		align	16

+xloop:

+		movq	mm0, [esi]

+		movq	mm2, mm7

+		pxor	mm0, mm7

+

+		pxor	mm2, [esi+edx*2]

+		pavgb	mm0, mm2

+		pxor	mm0, mm7

+

+		pavgb	mm0, [esi+edx]

+		add		esi,8

+

+		movq	[edi+esi-8],mm0

+		dec		ebp

+		jne		xloop

+

+		pop		ebx

+		pop		esi

+		pop		edi

+		pop		ebp

+		ret

+	};

+}

+#else

+static void asm_blend_row_clipped(void *dst0, const void *src0, uint32 w, ptrdiff_t srcpitch) {

+	uint32 *dst = (uint32 *)dst0;

+	const uint32 *src = (const uint32 *)src0;

+	const uint32 *src2 = (const uint32 *)((const char *)src + srcpitch);

+

+	do {

+		const uint32 x = *src++;

+		const uint32 y = *src2++;

+

+		*dst++ = (x|y) - (((x^y)&0xfefefefe)>>1);

+	} while(--w);

+}

+

+static void asm_blend_row(void *dst0, const void *src0, uint32 w, ptrdiff_t srcpitch) {

+	uint32 *dst = (uint32 *)dst0;

+	const uint32 *src = (const uint32 *)src0;

+	const uint32 *src2 = (const uint32 *)((const char *)src + srcpitch);

+	const uint32 *src3 = (const uint32 *)((const char *)src2 + srcpitch);

+

+	do {

+		const uint32 a = *src++;

+		const uint32 b = *src2++;

+		const uint32 c = *src3++;

+		const uint32 hi = (a & 0xfcfcfc) + 2*(b & 0xfcfcfc) + (c & 0xfcfcfc);

+		const uint32 lo = (a & 0x030303) + 2*(b & 0x030303) + (c & 0x030303) + 0x020202;

+

+		*dst++ = (hi + (lo & 0x0c0c0c))>>2;

+	} while(--w);

+}

+#endif

+

+#if defined(VD_CPU_X86) || defined(VD_CPU_AMD64)

+	static void asm_blend_row_SSE2(void *dst, const void *src, uint32 w, ptrdiff_t srcpitch) {

+		__m128i zero = _mm_setzero_si128();

+		__m128i inv = _mm_cmpeq_epi8(zero, zero);

+

+		w = (w + 3) >> 2;

+

+		const __m128i *src1 = (const __m128i *)src;

+		const __m128i *src2 = (const __m128i *)((const char *)src + srcpitch);

+		const __m128i *src3 = (const __m128i *)((const char *)src + srcpitch*2);

+		__m128i *dstrow = (__m128i *)dst;

+		do {

+			__m128i a = *src1++;

+			__m128i b = *src2++;

+			__m128i c = *src3++;

+

+			*dstrow++ = _mm_avg_epu8(_mm_xor_si128(_mm_avg_epu8(_mm_xor_si128(a, inv), _mm_xor_si128(c, inv)), inv), b);

+		} while(--w);

+	}

+

+#endif

+

+namespace {

+

+	void Average_scalar(void *dst, ptrdiff_t dstPitch, const void *src1, const void *src2, ptrdiff_t srcPitch, uint32 w16, uint32 h) {

+		uint32 w4 = w16 << 2;

+		do {

+			uint32 *dstv = (uint32 *)dst;

+			uint32 *src1v = (uint32 *)src1;

+			uint32 *src2v = (uint32 *)src2;

+

+			for(uint32 i=0; i<w4; ++i) {

+				uint32 a = src1v[i];

+				uint32 b = src2v[i];

+

+				dstv[i] = (a|b) - (((a^b) & 0xfefefefe) >> 1);

+			}

+

+			dst = (char *)dst + dstPitch;

+			src1 = (char *)src1 + srcPitch;

+			src2 = (char *)src2 + srcPitch;

+		} while(--h);

+	}

+

+#if defined(VD_CPU_X86)

+	void __declspec(naked) __cdecl Average_MMX(void *dst, ptrdiff_t dstPitch, const void *src1, const void *src2, ptrdiff_t srcPitch, uint32 w16, uint32 h) {

+		static const __declspec(align(8)) uint64 x7fb = 0x7f7f7f7f7f7f7f7f;

+		static const __declspec(align(8)) uint64 xfeb = 0xfefefefefefefefe;

+

+		__asm {

+			push		ebp

+			push		edi

+			push		esi

+			push		ebx

+

+			mov			esi, [esp+24+16]

+			mov			eax, [esp+4+16]

+			shl			esi, 4

+			mov			ecx, [esp+12+16]

+			mov			edx, [esp+16+16]

+			mov			ebp, [esp+20+16]

+			mov			edi, [esp+8+16]

+			sub			edi, esi

+			sub			ebp, esi

+

+			movq		mm6, x7fb

+			movq		mm7, xfeb

+

+			mov			esi, [esp+28+16]

+yloop:

+			mov			ebx, [esp+24+16]

+mainRowLoop:

+			movq		mm0, [ecx]

+			movq		mm3, [ecx + 8]

+			movq		mm1, mm0

+			movq		mm2, [edx]

+			movq		mm4, mm3

+			movq		mm5, [edx + 8]

+			por			mm1, mm2

+			pxor		mm0, mm2

+			por			mm4, mm5

+			pxor		mm3, mm5

+			psrlq		mm0, 1

+			pand		mm3, mm7

+			pand		mm0, mm6

+			psrlq		mm3, 1

+			psubb		mm1, mm0

+			psubb		mm4, mm3

+			add			ecx, 16

+			movq		[eax], mm1

+			movq		[eax+8], mm4

+			add			edx, 16

+			add			eax, 16

+			dec			ebx

+			jne			mainRowLoop

+

+			add			eax, edi

+			add			ecx, ebp

+			add			edx, ebp

+			dec			esi

+			jne			yloop

+

+			emms

+			pop			ebx

+			pop			esi

+			pop			edi

+			pop			ebp

+			ret	

+		}

+	}

+

+	void __declspec(naked) __cdecl Average_ISSE(void *dst, ptrdiff_t dstPitch, const void *src1, const void *src2, ptrdiff_t srcPitch, uint32 w16, uint32 h) {

+		static const __declspec(align(8)) uint64 x7fb = 0x7f7f7f7f7f7f7f7f;

+		static const __declspec(align(8)) uint64 xfeb = 0xfefefefefefefefe;

+

+		__asm {

+			push		ebp

+			push		edi

+			push		esi

+			push		ebx

+

+			mov			esi, [esp+24+16]

+			mov			eax, [esp+4+16]

+			shl			esi, 4

+			mov			ecx, [esp+12+16]

+			mov			edx, [esp+16+16]

+			mov			ebp, [esp+20+16]

+			mov			edi, [esp+8+16]

+			sub			edi, esi

+			sub			ebp, esi

+

+			movq		mm6, x7fb

+			movq		mm7, xfeb

+

+			mov			esi, [esp+28+16]

+yloop:

+			mov			ebx, [esp+24+16]

+mainRowLoop:

+			movq		mm0, [ecx]

+			movq		mm1, [ecx + 8]

+			movq		mm2, [edx]

+			movq		mm3, [edx + 8]

+			pavgb		mm0, mm2

+			pavgb		mm1, mm3

+			movq		[eax], mm0

+			add			ecx, 16

+			add			edx, 16

+			movq		[eax+8], mm1

+			add			eax, 16

+			dec			ebx

+			jne			mainRowLoop

+

+			add			eax, edi

+			add			ecx, ebp

+			add			edx, ebp

+			dec			esi

+			jne			yloop

+

+			emms

+			pop			ebx

+			pop			esi

+			pop			edi

+			pop			ebp

+			ret	

+		}

+	}

+#endif

+

+#if defined(VD_CPU_X86) || defined(VD_CPU_AMD64)

+	void Average_SSE2(void *dst, ptrdiff_t dstPitch, const void *src1, const void *src2, ptrdiff_t srcPitch, uint32 w16, uint32 h) {

+		do {

+			__m128i *dstv = (__m128i *)dst;

+			__m128i *src1v = (__m128i *)src1;

+			__m128i *src2v = (__m128i *)src2;

+

+			for(uint32 i=0; i<w16; ++i)

+				dstv[i] = _mm_avg_epu8(src1v[i], src2v[i]);

+

+			dst = (char *)dst + dstPitch;

+			src1 = (char *)src1 + srcPitch;

+			src2 = (char *)src2 + srcPitch;

+		} while(--h);

+	}

+#endif

+

+	void InterpPlane_Bob(void *dst, ptrdiff_t dstpitch, const void *src, ptrdiff_t srcpitch, uint32 w, uint32 h, bool interpField2) {

+		void (*blend_func)(void *dst, ptrdiff_t dstPitch, const void *src1, const void *src2, ptrdiff_t srcPitch, uint32 w16, uint32 h);

+#if defined(VD_CPU_X86)

+		if (SSE2_enabled)

+			blend_func = Average_SSE2;

+		else if (ISSE_enabled)

+			blend_func = Average_ISSE;

+		else if (MMX_enabled)

+			blend_func = Average_MMX;

+		else

+			blend_func = Average_scalar;

+#else

+		blend_func = Average_SSE2;

+#endif

+

+		w = (w + 3) >> 2;

+

+		int y0 = interpField2 ? 1 : 2;

+

+		if (!interpField2)

+			memcpy(dst, src, w * 4);

+

+		if (h > y0) {

+			ASSERT(((UINT_PTR)dst & 0xF) == 0);

+			ASSERT((dstpitch & 0xF) == 0);

+			ASSERT(((UINT_PTR)src & 0xF) == 0);

+			ASSERT((srcpitch*(y0 - 1) & 0xF) == 0);

+			blend_func((char *)dst + dstpitch*y0,

+				dstpitch*2,

+				(const char *)src + srcpitch*(y0 - 1),

+				(const char *)src + srcpitch*(y0 + 1),

+				srcpitch*2,

+				(w + 3) >> 2,

+				(h - y0) >> 1);

+		}

+

+		if (interpField2)

+			memcpy((char *)dst + dstpitch*(h - 1), (const char *)src + srcpitch*(h - 1), w*4);

+

+#ifdef _M_IX86

+		if (MMX_enabled)

+			__asm emms

+#endif

+	}

+

+	void BlendPlane(void *dst, ptrdiff_t dstpitch, const void *src, ptrdiff_t srcpitch, uint32 w, uint32 h) {

+		void (*blend_func)(void *, const void *, uint32, ptrdiff_t);

+#if defined(VD_CPU_X86)

+		if (SSE2_enabled)

+			blend_func = asm_blend_row_SSE2;

+		else

+			blend_func = ISSE_enabled ? asm_blend_row_ISSE : MMX_enabled ? asm_blend_row_MMX : asm_blend_row;

+#else

+		blend_func = asm_blend_row_SSE2;

+#endif

+

+		w = (w + 3) >> 2;

+

+		asm_blend_row_clipped(dst, src, w, srcpitch);

+		if (h-=2)

+			do {

+				dst = ((char *)dst + dstpitch);

+

+				blend_func(dst, src, w, srcpitch);

+

+				src = ((char *)src + srcpitch);

+			} while(--h);

+

+		asm_blend_row_clipped((char *)dst + dstpitch, src, w, srcpitch);

+

+#ifdef _M_IX86

+		if (MMX_enabled)

+			__asm emms

+#endif

+	}

+}

+

+void DeinterlaceBlend(BYTE* dst, BYTE* src, DWORD w, DWORD h, DWORD dstpitch, DWORD srcpitch)

+{

+	BlendPlane(dst, dstpitch, src, srcpitch, w, h);

+}

+

+void DeinterlaceBob(BYTE* dst, BYTE* src, DWORD w, DWORD h, DWORD dstpitch, DWORD srcpitch, bool topfield)

+{

+	topfield = !topfield;

+

+	InterpPlane_Bob(dst, dstpitch, src, srcpitch, w, h, topfield);

+}

diff --git a/src/DSUtil/dsutil.vcproj b/src/DSUtil/dsutil.vcproj
index 49162423b..69a625c18 100644
--- a/src/DSUtil/dsutil.vcproj
+++ b/src/DSUtil/dsutil.vcproj
@@ -44,7 +44,7 @@
 			<Tool

 				Name="VCCLCompilerTool"

 				AdditionalOptions="/MP"

-				AdditionalIncludeDirectories="..\..\include;..\filters\BaseClasses;&quot;$(DXSDK_DIR)Include&quot;"

+				AdditionalIncludeDirectories="..\..\include;..\filters\BaseClasses;&quot;$(DXSDK_DIR)Include&quot;;..\thirdparty\VirtualDub\h"

 				PreprocessorDefinitions="WIN32;_DEBUG;_LIB;"

 			/>

 			<Tool

@@ -105,7 +105,7 @@
 			<Tool

 				Name="VCCLCompilerTool"

 				AdditionalOptions="/MP"

-				AdditionalIncludeDirectories="..\..\include;..\filters\BaseClasses;&quot;$(DXSDK_DIR)Include&quot;"

+				AdditionalIncludeDirectories="..\..\include;..\filters\BaseClasses;&quot;$(DXSDK_DIR)Include&quot;;..\thirdparty\VirtualDub\h"

 				PreprocessorDefinitions="_WIN64;_DEBUG;_LIB;"

 				Detect64BitPortabilityProblems="false"

 				DebugInformationFormat="3"

@@ -167,7 +167,7 @@
 			<Tool

 				Name="VCCLCompilerTool"

 				AdditionalOptions="/MP"

-				AdditionalIncludeDirectories="..\..\include;..\filters\BaseClasses;&quot;$(DXSDK_DIR)Include&quot;"

+				AdditionalIncludeDirectories="..\..\include;..\filters\BaseClasses;&quot;$(DXSDK_DIR)Include&quot;;..\thirdparty\VirtualDub\h"

 				PreprocessorDefinitions="WIN32;NDEBUG;_LIB"

 				BufferSecurityCheck="true"

 				EnableEnhancedInstructionSet="1"

@@ -233,7 +233,7 @@
 			<Tool

 				Name="VCCLCompilerTool"

 				AdditionalOptions="/MP"

-				AdditionalIncludeDirectories="..\..\include;..\filters\BaseClasses;&quot;$(DXSDK_DIR)Include&quot;"

+				AdditionalIncludeDirectories="..\..\include;..\filters\BaseClasses;&quot;$(DXSDK_DIR)Include&quot;;..\thirdparty\VirtualDub\h"

 				PreprocessorDefinitions="_WIN64;NDEBUG;_LIB"

 				BufferSecurityCheck="true"

 				EnableEnhancedInstructionSet="0"

@@ -278,130 +278,8 @@
 			Filter="cpp;c;cxx;def;odl;idl;hpj;bat;asm"

 			>

 			<File

-				RelativePath=".\a_yuv2rgb.asm"

+				RelativePath=".\deinterlace.cpp"

 				>

-				<FileConfiguration

-					Name="Debug Unicode|Win32"

-					>

-					<Tool

-						Name="VCCustomBuildTool"

-						CommandLine="yasm -X vc -g cv8 -f &quot;$(PlatformName)&quot; -o &quot;$(OutDir)\$(InputName).obj&quot; &quot;$(InputPath)&quot;&#x0D;&#x0A;"

-						Outputs="$(OutDir)\$(InputName).obj"

-					/>

-				</FileConfiguration>

-				<FileConfiguration

-					Name="Debug Unicode|x64"

-					ExcludedFromBuild="true"

-					>

-					<Tool

-						Name="VCCustomBuildTool"

-						CommandLine="ml64 /c /coff /Cx /nologo /Fo&quot;$(OutDir)\$(InputName).obj&quot; &quot;$(InputPath)&#x0D;&#x0A;"

-						Outputs="$(OutDir)\$(InputName).obj"

-					/>

-				</FileConfiguration>

-				<FileConfiguration

-					Name="Release Unicode|Win32"

-					>

-					<Tool

-						Name="VCCustomBuildTool"

-						CommandLine="yasm -X vc -g cv8 -f &quot;$(PlatformName)&quot; -o &quot;$(OutDir)\$(InputName).obj&quot; &quot;$(InputPath)&quot;&#x0D;&#x0A;"

-						Outputs="$(OutDir)\$(InputName).obj"

-					/>

-				</FileConfiguration>

-				<FileConfiguration

-					Name="Release Unicode|x64"

-					ExcludedFromBuild="true"

-					>

-					<Tool

-						Name="VCCustomBuildTool"

-						CommandLine="yasm -X vc -g cv8 -f &quot;$(PlatformName)&quot; -o &quot;$(OutDir)\$(InputName).obj&quot; &quot;$(InputPath)&quot;&#x0D;&#x0A;"

-						Outputs="$(OutDir)\$(InputName).obj"

-					/>

-				</FileConfiguration>

-			</File>

-			<File

-				RelativePath=".\a_yuvtable.asm"

-				>

-				<FileConfiguration

-					Name="Debug Unicode|Win32"

-					>

-					<Tool

-						Name="VCCustomBuildTool"

-						CommandLine="yasm -X vc -g cv8 -f &quot;$(PlatformName)&quot; -o &quot;$(OutDir)\$(InputName).obj&quot; &quot;$(InputPath)&quot;&#x0D;&#x0A;"

-						Outputs="$(OutDir)\$(InputName).obj"

-					/>

-				</FileConfiguration>

-				<FileConfiguration

-					Name="Debug Unicode|x64"

-					ExcludedFromBuild="true"

-					>

-					<Tool

-						Name="VCCustomBuildTool"

-						CommandLine="ml64 /c /coff /Cx /nologo /Fo&quot;$(OutDir)\$(InputName).obj&quot; &quot;$(InputPath)&#x0D;&#x0A;"

-						Outputs="$(OutDir)\$(InputName).obj"

-					/>

-				</FileConfiguration>

-				<FileConfiguration

-					Name="Release Unicode|Win32"

-					>

-					<Tool

-						Name="VCCustomBuildTool"

-						CommandLine="yasm -X vc -g cv8 -f &quot;$(PlatformName)&quot; -o &quot;$(OutDir)\$(InputName).obj&quot; &quot;$(InputPath)&quot;&#x0D;&#x0A;"

-						Outputs="$(OutDir)\$(InputName).obj"

-					/>

-				</FileConfiguration>

-				<FileConfiguration

-					Name="Release Unicode|x64"

-					ExcludedFromBuild="true"

-					>

-					<Tool

-						Name="VCCustomBuildTool"

-						CommandLine="ml64 /c /coff /Cx /nologo /Fo&quot;$(OutDir)\$(InputName).obj&quot; &quot;$(InputPath)&#x0D;&#x0A;"

-						Outputs="$(OutDir)\$(InputName).obj"

-					/>

-				</FileConfiguration>

-			</File>

-			<File

-				RelativePath=".\convert_a.asm"

-				>

-				<FileConfiguration

-					Name="Debug Unicode|Win32"

-					>

-					<Tool

-						Name="VCCustomBuildTool"

-						CommandLine="ml /c /coff /Cx /nologo /Fo&quot;$(OutDir)\$(InputName).obj&quot; &quot;$(InputPath)&quot;&#x0D;&#x0A;"

-						Outputs="$(OutDir)\$(InputName).obj"

-					/>

-				</FileConfiguration>

-				<FileConfiguration

-					Name="Debug Unicode|x64"

-					ExcludedFromBuild="true"

-					>

-					<Tool

-						Name="VCCustomBuildTool"

-						CommandLine="ml64 /c /coff /Cx /nologo /Fo&quot;$(OutDir)\$(InputName).obj&quot; &quot;$(InputPath)&quot;&#x0D;&#x0A;"

-						Outputs="$(OutDir)\$(InputName).obj"

-					/>

-				</FileConfiguration>

-				<FileConfiguration

-					Name="Release Unicode|Win32"

-					>

-					<Tool

-						Name="VCCustomBuildTool"

-						CommandLine="ml /c /coff /Cx /nologo /Fo&quot;$(OutDir)\$(InputName).obj&quot; &quot;$(InputPath)&quot;&#x0D;&#x0A;"

-						Outputs="$(OutDir)\$(InputName).obj"

-					/>

-				</FileConfiguration>

-				<FileConfiguration

-					Name="Release Unicode|x64"

-					ExcludedFromBuild="true"

-					>

-					<Tool

-						Name="VCCustomBuildTool"

-						CommandLine="ml64 /c /coff /Cx /nologo /Fo&quot;$(OutDir)\$(InputName).obj&quot; &quot;$(InputPath)&quot;&#x0D;&#x0A;"

-						Outputs="$(OutDir)\$(InputName).obj"

-					/>

-				</FileConfiguration>

 			</File>

 			<File

 				RelativePath=".\DSMPropertyBag.cpp"

diff --git a/src/DSUtil/vd.cpp b/src/DSUtil/vd.cpp
index 063fbfe06..0d7f77aaa 100644
--- a/src/DSUtil/vd.cpp
+++ b/src/DSUtil/vd.cpp
@@ -1,5 +1,6 @@
 //	VirtualDub - Video processing and capture application

-//	Copyright (C) 1998-2001 Avery Lee

+//	Graphics support library

+//	Copyright (C) 1998-2007 Avery Lee

 //

 //	This program is free software; you can redistribute it and/or modify

 //	it under the terms of the GNU General Public License as published by

@@ -16,8 +17,7 @@
 //	Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.

 //

 //  Notes: 

-//  - BitBltFromI420ToRGB is from VirtualDub

-//	- The core assembly function of CCpuID is from DVD2AVI

+//  - VDPixmapBlt is from VirtualDub

 //  - sse2 yv12 to yuy2 conversion by Haali

 //	(- vd.cpp/h should be renamed to something more sensible already :)

 

@@ -27,300 +27,272 @@
 #include "vd_asm.h"

 #include <intrin.h>

 

+#include <vd2/system/cpuaccel.h>

+#include <vd2/system/memory.h>

+

+#include <vd2/Kasumi/pixmap.h>

+#include <vd2/Kasumi/pixmaputils.h>

+#include <vd2/Kasumi/pixmapops.h>

+

 #pragma warning(disable : 4799) // no emms... blahblahblah

 

+void VDCPUTest() {

+	SYSTEM_INFO si;

+

+	long lEnableFlags = CPUCheckForExtensions();

+

+	GetSystemInfo(&si);

+

+	if (si.wProcessorArchitecture == PROCESSOR_ARCHITECTURE_INTEL)

+		if (si.wProcessorLevel < 4)

+			lEnableFlags &= ~CPUF_SUPPORTS_FPU;		// Not strictly true, but very slow anyway

+

+	// Enable FPU support...

+

+	CPUEnableExtensions(lEnableFlags);

+

+	VDFastMemcpyAutodetect();

+}

+

 CCpuID g_cpuid;

 

 CCpuID::CCpuID()

-{       

-	int CPUInfo[4] = {-1};

-	__cpuid(CPUInfo, 1);

-	int t = CPUInfo[3];

-

-	int mflags = 0;

-	mflags |= ((t&0x00800000)!=0) ? mmx : 0;			// STD MMX

-	mflags |= ((t&0x02000000)!=0) ? ssemmx+ssefpu : 0;	// STD SSE

-	mflags |= ((t&0x04000000)!=0) ? sse2 : 0;			// SSE2

+{

+	VDCPUTest();

 

-	t = CPUInfo[2];

-	mflags |= ((t&0x00000001)!=0) ? sse3 : 0;			// SSE3

+	long lEnableFlags = CPUGetEnabledExtensions();

 

-	// 3dnow

-	__cpuid(CPUInfo, 0x80000001);

-	t = CPUInfo[3];

-	mflags |= ((t&0x80000000)!=0) ? _3dnow : 0;			// 3D NOW

-	mflags |= ((t&0x00400000)!=0) ? ssemmx : 0;			// SSE MMX

+	int flags = 0;

+	flags |= !!(lEnableFlags & CPUF_SUPPORTS_MMX)			? mmx		: 0;			// STD MMX

+	flags |= !!(lEnableFlags & CPUF_SUPPORTS_INTEGER_SSE)	? ssemmx	: 0;			// SSE MMX

+	flags |= !!(lEnableFlags & CPUF_SUPPORTS_SSE)			? ssefpu	: 0;			// STD SSE

+	flags |= !!(lEnableFlags & CPUF_SUPPORTS_SSE2)			? sse2		: 0;			// SSE2

+	flags |= !!(lEnableFlags & CPUF_SUPPORTS_3DNOW)			? _3dnow	: 0;			// 3DNow

 

 	// result

-	m_flags = (flag_t)mflags;

+	m_flags = (flag_t)flags;

 }

 

-void memcpy_accel(void* dst, const void* src, size_t len)

+bool BitBltFromI420ToI420(int w, int h, BYTE* dsty, BYTE* dstu, BYTE* dstv, int dstpitch, BYTE* srcy, BYTE* srcu, BYTE* srcv, int srcpitch)

 {

-#ifndef _WIN64

-	if((g_cpuid.m_flags & CCpuID::ssefpu) && len >= 128 

-		&& !((DWORD)src&15) && !((DWORD)dst&15))

-	{

-		__asm

-		{

-			mov     esi, dword ptr [src]

-			mov     edi, dword ptr [dst]

-			mov     ecx, len

-			shr     ecx, 7

-	memcpy_accel_sse_loop:

-			prefetchnta	[esi+16*8]

-			movaps		xmm0, [esi]

-			movaps		xmm1, [esi+16*1]

-			movaps		xmm2, [esi+16*2]

-			movaps		xmm3, [esi+16*3]

-			movaps		xmm4, [esi+16*4]

-			movaps		xmm5, [esi+16*5]

-			movaps		xmm6, [esi+16*6]

-			movaps		xmm7, [esi+16*7]

-			movntps		[edi], xmm0

-			movntps		[edi+16*1], xmm1

-			movntps		[edi+16*2], xmm2

-			movntps		[edi+16*3], xmm3

-			movntps		[edi+16*4], xmm4

-			movntps		[edi+16*5], xmm5

-			movntps		[edi+16*6], xmm6

-			movntps		[edi+16*7], xmm7

-			add			esi, 128

-			add			edi, 128

-			dec			ecx

-			jne			memcpy_accel_sse_loop

-			mov     ecx, len

-			and     ecx, 127

-			cmp     ecx, 0

-			je		memcpy_accel_sse_end

-	memcpy_accel_sse_loop2:

-			mov		dl, byte ptr[esi] 

-			mov		byte ptr[edi], dl

-			inc		esi

-			inc		edi

-			dec		ecx

-			jne		memcpy_accel_sse_loop2

-	memcpy_accel_sse_end:

-			emms

-			sfence

-		}

-	}

-	else if((g_cpuid.m_flags & CCpuID::mmx) && len >= 64

-		&& !((DWORD)src&7) && !((DWORD)dst&7))

-	{

-		__asm 

-		{

-			mov     esi, dword ptr [src]

-			mov     edi, dword ptr [dst]

-			mov     ecx, len

-			shr     ecx, 6

-	memcpy_accel_mmx_loop:

-			movq    mm0, qword ptr [esi]

-			movq    mm1, qword ptr [esi+8*1]

-			movq    mm2, qword ptr [esi+8*2]

-			movq    mm3, qword ptr [esi+8*3]

-			movq    mm4, qword ptr [esi+8*4]

-			movq    mm5, qword ptr [esi+8*5]

-			movq    mm6, qword ptr [esi+8*6]

-			movq    mm7, qword ptr [esi+8*7]

-			movq    qword ptr [edi], mm0

-			movq    qword ptr [edi+8*1], mm1

-			movq    qword ptr [edi+8*2], mm2

-			movq    qword ptr [edi+8*3], mm3

-			movq    qword ptr [edi+8*4], mm4

-			movq    qword ptr [edi+8*5], mm5

-			movq    qword ptr [edi+8*6], mm6

-			movq    qword ptr [edi+8*7], mm7

-			add     esi, 64

-			add     edi, 64

-			loop	memcpy_accel_mmx_loop

-			mov     ecx, len

-			and     ecx, 63

-			cmp     ecx, 0

-			je		memcpy_accel_mmx_end

-	memcpy_accel_mmx_loop2:

-			mov		dl, byte ptr [esi] 

-			mov		byte ptr [edi], dl

-			inc		esi

-			inc		edi

-			dec		ecx

-			jne		memcpy_accel_mmx_loop2

-	memcpy_accel_mmx_end:

-			emms

-		}

-	}

-	else

-#endif

-	{

-		memcpy(dst, src, len);

-	}

+	VDPixmap srcbm = {0};

+

+	srcbm.data		= srcy;

+	srcbm.pitch		= srcpitch;

+	srcbm.w			= w;

+	srcbm.h			= h;

+	srcbm.format	= nsVDPixmap::kPixFormat_YUV420_Planar;

+	srcbm.data2		= srcu;

+	srcbm.pitch2	= srcpitch / 2;

+	srcbm.data3		= srcv;

+	srcbm.pitch3	= srcpitch / 2;

+

+	VDPixmap dstpxm = {0};

+

+	dstpxm.data		= dsty;

+	dstpxm.pitch	= dstpitch;

+	dstpxm.w		= w;

+	dstpxm.h		= h;

+	dstpxm.format	= nsVDPixmap::kPixFormat_YUV420_Planar;

+	dstpxm.data2	= dstu;

+	dstpxm.pitch2	= dstpitch / 2;

+	dstpxm.data3	= dstv;

+	dstpxm.pitch3	= dstpitch / 2;

+

+	return VDPixmapBlt(dstpxm, srcbm);

 }

 

-static void yuvtoyuy2row_c(BYTE* dst, BYTE* srcy, BYTE* srcu, BYTE* srcv, DWORD width)

+bool BitBltFromYUY2ToYUY2(int w, int h, BYTE* dst, int dstpitch, BYTE* src, int srcpitch)

 {

-	WORD* dstw = (WORD*)dst;

-	for(; width > 1; width -= 2)

-	{

-		*dstw++ = (*srcu++<<8)|*srcy++;

-		*dstw++ = (*srcv++<<8)|*srcy++;

-	}

+	VDPixmap srcbm = {0};

+

+	srcbm.data		= src;

+	srcbm.pitch		= srcpitch;

+	srcbm.w			= w;

+	srcbm.h			= h;

+	srcbm.format	= nsVDPixmap::kPixFormat_YUV422_YUYV;

+

+	VDPixmap dstpxm = {

+		dst,

+		NULL,

+		w,

+		h,

+		dstpitch

+	};

+

+	dstpxm.format = nsVDPixmap::kPixFormat_YUV422_YUYV;

+

+	return VDPixmapBlt(dstpxm, srcbm);

 }

 

-static void yuvtoyuy2row_avg_c(BYTE* dst, BYTE* srcy, BYTE* srcu, BYTE* srcv, DWORD width, DWORD pitchuv)

+bool BitBltFromI420ToRGB(int w, int h, BYTE* dst, int dstpitch, int dbpp, BYTE* srcy, BYTE* srcu, BYTE* srcv, int srcpitch)

 {

-	WORD* dstw = (WORD*)dst;

-	for(; width > 1; width -= 2, srcu++, srcv++)

-	{

-		*dstw++ = (((srcu[0]+srcu[pitchuv])>>1)<<8)|*srcy++;

-		*dstw++ = (((srcv[0]+srcv[pitchuv])>>1)<<8)|*srcy++;

-	}

+	VDPixmap srcbm = {0};

+

+	srcbm.data		= srcy;

+	srcbm.pitch		= srcpitch;

+	srcbm.w			= w;

+	srcbm.h			= h;

+	srcbm.format	= nsVDPixmap::kPixFormat_YUV420_Planar;

+	srcbm.data2		= srcu;

+	srcbm.pitch2	= srcpitch/2;

+	srcbm.data3		= srcv;

+	srcbm.pitch3	= srcpitch/2;

+

+	VDPixmap dstpxm = {

+		(char *)dst + dstpitch * (h - 1),

+		NULL,

+		w,

+		h,

+		-dstpitch

+	};

+

+	switch(dbpp) {

+	case 16:	dstpxm.format = nsVDPixmap::kPixFormat_XRGB1555; break;

+	case 24:	dstpxm.format = nsVDPixmap::kPixFormat_RGB888; break;

+	case 32:	dstpxm.format = nsVDPixmap::kPixFormat_XRGB8888; break;

+	default:

+		VDASSERT(false);

+	}

+

+	// TODO: check correct conversion work (555->565) when dpp == 16

+

+	return VDPixmapBlt(dstpxm, srcbm);

 }

 

-static void asm_blend_row_clipped_c(BYTE* dst, BYTE* src, DWORD w, DWORD srcpitch)

+bool BitBltFromI420ToYUY2(int w, int h, BYTE* dst, int dstpitch, BYTE* srcy, BYTE* srcu, BYTE* srcv, int srcpitch)

 {

-	BYTE* src2 = src + srcpitch;

-	do 

+	if(srcpitch == 0) srcpitch = w;

+

+#ifndef _WIN64

+	if((g_cpuid.m_flags & CCpuID::sse2)

+		&& !((DWORD_PTR)srcy&15) && !((DWORD_PTR)srcu&15) && !((DWORD_PTR)srcv&15) && !(srcpitch&31) 

+		&& !((DWORD_PTR)dst&15) && !(dstpitch&15))

 	{

-		*dst++ = (*src++ + *src2++ + 1) >> 1;

-	} while(w--);

+		if(w<=0 || h<=0 || (w&1) || (h&1))

+			return(false);

+

+		yv12_yuy2_sse2(srcy, srcu, srcv, srcpitch/2, w/2, h, dst, dstpitch);

+		return(true);

+	}

+#endif

+

+	VDPixmap srcbm = {0};

+

+	srcbm.data		= srcy;

+	srcbm.pitch		= srcpitch;

+	srcbm.w			= w;

+	srcbm.h			= h;

+	srcbm.format	= nsVDPixmap::kPixFormat_YUV420_Planar;

+	srcbm.data2		= srcu;

+	srcbm.pitch2	= srcpitch/2;

+	srcbm.data3		= srcv;

+	srcbm.pitch3	= srcpitch/2;

+

+	VDPixmap dstpxm = {

+		dst,

+		NULL,

+		w,

+		h,

+		dstpitch

+	};

+

+	dstpxm.format = nsVDPixmap::kPixFormat_YUV422_YUYV;

+

+	return VDPixmapBlt(dstpxm, srcbm);

 }

 

-static void asm_blend_row_c(BYTE* dst, BYTE* src, DWORD w, DWORD srcpitch)

+bool BitBltFromRGBToRGB(int w, int h, BYTE* dst, int dstpitch, int dbpp, BYTE* src, int srcpitch, int sbpp)

 {

-	BYTE* src2 = src + srcpitch;

-	BYTE* src3 = src2 + srcpitch;

-	do

-	{

-		*dst++ = (*src++ + (*src2++ << 1) + *src3++ + 2) >> 2;

-	} while(w--);

+	VDPixmap srcbm = {

+		(char *)src + srcpitch * (h - 1),

+		NULL,

+		w,

+		h,

+		-srcpitch

+	};

+

+	switch(dbpp) {

+	case 8:		srcbm.format = nsVDPixmap::kPixFormat_Pal8; break;

+	case 16:	srcbm.format = nsVDPixmap::kPixFormat_XRGB1555; break;

+	case 24:	srcbm.format = nsVDPixmap::kPixFormat_RGB888; break;

+	case 32:	srcbm.format = nsVDPixmap::kPixFormat_XRGB8888; break;

+	default:

+		VDASSERT(false);

+	}

+

+	VDPixmap dstpxm = {

+		(char *)dst + dstpitch * (h - 1),

+		NULL,

+		w,

+		h,

+		-dstpitch

+	};

+

+	switch(dbpp) {

+	case 8:		dstpxm.format = nsVDPixmap::kPixFormat_Pal8; break;

+	case 16:	dstpxm.format = nsVDPixmap::kPixFormat_XRGB1555; break;

+	case 24:	dstpxm.format = nsVDPixmap::kPixFormat_RGB888; break;

+	case 32:	dstpxm.format = nsVDPixmap::kPixFormat_XRGB8888; break;

+	default:

+		VDASSERT(false);

+	}

+

+	return VDPixmapBlt(dstpxm, srcbm);

 }

 

-bool BitBltFromI420ToI420(int w, int h, BYTE* dsty, BYTE* dstu, BYTE* dstv, int dstpitch, BYTE* srcy, BYTE* srcu, BYTE* srcv, int srcpitch)

+bool BitBltFromYUY2ToRGB(int w, int h, BYTE* dst, int dstpitch, int dbpp, BYTE* src, int srcpitch)

 {

-	if((w&1))

-		return(false);

-

-	if(w > 0 && w == srcpitch && w == dstpitch)

-	{

-		memcpy_accel(dsty, srcy, h*srcpitch);

-		memcpy_accel(dstu, srcu, h/2*srcpitch/2);

-		memcpy_accel(dstv, srcv, h/2*srcpitch/2);

-	}

-	else

-	{

-		int pitch = min(abs(srcpitch), abs(dstpitch));

-

-		for(ptrdiff_t y = 0; y < h; y++, srcy += srcpitch, dsty += dstpitch)

-			memcpy_accel(dsty, srcy, pitch);

+	if(srcpitch == 0) srcpitch = w;

 

-		srcpitch >>= 1;

-		dstpitch >>= 1;

+	VDPixmap srcbm = {0};

 

-		pitch = min(abs(srcpitch), abs(dstpitch));

+	srcbm.data		= src;

+	srcbm.pitch		= srcpitch;

+	srcbm.w			= w;

+	srcbm.h			= h;

+	srcbm.format	= nsVDPixmap::kPixFormat_YUV422_YUYV;

 

-		for(ptrdiff_t y = 0; y < h; y+=2, srcu += srcpitch, dstu += dstpitch)

-			memcpy_accel(dstu, srcu, pitch);

+	VDPixmap dstpxm = {

+		(char *)dst + dstpitch * (h - 1),

+		NULL,

+		w,

+		h,

+		-dstpitch

+	};

 

-		for(ptrdiff_t y = 0; y < h; y+=2, srcv += srcpitch, dstv += dstpitch)

-			memcpy_accel(dstv, srcv, pitch);

+	switch(dbpp) {

+	case 16:	dstpxm.format = nsVDPixmap::kPixFormat_XRGB1555; break;

+	case 24:	dstpxm.format = nsVDPixmap::kPixFormat_RGB888; break;

+	case 32:	dstpxm.format = nsVDPixmap::kPixFormat_XRGB8888; break;

+	default:

+		VDASSERT(false);

 	}

 

-	return(true);

+	return VDPixmapBlt(dstpxm, srcbm);

 }

 

-bool BitBltFromYUY2ToYUY2(int w, int h, BYTE* dst, int dstpitch, BYTE* src, int srcpitch)

+static void yuvtoyuy2row_c(BYTE* dst, BYTE* srcy, BYTE* srcu, BYTE* srcv, DWORD width)

 {

-	if(w > 0 && w == srcpitch && w == dstpitch)

-	{

-		memcpy_accel(dst, src, h*srcpitch);

-	}

-	else

+	WORD* dstw = (WORD*)dst;

+	for(; width > 1; width -= 2)

 	{

-		int pitch = min(abs(srcpitch), abs(dstpitch));

-

-		for(ptrdiff_t y = 0; y < h; y++, src += srcpitch, dst += dstpitch)

-			memcpy_accel(dst, src, pitch);

+		*dstw++ = (*srcu++<<8)|*srcy++;

+		*dstw++ = (*srcv++<<8)|*srcy++;

 	}

-

-	return(true);

 }

 

-#ifndef _WIN64

-extern "C" void asm_YUVtoRGB32_row(void* ARGB1, void* ARGB2, BYTE* Y1, BYTE* Y2, BYTE* U, BYTE* V, long width);

-extern "C" void asm_YUVtoRGB24_row(void* ARGB1, void* ARGB2, BYTE* Y1, BYTE* Y2, BYTE* U, BYTE* V, long width);

-extern "C" void asm_YUVtoRGB16_row(void* ARGB1, void* ARGB2, BYTE* Y1, BYTE* Y2, BYTE* U, BYTE* V, long width);

-extern "C" void asm_YUVtoRGB32_row_MMX(void* ARGB1, void* ARGB2, BYTE* Y1, BYTE* Y2, BYTE* U, BYTE* V, long width);

-extern "C" void asm_YUVtoRGB24_row_MMX(void* ARGB1, void* ARGB2, BYTE* Y1, BYTE* Y2, BYTE* U, BYTE* V, long width);

-extern "C" void asm_YUVtoRGB16_row_MMX(void* ARGB1, void* ARGB2, BYTE* Y1, BYTE* Y2, BYTE* U, BYTE* V, long width);

-extern "C" void asm_YUVtoRGB32_row_ISSE(void* ARGB1, void* ARGB2, BYTE* Y1, BYTE* Y2, BYTE* U, BYTE* V, long width);

-extern "C" void asm_YUVtoRGB24_row_ISSE(void* ARGB1, void* ARGB2, BYTE* Y1, BYTE* Y2, BYTE* U, BYTE* V, long width);

-extern "C" void asm_YUVtoRGB16_row_ISSE(void* ARGB1, void* ARGB2, BYTE* Y1, BYTE* Y2, BYTE* U, BYTE* V, long width);

-#endif

-

-bool BitBltFromI420ToRGB(int w, int h, BYTE* dst, int dstpitch, int dbpp, BYTE* srcy, BYTE* srcu, BYTE* srcv, int srcpitch)

+static void yuvtoyuy2row_avg_c(BYTE* dst, BYTE* srcy, BYTE* srcu, BYTE* srcv, DWORD width, DWORD pitchuv)

 {

-	if(w<=0 || h<=0 || (w&1) || (h&1))

-		return(false);

-

-#ifndef _WIN64

-	void (*asm_YUVtoRGB_row)(void* ARGB1, void* ARGB2, BYTE* Y1, BYTE* Y2, BYTE* U, BYTE* V, long width) = NULL;;

-

-	if((g_cpuid.m_flags & CCpuID::ssefpu) && !(w&7))

-	{

-		switch(dbpp)

-		{

-		case 16: asm_YUVtoRGB_row = asm_YUVtoRGB16_row/*_ISSE*/; break; // TODO: fix _ISSE (555->565)

-		case 24: asm_YUVtoRGB_row = asm_YUVtoRGB24_row_ISSE; break;

-		case 32: asm_YUVtoRGB_row = asm_YUVtoRGB32_row_ISSE; break;

-		}

-	}

-	else if((g_cpuid.m_flags & CCpuID::mmx) && !(w&7))

-	{

-		switch(dbpp)

-		{

-		case 16: asm_YUVtoRGB_row = asm_YUVtoRGB16_row/*_MMX*/; break; // TODO: fix _MMX (555->565)

-		case 24: asm_YUVtoRGB_row = asm_YUVtoRGB24_row_MMX; break;

-		case 32: asm_YUVtoRGB_row = asm_YUVtoRGB32_row_MMX; break;

-		}

-	}

-	else

-	{

-		switch(dbpp)

-		{

-		case 16: asm_YUVtoRGB_row = asm_YUVtoRGB16_row; break;

-		case 24: asm_YUVtoRGB_row = asm_YUVtoRGB24_row; break;

-		case 32: asm_YUVtoRGB_row = asm_YUVtoRGB32_row; break;

-		}

-	}

-

-	if(!asm_YUVtoRGB_row) 

-		return(false);

-

-	do

+	WORD* dstw = (WORD*)dst;

+	for(; width > 1; width -= 2, srcu++, srcv++)

 	{

-		asm_YUVtoRGB_row(dst + dstpitch, dst, srcy + srcpitch, srcy, srcu, srcv, w/2);

-

-		dst += 2*dstpitch;

-		srcy += srcpitch*2;

-		srcu += srcpitch/2;

-		srcv += srcpitch/2;

+		*dstw++ = (((srcu[0]+srcu[pitchuv])>>1)<<8)|*srcy++;

+		*dstw++ = (((srcv[0]+srcv[pitchuv])>>1)<<8)|*srcy++;

 	}

-	while(h -= 2);

-

-	if(g_cpuid.m_flags & CCpuID::mmx)

-		__asm emms

-

-	if(g_cpuid.m_flags & CCpuID::ssefpu)

-		__asm sfence

-

-	return(true);

-#else

-	ASSERT(FALSE);

-	return(false);

-#endif

 }

 

-bool BitBltFromI420ToYUY2(int w, int h, BYTE* dst, int dstpitch, BYTE* srcy, BYTE* srcu, BYTE* srcv, int srcpitch, bool fInterlaced)

+bool BitBltFromI420ToYUY2Interlaced(int w, int h, BYTE* dst, int dstpitch, BYTE* srcy, BYTE* srcu, BYTE* srcv, int srcpitch)

 {

 	if(w<=0 || h<=0 || (w&1) || (h&1))

 		return(false);

@@ -332,16 +304,15 @@ bool BitBltFromI420ToYUY2(int w, int h, BYTE* dst, int dstpitch, BYTE* srcy, BYT
 

 #ifndef _WIN64

 	if((g_cpuid.m_flags & CCpuID::sse2) 

-        && !((DWORD_PTR)srcy&15) && !((DWORD_PTR)srcu&15) && !((DWORD_PTR)srcv&15) && !(srcpitch&31) 

-        && !((DWORD_PTR)dst&15) && !(dstpitch&15))

+		&& !((DWORD_PTR)srcy&15) && !((DWORD_PTR)srcu&15) && !((DWORD_PTR)srcv&15) && !(srcpitch&31) 

+		&& !((DWORD_PTR)dst&15) && !(dstpitch&15))

 	{

-		if(!fInterlaced) yv12_yuy2_sse2(srcy, srcu, srcv, srcpitch/2, w/2, h, dst, dstpitch);

-		else yv12_yuy2_sse2_interlaced(srcy, srcu, srcv, srcpitch/2, w/2, h, dst, dstpitch);

+		yv12_yuy2_sse2_interlaced(srcy, srcu, srcv, srcpitch/2, w/2, h, dst, dstpitch);

 		return(true);

 	}

 	else

 	{

-		ASSERT(!fInterlaced);

+		ASSERT(FALSE);

 	}

 

 	if((g_cpuid.m_flags & CCpuID::mmx) && !(w&7))

@@ -359,15 +330,16 @@ bool BitBltFromI420ToYUY2(int w, int h, BYTE* dst, int dstpitch, BYTE* srcy, BYT
 	if(!yuvtoyuy2row) 

 		return(false);

 

+	int halfsrcpitch = srcpitch/2;

 	do

 	{

 		yuvtoyuy2row(dst, srcy, srcu, srcv, w);

-		yuvtoyuy2row_avg(dst + dstpitch, srcy + srcpitch, srcu, srcv, w, srcpitch/2);

+		yuvtoyuy2row_avg(dst + dstpitch, srcy + srcpitch, srcu, srcv, w, halfsrcpitch);

 

 		dst += 2*dstpitch;

-		srcy += srcpitch*2;

-		srcu += srcpitch/2;

-		srcv += srcpitch/2;

+		srcy += halfsrcpitch;

+		srcu += halfsrcpitch;

+		srcv += halfsrcpitch;

 	}

 	while((h -= 2) > 2);

 

@@ -381,481 +353,3 @@ bool BitBltFromI420ToYUY2(int w, int h, BYTE* dst, int dstpitch, BYTE* srcy, BYT
 

 	return(true);

 }

-

-bool BitBltFromRGBToRGB(int w, int h, BYTE* dst, int dstpitch, int dbpp, BYTE* src, int srcpitch, int sbpp)

-{

-	if(dbpp == sbpp)

-	{

-		int rowbytes = w*dbpp>>3;

-

-		if(rowbytes > 0 && rowbytes == srcpitch && rowbytes == dstpitch)

-		{

-			memcpy_accel(dst, src, h*rowbytes);

-		}

-		else

-		{

-			for(ptrdiff_t y = 0; y < h; y++, src += srcpitch, dst += dstpitch)

-				memcpy_accel(dst, src, rowbytes);

-		}

-

-		return(true);

-	}

-	

-	if(sbpp != 16 && sbpp != 24 && sbpp != 32

-	|| dbpp != 16 && dbpp != 24 && dbpp != 32)

-		return(false);

-

-	if(dbpp == 16)

-	{

-		for(ptrdiff_t y = 0; y < h; y++, src += srcpitch, dst += dstpitch)

-		{

-			if(sbpp == 24)

-			{

-				BYTE* s = (BYTE*)src;

-				WORD* d = (WORD*)dst;

-				for(ptrdiff_t x = 0; x < w; x++, s+=3, d++)

-					*d = (WORD)(((*((DWORD*)s)>>8)&0xf800)|((*((DWORD*)s)>>5)&0x07e0)|((*((DWORD*)s)>>3)&0x1f));

-			}

-			else if(sbpp == 32)

-			{

-				DWORD* s = (DWORD*)src;

-				WORD* d = (WORD*)dst;

-				for(ptrdiff_t x = 0; x < w; x++, s++, d++)

-					*d = (WORD)(((*s>>8)&0xf800)|((*s>>5)&0x07e0)|((*s>>3)&0x1f));

-			}

-		}

-	}

-	else if(dbpp == 24)

-	{

-		for(ptrdiff_t y = 0; y < h; y++, src += srcpitch, dst += dstpitch)

-		{

-			if(sbpp == 16)

-			{

-				WORD* s = (WORD*)src;

-				BYTE* d = (BYTE*)dst;

-				for(ptrdiff_t x = 0; x < w; x++, s++, d+=3)

-				{	// not tested, r-g-b might be in reverse

-					d[0] = (*s&0x001f)<<3;

-					d[1] = (*s&0x07e0)<<5;

-					d[2] = (*s&0xf800)<<8;

-				}

-			}

-			else if(sbpp == 32)

-			{

-				BYTE* s = (BYTE*)src;

-				BYTE* d = (BYTE*)dst;

-				for(ptrdiff_t x = 0; x < w; x++, s+=4, d+=3)

-					{d[0] = s[0]; d[1] = s[1]; d[2] = s[2];}

-			}

-		}

-	}

-	else if(dbpp == 32)

-	{

-		for(ptrdiff_t y = 0; y < h; y++, src += srcpitch, dst += dstpitch)

-		{

-			if(sbpp == 16)

-			{

-				WORD* s = (WORD*)src;

-				DWORD* d = (DWORD*)dst;

-				for(ptrdiff_t x = 0; x < w; x++, s++, d++)

-					*d = ((*s&0xf800)<<8)|((*s&0x07e0)<<5)|((*s&0x001f)<<3);

-			}

-			else if(sbpp == 24)

-			{	

-				BYTE* s = (BYTE*)src;

-				DWORD* d = (DWORD*)dst;

-				for(ptrdiff_t x = 0; x < w; x++, s+=3, d++)

-					*d = *((DWORD*)s)&0xffffff;

-			}

-		}

-	}

-

-	return(true);

-}

-

-void DeinterlaceBlend(BYTE* dst, BYTE* src, DWORD rowbytes, DWORD h, DWORD dstpitch, DWORD srcpitch)

-{

-	void (*blend_row_clipped)(BYTE* dst, BYTE* src, DWORD w, DWORD srcpitch) = NULL;

-	void (*blend_row)(BYTE* dst, BYTE* src, DWORD w, DWORD srcpitch) = NULL;

-

-#ifndef _WIN64

-	if((g_cpuid.m_flags & CCpuID::sse2) && !((DWORD)src&0xf) && !((DWORD)dst&0xf) && !(srcpitch&0xf))

-	{

-		blend_row_clipped = asm_blend_row_clipped_SSE2;

-		blend_row = asm_blend_row_SSE2;

-	}

-	else if(g_cpuid.m_flags & CCpuID::mmx)

-	{

-		blend_row_clipped = asm_blend_row_clipped_MMX;

-		blend_row = asm_blend_row_MMX;

-	}

-	else

-#endif

-	{

-		blend_row_clipped = asm_blend_row_clipped_c;

-		blend_row = asm_blend_row_c;

-	}

-

-	if(!blend_row_clipped)

-		return;

-

-	blend_row_clipped(dst, src, rowbytes, srcpitch);

-

-	if((h -= 2) > 0) do

-	{

-		dst += dstpitch;

-		blend_row(dst, src, rowbytes, srcpitch);

-        src += srcpitch;

-	}

-	while(--h);

-

-	blend_row_clipped(dst + dstpitch, src, rowbytes, srcpitch);

-

-#ifndef _WIN64

-	if(g_cpuid.m_flags & CCpuID::mmx)

-		__asm emms

-#endif

-}

-

-void DeinterlaceBob(BYTE* dst, BYTE* src, DWORD rowbytes, DWORD h, DWORD dstpitch, DWORD srcpitch, bool topfield)

-{

-	if(topfield)

-	{

-		BitBltFromRGBToRGB(rowbytes, h/2, dst, dstpitch*2, 8, src, srcpitch*2, 8);

-		AvgLines8(dst, h, dstpitch);

-	}

-	else

-	{

-		BitBltFromRGBToRGB(rowbytes, h/2, dst + dstpitch, dstpitch*2, 8, src + srcpitch, srcpitch*2, 8);

-		AvgLines8(dst + dstpitch, h-1, dstpitch);

-	}

-}

-

-void AvgLines8(BYTE* dst, DWORD h, DWORD pitch)

-{

-	if(h <= 1)

-		return;

-

-	BYTE* s = dst;

-	BYTE* d = dst + (h-2)*pitch;

-

-	for(; s < d; s += pitch*2)

-	{

-		BYTE* tmp = s;

-

-#ifndef _WIN64

-		if((g_cpuid.m_flags & CCpuID::sse2) && !((DWORD)tmp&0xf) && !((DWORD)pitch&0xf))

-		{

-			__asm

-			{

-				mov		esi, tmp

-				mov		ebx, pitch

-

-				mov		ecx, ebx

-				shr		ecx, 4

-

-AvgLines8_sse2_loop:

-				movdqa	xmm0, [esi]

-				pavgb	xmm0, [esi+ebx*2]

-				movdqa	[esi+ebx], xmm0

-				add		esi, 16

-

-				dec		ecx

-				jnz		AvgLines8_sse2_loop

-

-				mov		tmp, esi

-			}

-

-			for(ptrdiff_t i = pitch&7; i--; tmp++)

-			{

-				tmp[pitch] = (tmp[0] + tmp[pitch<<1] + 1) >> 1;

-			}

-		}

-		else if(g_cpuid.m_flags & CCpuID::mmx)

-		{

-			__asm

-			{

-				mov		esi, tmp

-				mov		ebx, pitch

-

-				mov		ecx, ebx

-				shr		ecx, 3

-

-				pxor	mm7, mm7

-AvgLines8_mmx_loop:

-				movq	mm0, [esi]

-				movq	mm1, mm0

-

-				punpcklbw	mm0, mm7

-				punpckhbw	mm1, mm7

-

-				movq	mm2, [esi+ebx*2]

-				movq	mm3, mm2

-

-				punpcklbw	mm2, mm7

-				punpckhbw	mm3, mm7

-

-				paddw	mm0, mm2

-				psrlw	mm0, 1

-

-				paddw	mm1, mm3

-				psrlw	mm1, 1

-

-				packuswb	mm0, mm1

-

-				movq	[esi+ebx], mm0

-

-				lea		esi, [esi+8]

-

-				dec		ecx

-				jnz		AvgLines8_mmx_loop

-

-				mov		tmp, esi

-			}

-

-			for(ptrdiff_t i = pitch&7; i--; tmp++)

-			{

-				tmp[pitch] = (tmp[0] + tmp[pitch<<1] + 1) >> 1;

-			}

-		}

-		else

-#endif

-		{

-			for(ptrdiff_t i = pitch; i--; tmp++)

-			{

-				tmp[pitch] = (tmp[0] + tmp[pitch<<1] + 1) >> 1;

-			}

-		}

-	}

-

-	if(!(h&1) && h >= 2)

-	{

-		dst += (h-2)*pitch;

-		memcpy_accel(dst + pitch, dst, pitch);

-	}

-

-#ifndef _WIN64

-	__asm emms;

-#endif

-}

-

-void AvgLines555(BYTE* dst, DWORD h, DWORD pitch)

-{

-	if(h <= 1)

-		return;

-

-	unsigned __int64 __0x03e003e003e003e0 = 0x03e003e003e003e0;

-	unsigned __int64 __0x001f001f001f001f = 0x001f001f001f001f;

-

-	BYTE* s = dst;

-	BYTE* d = dst + (h-2)*pitch;

-

-	for(; s < d; s += pitch*2)

-	{

-		BYTE* tmp = s;

-

-#ifndef _WIN64

-		__asm

-		{

-			mov		esi, tmp

-			mov		ebx, pitch

-

-			mov		ecx, ebx

-			shr		ecx, 3

-

-			movq	mm6, __0x03e003e003e003e0

-			movq	mm7, __0x001f001f001f001f

-

-AvgLines555_loop:

-			movq	mm0, [esi]

-			movq	mm1, mm0

-			movq	mm2, mm0

-

-			psrlw	mm0, 10				// red1 bits: mm0 = 001f001f001f001f

-			pand	mm1, mm6			// green1 bits: mm1 = 03e003e003e003e0

-			pand	mm2, mm7			// blue1 bits: mm2 = 001f001f001f001f

-

-			movq	mm3, [esi+ebx*2]

-			movq	mm4, mm3

-			movq	mm5, mm3

-

-			psrlw	mm3, 10				// red2 bits: mm3 = 001f001f001f001f

-			pand	mm4, mm6			// green2 bits: mm4 = 03e003e003e003e0

-			pand	mm5, mm7			// blue2 bits: mm5 = 001f001f001f001f

-

-			paddw	mm0, mm3

-			psrlw	mm0, 1				// (red1+red2)/2

-			psllw	mm0, 10				// red bits at 7c007c007c007c00

-

-			paddw	mm1, mm4

-			psrlw	mm1, 1				// (green1+green2)/2

-			pand	mm1, mm6			// green bits at 03e003e003e003e0

-

-			paddw	mm2, mm5

-			psrlw	mm2, 1				// (blue1+blue2)/2

-										// blue bits at 001f001f001f001f (no need to pand, lower bits were discareded)

-

-			por		mm0, mm1

-			por		mm0, mm2

-

-			movq	[esi+ebx], mm0

-

-			lea		esi, [esi+8]

-

-			dec		ecx

-			jnz		AvgLines555_loop

-

-			mov		tmp, esi

-		}

-#endif

-

-		for(ptrdiff_t i = (pitch&7)>>1; i--; tmp++)

-		{

-			tmp[pitch] = 

-				((((*tmp&0x7c00) + (tmp[pitch<<1]&0x7c00)) >> 1)&0x7c00)|

-				((((*tmp&0x03e0) + (tmp[pitch<<1]&0x03e0)) >> 1)&0x03e0)|

-				((((*tmp&0x001f) + (tmp[pitch<<1]&0x001f)) >> 1)&0x001f);

-		}

-	}

-

-	if(!(h&1) && h >= 2)

-	{

-		dst += (h-2)*pitch;

-		memcpy_accel(dst + pitch, dst, pitch);

-	}

-

-#ifndef _WIN64

-	__asm emms;

-#endif

-}

-

-void AvgLines565(BYTE* dst, DWORD h, DWORD pitch)

-{

-	if(h <= 1)

-		return;

-

-	unsigned __int64 __0x07e007e007e007e0 = 0x07e007e007e007e0;

-	unsigned __int64 __0x001f001f001f001f = 0x001f001f001f001f;

-

-	BYTE* s = dst;

-	BYTE* d = dst + (h-2)*pitch;

-

-	for(; s < d; s += pitch*2)

-	{

-		WORD* tmp = (WORD*)s;

-

-#ifndef _WIN64

-		__asm

-		{

-			mov		esi, tmp

-			mov		ebx, pitch

-

-			mov		ecx, ebx

-			shr		ecx, 3

-

-			movq	mm6, __0x07e007e007e007e0

-			movq	mm7, __0x001f001f001f001f

-

-AvgLines565_loop:

-			movq	mm0, [esi]

-			movq	mm1, mm0

-			movq	mm2, mm0

-

-			psrlw	mm0, 11				// red1 bits: mm0 = 001f001f001f001f

-			pand	mm1, mm6			// green1 bits: mm1 = 07e007e007e007e0

-			pand	mm2, mm7			// blue1 bits: mm2 = 001f001f001f001f

-

-			movq	mm3, [esi+ebx*2]

-			movq	mm4, mm3

-			movq	mm5, mm3

-

-			psrlw	mm3, 11				// red2 bits: mm3 = 001f001f001f001f

-			pand	mm4, mm6			// green2 bits: mm4 = 07e007e007e007e0

-			pand	mm5, mm7			// blue2 bits: mm5 = 001f001f001f001f

-

-			paddw	mm0, mm3

-			psrlw	mm0, 1				// (red1+red2)/2

-			psllw	mm0, 11				// red bits at f800f800f800f800

-

-			paddw	mm1, mm4

-			psrlw	mm1, 1				// (green1+green2)/2

-			pand	mm1, mm6			// green bits at 03e003e003e003e0

-

-			paddw	mm2, mm5

-			psrlw	mm2, 1				// (blue1+blue2)/2

-										// blue bits at 001f001f001f001f (no need to pand, lower bits were discareded)

-

-			por		mm0, mm1

-			por		mm0, mm2

-

-			movq	[esi+ebx], mm0

-

-			lea		esi, [esi+8]

-

-			dec		ecx

-			jnz		AvgLines565_loop

-

-			mov		tmp, esi

-		}

-#else

-		for(ptrdiff_t wd=(pitch>>3);wd--;tmp++)

-		{

-			tmp[0] = 

-				((((*tmp&0xf800) + (tmp[pitch<<1]&0xf800)) >> 1)&0xf800)|

-				((((*tmp&0x07e0) + (tmp[pitch<<1]&0x07e0)) >> 1)&0x07e0)|

-				((((*tmp&0x001f) + (tmp[pitch<<1]&0x001f)) >> 1)&0x001f);

-		}

-#endif

-

-		for(ptrdiff_t i = (pitch&7)>>1; i--; tmp++)

-		{

-			tmp[pitch] = 

-				((((*tmp&0xf800) + (tmp[pitch<<1]&0xf800)) >> 1)&0xf800)|

-				((((*tmp&0x07e0) + (tmp[pitch<<1]&0x07e0)) >> 1)&0x07e0)|

-				((((*tmp&0x001f) + (tmp[pitch<<1]&0x001f)) >> 1)&0x001f);

-		}

-	}

-

-	if(!(h&1) && h >= 2)

-	{

-		dst += (h-2)*pitch;

-		memcpy_accel(dst + pitch, dst, pitch);

-	}

-

-#ifndef _WIN64

-	__asm emms;

-#endif

-}

-

-#ifndef _WIN64

-extern "C" void mmx_YUY2toRGB24(const BYTE* src, BYTE* dst, const BYTE* src_end, int src_pitch, int row_size, bool rec709);

-extern "C" void mmx_YUY2toRGB32(const BYTE* src, BYTE* dst, const BYTE* src_end, int src_pitch, int row_size, bool rec709);

-#endif

-

-bool BitBltFromYUY2ToRGB(int w, int h, BYTE* dst, int dstpitch, int dbpp, BYTE* src, int srcpitch)

-{

-	void (* YUY2toRGB)(const BYTE* src, BYTE* dst, const BYTE* src_end, int src_pitch, int row_size, bool rec709) = NULL;

-

-#ifndef _WIN64

-	if(g_cpuid.m_flags & CCpuID::mmx)

-	{

-		YUY2toRGB = 

-			dbpp == 32 ? mmx_YUY2toRGB32 :

-			dbpp == 24 ? mmx_YUY2toRGB24 :

-			// dbpp == 16 ? mmx_YUY2toRGB16 : // TODO

-			NULL;

-	}

-	else

-#endif

-	{

-		ASSERT(FALSE);

-		// TODO

-	}

-

-	if(!YUY2toRGB)

-		return(false);

-

-	YUY2toRGB(src, dst, src + h*srcpitch, srcpitch, w, false);

-

-	return(true);

-}

diff --git a/src/DSUtil/vd.h b/src/DSUtil/vd.h
index a69e406c0..0db586cec 100644
--- a/src/DSUtil/vd.h
+++ b/src/DSUtil/vd.h
@@ -1,5 +1,6 @@
 //	VirtualDub - Video processing and capture application

-//	Copyright (C) 1998-2001 Avery Lee

+//	Graphics support library

+//	Copyright (C) 1998-2007 Avery Lee

 //

 //	This program is free software; you can redistribute it and/or modify

 //	it under the terms of the GNU General Public License as published by

@@ -16,25 +17,22 @@
 //	Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.

 //

 //  Notes: 

-//  - BitBltFromI420ToRGB is from VirtualDub

-//  - BitBltFromYUY2ToRGB is from AviSynth 2.52

+//  - VDPixmapBlt is from VirtualDub

+//  - sse2 yv12 to yuy2 conversion by Haali

 //	(- vd.cpp/h should be renamed to something more sensible already :)

 

 #pragma once

 

-class CCpuID {public: CCpuID(); enum flag_t {mmx=1, ssemmx=2, ssefpu=4, sse2=8, _3dnow=16, sse3=32} m_flags;};

+class CCpuID {public: CCpuID(); enum flag_t {mmx=1, ssemmx=2, ssefpu=4, sse2=8, _3dnow=16} m_flags;};

 extern CCpuID g_cpuid;

 

 extern bool BitBltFromI420ToI420(int w, int h, BYTE* dsty, BYTE* dstu, BYTE* dstv, int dstpitch, BYTE* srcy, BYTE* srcu, BYTE* srcv, int srcpitch);

-extern bool BitBltFromI420ToYUY2(int w, int h, BYTE* dst, int dstpitch, BYTE* srcy, BYTE* srcu, BYTE* srcv, int srcpitch, bool fInterlaced = false);

+extern bool BitBltFromI420ToYUY2(int w, int h, BYTE* dst, int dstpitch, BYTE* srcy, BYTE* srcu, BYTE* srcv, int srcpitch);

+extern bool BitBltFromI420ToYUY2Interlaced(int w, int h, BYTE* dst, int dstpitch, BYTE* srcy, BYTE* srcu, BYTE* srcv, int srcpitch);

 extern bool BitBltFromI420ToRGB(int w, int h, BYTE* dst, int dstpitch, int dbpp, BYTE* srcy, BYTE* srcu, BYTE* srcv, int srcpitch /* TODO: , bool fInterlaced = false */);

 extern bool BitBltFromYUY2ToYUY2(int w, int h, BYTE* dst, int dstpitch, BYTE* src, int srcpitch);

 extern bool BitBltFromYUY2ToRGB(int w, int h, BYTE* dst, int dstpitch, int dbpp, BYTE* src, int srcpitch);

 extern bool BitBltFromRGBToRGB(int w, int h, BYTE* dst, int dstpitch, int dbpp, BYTE* src, int srcpitch, int sbpp);

 

 extern void DeinterlaceBlend(BYTE* dst, BYTE* src, DWORD rowbytes, DWORD h, DWORD dstpitch, DWORD srcpitch);

-extern void DeinterlaceBob(BYTE* dst, BYTE* src, DWORD rowbytes, DWORD h, DWORD dstpitch, DWORD srcpitch, bool topfield);

-

-extern void AvgLines8(BYTE* dst, DWORD h, DWORD pitch);

-extern void AvgLines555(BYTE* dst, DWORD h, DWORD pitch);

-extern void AvgLines565(BYTE* dst, DWORD h, DWORD pitch);
\ No newline at end of file
+extern void DeinterlaceBob(BYTE* dst, BYTE* src, DWORD rowbytes, DWORD h, DWORD dstpitch, DWORD srcpitch, bool topfield);
\ No newline at end of file
diff --git a/src/DSUtil/vd_asm.cpp b/src/DSUtil/vd_asm.cpp
index 851449089..3fc521844 100644
--- a/src/DSUtil/vd_asm.cpp
+++ b/src/DSUtil/vd_asm.cpp
@@ -1,5 +1,6 @@
 //	VirtualDub - Video processing and capture application

-//	Copyright (C) 1998-2001 Avery Lee

+//	Graphics support library

+//	Copyright (C) 1998-2007 Avery Lee

 //

 //	This program is free software; you can redistribute it and/or modify

 //	it under the terms of the GNU General Public License as published by

@@ -16,7 +17,7 @@
 //	Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.

 //

 //  Notes: 

-//  - BitBltFromI420ToRGB is from VirtualDub

+//  - VDPixmapBlt is from VirtualDub

 //  - sse2 yv12 to yuy2 conversion by Haali

 //	(- vd.cpp/h should be renamed to something more sensible already :)

 

@@ -428,289 +429,4 @@ last4:
     ret

   };

 }

-

-void __declspec(naked) asm_blend_row_clipped_MMX(BYTE* dst, BYTE* src, DWORD w, DWORD srcpitch)

-{

-	static const __int64 _x0001000100010001 = 0x0001000100010001;

-

-	__asm {

-		push	ebp

-		push	edi

-		push	esi

-		push	ebx

-

-		mov		edi,[esp+20]

-		mov		esi,[esp+24]

-		sub		edi,esi

-		mov		ebp,[esp+28]

-		mov		edx,[esp+32]

-

-		shr		ebp, 3

-

-		movq	mm6, _x0001000100010001

-		pxor	mm7, mm7

-

-xloop:

-		movq		mm0, [esi]

-		movq		mm3, mm0

-		punpcklbw	mm0, mm7

-		punpckhbw	mm3, mm7

-

-		movq		mm1, [esi+edx]

-		movq		mm4, mm1

-		punpcklbw	mm1, mm7

-		punpckhbw	mm4, mm7

-

-		paddw		mm1, mm0

-		paddw		mm1, mm6

-		psrlw		mm1, 1

-

-		paddw		mm4, mm3

-		paddw		mm4, mm6

-		psrlw		mm4, 1

-

-		add			esi, 8

-		packuswb	mm1, mm4

-		movq		[edi+esi-8], mm1

-

-		dec		ebp

-		jne		xloop

-

-		pop		ebx

-		pop		esi

-		pop		edi

-		pop		ebp

-		ret

-	};

-}

-

-void __declspec(naked) asm_blend_row_MMX(BYTE* dst, BYTE* src, DWORD w, DWORD srcpitch)

-{

-	static const __int64 mask0 = 0xfcfcfcfcfcfcfcfci64;

-	static const __int64 mask1 = 0x7f7f7f7f7f7f7f7fi64;

-	static const __int64 mask2 = 0x3f3f3f3f3f3f3f3fi64;

-	static const __int64 _x0002000200020002 = 0x0002000200020002;

-

-	__asm {

-		push	ebp

-		push	edi

-		push	esi

-		push	ebx

-

-		mov		edi, [esp+20]

-		mov		esi, [esp+24]

-		sub		edi, esi

-		mov		ebp, [esp+28]

-		mov		edx, [esp+32]

-

-		shr		ebp, 3

-

-		movq	mm6, _x0002000200020002

-		pxor	mm7, mm7

-

-xloop:

-		movq		mm0, [esi]

-		movq		mm3, mm0

-		punpcklbw	mm0, mm7

-		punpckhbw	mm3, mm7

-

-		movq		mm1, [esi+edx]

-		movq		mm4, mm1

-		punpcklbw	mm1, mm7

-		punpckhbw	mm4, mm7

-

-		movq		mm2, [esi+edx*2]

-		movq		mm5, mm2

-		punpcklbw	mm2, mm7

-		punpckhbw	mm5, mm7

-

-		psllw		mm1, 1

-		paddw		mm1, mm0

-		paddw		mm1, mm2

-		paddw		mm1, mm6

-		psrlw		mm1, 2

-

-		psllw		mm4, 1

-		paddw		mm4, mm3

-		paddw		mm4, mm5

-		paddw		mm4, mm6

-		psrlw		mm4, 2

-

-		add			esi, 8

-		packuswb	mm1, mm4

-		movq		[edi+esi-8], mm1

-

-		dec		ebp

-		jne		xloop

-

-		// sadly the original code makes a lot of visible banding artifacts on yuv

-		// (it seems those shiftings without rounding introduce too much error)

-/*

-		mov		edi,[esp+20]

-		mov		esi,[esp+24]

-		sub		edi,esi

-		mov		ebp,[esp+28]

-		mov		edx,[esp+32]

-

-		movq	mm5,mask0

-		movq	mm6,mask1

-		movq	mm7,mask2

-		shr		ebp,1

-		jz		oddpart

-

-xloop:

-		movq	mm2,[esi]

-		movq	mm0,mm5

-

-		movq	mm1,[esi+edx]

-		pand	mm0,mm2

-

-		psrlq	mm1,1

-		movq	mm2,[esi+edx*2]

-

-		psrlq	mm2,2

-		pand	mm1,mm6

-

-		psrlq	mm0,2

-		pand	mm2,mm7

-

-		paddb	mm0,mm1

-		add		esi,8

-

-		paddb	mm0,mm2

-		dec		ebp

-

-		movq	[edi+esi-8],mm0

-		jne		xloop

-

-oddpart:

-		test	byte ptr [esp+28],1

-		jz		nooddpart

-

-		mov		ecx,[esi]

-		mov		eax,0fcfcfcfch

-		mov		ebx,[esi+edx]

-		and		eax,ecx

-		shr		ebx,1

-		mov		ecx,[esi+edx*2]

-		shr		ecx,2

-		and		ebx,07f7f7f7fh

-		shr		eax,2

-		and		ecx,03f3f3f3fh

-		add		eax,ebx

-		add		eax,ecx

-		mov		[edi+esi],eax

-

-nooddpart:

-*/

-		pop		ebx

-		pop		esi

-		pop		edi

-		pop		ebp

-		ret

-	};

-}

-

-__declspec(align(16)) static BYTE const_1_16_bytes[] = {1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1};

-

-void asm_blend_row_SSE2(BYTE* dst, BYTE* src, DWORD w, DWORD srcpitch)

-{

-	__asm

-	{

-		mov edx, srcpitch

-		mov esi, src

-		mov edi, dst

-		sub edi, esi

-		mov ecx, w

-		mov ebx, ecx

-		shr ecx, 4

-		and ebx, 15

-

-		movdqa xmm7, [const_1_16_bytes] 

-

-asm_blend_row_SSE2_loop:

-		movdqa xmm0, [esi]

-		movdqa xmm1, [esi+edx]

-		movdqa xmm2, [esi+edx*2]

-		pavgb xmm0, xmm1

-		pavgb xmm2, xmm1

-		psubusb xmm0, xmm7

-		pavgb xmm0, xmm2

-		movdqa [esi+edi], xmm0

-		add esi, 16

-		dec	ecx

-		jnz asm_blend_row_SSE2_loop

-

-		test ebx,15

-		jz asm_blend_row_SSE2_end

-

-		mov ecx, ebx

-		xor ax, ax

-		xor bx, bx

-		xor dx, dx

-asm_blend_row_SSE2_loop2:

-		mov al, [esi]

-		mov bl, [esi+edx]

-		mov dl, [esi+edx*2]

-		add ax, bx

-		inc ax

-		shr ax, 1

-		add dx, bx

-		inc dx

-		shr dx, 1

-		add ax, dx

-		shr ax, 1

-		mov [esi+edi], al

-		inc esi

-		dec	ecx

-		jnz asm_blend_row_SSE2_loop2

-

-asm_blend_row_SSE2_end:

-	}

-}

-

-void asm_blend_row_clipped_SSE2(BYTE* dst, BYTE* src, DWORD w, DWORD srcpitch)

-{

-	__asm

-	{

-		mov edx, srcpitch

-		mov esi, src

-		mov edi, dst

-		sub edi, esi

-		mov ecx, w

-		mov ebx, ecx

-		shr ecx, 4

-		and ebx, 15

-

-		movdqa xmm7, [const_1_16_bytes] 

-

-asm_blend_row_clipped_SSE2_loop:

-		movdqa xmm0, [esi]

-		movdqa xmm1, [esi+edx]

-		pavgb xmm0, xmm1

-		movdqa [esi+edi], xmm0

-		add esi, 16

-		dec	ecx

-		jnz asm_blend_row_clipped_SSE2_loop

-

-		test ebx,15

-		jz asm_blend_row_clipped_SSE2_end

-

-		mov ecx, ebx

-		xor ax, ax

-		xor bx, bx

-asm_blend_row_clipped_SSE2_loop2:

-		mov al, [esi]

-		mov bl, [esi+edx]

-		add ax, bx

-		inc ax

-		shr ax, 1

-		mov [esi+edi], al

-		inc esi

-		dec	ecx

-		jnz asm_blend_row_clipped_SSE2_loop2

-

-asm_blend_row_clipped_SSE2_end:

-	}

-}

 #endif

diff --git a/src/DSUtil/vd_asm.h b/src/DSUtil/vd_asm.h
index c1c78f39b..7c1f2f134 100644
--- a/src/DSUtil/vd_asm.h
+++ b/src/DSUtil/vd_asm.h
@@ -1,5 +1,6 @@
 //	VirtualDub - Video processing and capture application

-//	Copyright (C) 1998-2001 Avery Lee

+//	Graphics support library

+//	Copyright (C) 1998-2007 Avery Lee

 //

 //	This program is free software; you can redistribute it and/or modify

 //	it under the terms of the GNU General Public License as published by

@@ -16,8 +17,7 @@
 //	Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.

 //

 //  Notes: 

-//  - BitBltFromI420ToRGB is from VirtualDub

-//  - BitBltFromYUY2ToRGB is from AviSynth 2.52

+//  - VDPixmapBlt is from VirtualDub

 //	(- vd.cpp/h should be renamed to something more sensible already :)

 

 #pragma once

@@ -31,9 +31,4 @@ void yv12_yuy2_row_sse2_linear();
 void yv12_yuy2_row_sse2_linear_interlaced();

 void yv12_yuy2_sse2(const BYTE *Y, const BYTE *U, const BYTE *V, int halfstride, unsigned halfwidth, unsigned height, BYTE *YUY2, int d_stride);

 void yv12_yuy2_sse2_interlaced(const BYTE *Y, const BYTE *U, const BYTE *V, int halfstride, unsigned halfwidth, unsigned height, BYTE *YUY2, int d_stride);

-

-void asm_blend_row_clipped_MMX(BYTE* dst, BYTE* src, DWORD w, DWORD srcpitch);

-void asm_blend_row_MMX(BYTE* dst, BYTE* src, DWORD w, DWORD srcpitch);

-void asm_blend_row_SSE2(BYTE* dst, BYTE* src, DWORD w, DWORD srcpitch);

-void asm_blend_row_clipped_SSE2(BYTE* dst, BYTE* src, DWORD w, DWORD srcpitch);

 #endif