GPU: Vectorized mosaic rendering now uses an aligned working buffer, improving performance for SSE2 and AVX2.

desmume/src/GPU.h

@@ -2,7 +2,7 @@
 	Copyright (C) 2006 yopyop
 	Copyright (C) 2006-2007 Theo Berkau
 	Copyright (C) 2007 shash
-	Copyright (C) 2009-2023 DeSmuME team
+	Copyright (C) 2009-2025 DeSmuME team
 
 	This file is free software: you can redistribute it and/or modify
 	it under the terms of the GNU General Public License as published by
@@ -102,7 +102,7 @@ enum DisplayCaptureSize
 	DisplayCaptureSize_128x128	= 0,
 	DisplayCaptureSize_256x64	= 1,
 	DisplayCaptureSize_256x128	= 2,
-	DisplayCaptureSize_256x192	= 3,
+	DisplayCaptureSize_256x192	= 3
 };
 
 typedef union
@@ -1464,7 +1464,7 @@ protected:
 	itemsForPriority_t _itemsForPriority[NB_PRIORITIES];
 	
 	struct MosaicColor {
-		CACHE_ALIGN u16 bg[4][GPU_FRAMEBUFFER_NATIVE_WIDTH + sizeof(u32)]; // Pad this buffer a little bit to avoid buffer overruns with vectorized gather instructions.
+		CACHE_ALIGN u16 bg[4][GPU_FRAMEBUFFER_NATIVE_WIDTH + 64]; // Pad this buffer a little bit to avoid buffer overruns with vectorized gather instructions.
 		struct Obj {
 			u16 color;
 			u8 alpha;
@@ -1843,6 +1843,8 @@ public:
 class GPUEventHandler
 {
 public:
+	virtual ~GPUEventHandler() {};
+	
 	virtual void DidFrameBegin(const size_t line, const bool isFrameSkipRequested, const size_t pageCount, u8 &selectedBufferIndexInOut) = 0;
 	virtual void DidFrameEnd(bool isFrameSkipped, const NDSDisplayInfo &latestDisplayInfo) = 0;
 	virtual void DidRender3DBegin() = 0;
@@ -1859,6 +1861,8 @@ public:
 class GPUEventHandlerDefault : public GPUEventHandler
 {
 public:
+	virtual ~GPUEventHandlerDefault() {};
+	
 	virtual void DidFrameBegin(const size_t line, const bool isFrameSkipRequested, const size_t pageCount, u8 &selectedBufferIndexInOut);
 	virtual void DidFrameEnd(bool isFrameSkipped, const NDSDisplayInfo &latestDisplayInfo) {};
 	virtual void DidRender3DBegin() {};

desmume/src/GPU_Operations_AVX2.cpp

@@ -2462,25 +2462,25 @@ void GPUEngineBase::_MosaicLine(GPUEngineCompositorInfo &compInfo)
 				_mm256_blendv_epi8(_mm256_and_si256(dstColor16[1], _mm256_set1_epi16(0x7FFF)), _mm256_set1_epi16(0xFFFF), idxMask16[1])
 			};
 			
-			const v256u16 mosaicSetColorMask8 = _mm256_permute4x64_epi64( _mm256_cmpeq_epi16(_mm256_loadu_si256((v256u8 *)(compInfo.renderState.mosaicWidthBG->begin + x)), _mm256_setzero_si256()), 0xD8 );
+			const v256u16 mosaicSetColorMask8 = _mm256_permute4x64_epi64( _mm256_cmpeq_epi16(_mm256_load_si256((v256u8 *)(compInfo.renderState.mosaicWidthBG->begin + x)), _mm256_setzero_si256()), 0xD8 );
 			const v256u16 mosaicSetColorMask16[2] = {
 				_mm256_unpacklo_epi8(mosaicSetColorMask8, mosaicSetColorMask8),
 				_mm256_unpackhi_epi8(mosaicSetColorMask8, mosaicSetColorMask8)
 			};
 			
 			__m256i mosaicColorOut[2];
-			mosaicColorOut[0] = _mm256_blendv_epi8(mosaicColor16[0], _mm256_loadu_si256((v256u16 *)(mosaicColorBG + x) + 0), mosaicSetColorMask16[0]);
-			mosaicColorOut[1] = _mm256_blendv_epi8(mosaicColor16[1], _mm256_loadu_si256((v256u16 *)(mosaicColorBG + x) + 1), mosaicSetColorMask16[1]);
+			mosaicColorOut[0] = _mm256_blendv_epi8(mosaicColor16[0], _mm256_load_si256((v256u16 *)(mosaicColorBG + x) + 0), mosaicSetColorMask16[0]);
+			mosaicColorOut[1] = _mm256_blendv_epi8(mosaicColor16[1], _mm256_load_si256((v256u16 *)(mosaicColorBG + x) + 1), mosaicSetColorMask16[1]);
 			
-			_mm256_storeu_si256((v256u16 *)(mosaicColorBG + x) + 0, mosaicColorOut[0]);
-			_mm256_storeu_si256((v256u16 *)(mosaicColorBG + x) + 1, mosaicColorOut[1]);
+			_mm256_store_si256((v256u16 *)(mosaicColorBG + x) + 0, mosaicColorOut[0]);
+			_mm256_store_si256((v256u16 *)(mosaicColorBG + x) + 1, mosaicColorOut[1]);
 		}
 		
 		const v256u32 outColor32idx[4] = {
-			_mm256_loadu_si256((v256u32 *)(compInfo.renderState.mosaicWidthBG->trunc32 + x) + 0),
-			_mm256_loadu_si256((v256u32 *)(compInfo.renderState.mosaicWidthBG->trunc32 + x) + 1),
-			_mm256_loadu_si256((v256u32 *)(compInfo.renderState.mosaicWidthBG->trunc32 + x) + 2),
-			_mm256_loadu_si256((v256u32 *)(compInfo.renderState.mosaicWidthBG->trunc32 + x) + 3)
+			_mm256_load_si256((v256u32 *)(compInfo.renderState.mosaicWidthBG->trunc32 + x) + 0),
+			_mm256_load_si256((v256u32 *)(compInfo.renderState.mosaicWidthBG->trunc32 + x) + 1),
+			_mm256_load_si256((v256u32 *)(compInfo.renderState.mosaicWidthBG->trunc32 + x) + 2),
+			_mm256_load_si256((v256u32 *)(compInfo.renderState.mosaicWidthBG->trunc32 + x) + 3)
 		};
 		
 		const v256u16 outColor32[4] = {

desmume/src/GPU_Operations_SSE2.cpp

@@ -2641,14 +2641,14 @@ void GPUEngineBase::_MosaicLine(GPUEngineCompositorInfo &compInfo)
 				_mm_blendv_epi8(_mm_and_si128(dstColor16[1], _mm_set1_epi16(0x7FFF)), _mm_set1_epi16(0xFFFF), idxMask16[1])
 			};
 			
-			const v128u16 mosaicSetColorMask8 = _mm_cmpeq_epi16( _mm_loadu_si128((v128u8 *)(compInfo.renderState.mosaicWidthBG->begin + x)), _mm_setzero_si128() );
+			const v128u16 mosaicSetColorMask8 = _mm_cmpeq_epi16( _mm_load_si128((v128u8 *)(compInfo.renderState.mosaicWidthBG->begin + x)), _mm_setzero_si128() );
 			const v128u16 mosaicSetColorMask16[2] = {
 				_mm_unpacklo_epi8(mosaicSetColorMask8, mosaicSetColorMask8),
 				_mm_unpackhi_epi8(mosaicSetColorMask8, mosaicSetColorMask8)
 			};
 			
-			_mm_storeu_si128( (v128u16 *)(mosaicColorBG + x) + 0, _mm_blendv_epi8(mosaicColor16[0], _mm_loadu_si128((v128u16 *)(mosaicColorBG + x) + 0), mosaicSetColorMask16[0]) );
-			_mm_storeu_si128( (v128u16 *)(mosaicColorBG + x) + 1, _mm_blendv_epi8(mosaicColor16[1], _mm_loadu_si128((v128u16 *)(mosaicColorBG + x) + 1), mosaicSetColorMask16[1]) );
+			_mm_store_si128( (v128u16 *)(mosaicColorBG + x) + 0, _mm_blendv_epi8(mosaicColor16[0], _mm_load_si128((v128u16 *)(mosaicColorBG + x) + 0), mosaicSetColorMask16[0]) );
+			_mm_store_si128( (v128u16 *)(mosaicColorBG + x) + 1, _mm_blendv_epi8(mosaicColor16[1], _mm_load_si128((v128u16 *)(mosaicColorBG + x) + 1), mosaicSetColorMask16[1]) );
 		}
 		
 		const v128u16 outColor16[2] = {

desmume/src/filter/hq2x.cpp

@@ -149,9 +149,9 @@
 //  }
 //}
 
-static void hq2x_32_def(u32 *__restrict dst0, u32 *__restrict dst1, const u32 *src0, const u32 *src1, const u32 *src2, unsigned count)
+static void hq2x_32_def(u32 *__restrict dst0, u32 *__restrict dst1, const u32 *src0, const u32 *src1, const u32 *src2, size_t count)
 {
-	for (unsigned i = 0; i < count; ++i)
+	for (size_t i = 0; i < count; ++i)
 	{
 		u8 mask = 0;
 		u32 c[9];
@@ -407,9 +407,9 @@ static void hq2x_32_def(u32 *__restrict dst0, u32 *__restrict dst1, const u32 *s
 //  }
 //}
 
-static void hq2xS_32_def(u32 *__restrict dst0, u32 *__restrict dst1, const u32 *src0, const u32 *src1, const u32 *src2, unsigned count)
+static void hq2xS_32_def(u32 *__restrict dst0, u32 *__restrict dst1, const u32 *src0, const u32 *src1, const u32 *src2, size_t count)
 {
-	for (unsigned i = 0; i < count; ++i)
+	for (size_t i = 0; i < count; ++i)
 	{
 		u8 mask = 0;
 		u32 c[9];
@@ -460,7 +460,7 @@ static void hq2xS_32_def(u32 *__restrict dst0, u32 *__restrict dst1, const u32 *
 			
 			brightArray[j] = bright;
 		}
-		int diffBright = ((maxBright - minBright) * 7) >> 4;
+		u32 diffBright = ((u32)(maxBright - minBright) * 7) >> 4;
 		if(diffBright > 7)
 		{
 			const int centerBright = brightArray[4];

desmume/src/filter/hq3x.cpp

@@ -39,9 +39,9 @@
  * This effect is a rewritten implementation of the hq3x effect made by Maxim Stepin
  */
 
-void hq3x_32_def(u32 *__restrict dst0, u32 *__restrict dst1, u32 *__restrict dst2, const u32 *src0, const u32 *src1, const u32 *src2, int count)
+void hq3x_32_def(u32 *__restrict dst0, u32 *__restrict dst1, u32 *__restrict dst2, const u32 *src0, const u32 *src1, const u32 *src2, size_t count)
 {
-	for (int i = 0; i < count; ++i)
+	for (size_t i = 0; i < count; ++i)
 	{
 		u8 mask = 0;
 		u32 c[9];
@@ -125,9 +125,9 @@ void hq3x_32_def(u32 *__restrict dst0, u32 *__restrict dst1, u32 *__restrict dst
 	}
 }
 
-void hq3xS_32_def(u32 *__restrict dst0, u32 *__restrict dst1, u32 *__restrict dst2, const u32 *src0, const u32 *src1, const u32 *src2, int count)
+void hq3xS_32_def(u32 *__restrict dst0, u32 *__restrict dst1, u32 *__restrict dst2, const u32 *src0, const u32 *src1, const u32 *src2, size_t count)
 {
-	for (int i = 0; i < count; ++i)
+	for (size_t i = 0; i < count; ++i)
 	{
 		u8 mask = 0;
 		u32 c[9];
@@ -178,7 +178,7 @@ void hq3xS_32_def(u32 *__restrict dst0, u32 *__restrict dst1, u32 *__restrict ds
 			
 			brightArray[j] = bright;
 		}
-		unsigned int diffBright = ((maxBright - minBright) * 7) >> 4;
+		u32 diffBright = ((u32)(maxBright - minBright) * 7) >> 4;
 		if(diffBright > 7)
 		{
 			const int centerBright = brightArray[4];

desmume/src/filter/hq4x.cpp

@@ -41,9 +41,9 @@
 
 void hq4x_32_def(u32 *__restrict dst0, u32 *__restrict dst1, u32 *__restrict dst2, u32 *__restrict dst3,
 				 const u32 *src0, const u32 *src1, const u32 *src2,
-				 unsigned count, unsigned flag)
+				 size_t count, unsigned flag)
 {
-	for (unsigned i = 0; i < count; ++i)
+	for (size_t i = 0; i < count; ++i)
 	{
 		u8 mask = 0;
 		u32 c[9];
@@ -130,9 +130,9 @@ void hq4x_32_def(u32 *__restrict dst0, u32 *__restrict dst1, u32 *__restrict dst
 
 void hq4xS_32_def(u32 *__restrict dst0, u32 *__restrict dst1, u32 *__restrict dst2, u32 *__restrict dst3,
 				  const u32 *src0, const u32 *src1, const u32 *src2,
-				  unsigned count, unsigned flag)
+				  size_t count, unsigned flag)
 {
-	for (unsigned i = 0; i < count; ++i)
+	for (size_t i = 0; i < count; ++i)
 	{
 		u8 mask = 0;
 		u32 c[9];
@@ -183,7 +183,7 @@ void hq4xS_32_def(u32 *__restrict dst0, u32 *__restrict dst1, u32 *__restrict ds
 			
 			brightArray[j] = bright;
 		}
-		unsigned int diffBright = ((maxBright - minBright) * 7) >> 4;
+		u32 diffBright = ((u32)(maxBright - minBright) * 7) >> 4;
 		if(diffBright > 7)
 		{
 			const int centerBright = brightArray[4];

desmume/src/filter/lq2x.cpp

@@ -270,13 +270,11 @@
 //   }
 //}
 
-static void lq2x_32_def(u32* dst0, u32* dst1, const u32* src0, const u32* src1, const u32* src2, unsigned count)
+static void lq2x_32_def(u32* dst0, u32* dst1, const u32* src0, const u32* src1, const u32* src2, size_t count)
 {
-	unsigned i;
-
-	for(i=0;i<count;++i) {
-		unsigned char mask;
-
+	for (size_t i = 0; i < count; ++i)
+	{
+		u8 mask = 0;
 		u32 c[9];
 
 		c[1] = src0[0];
@@ -303,8 +301,6 @@ static void lq2x_32_def(u32* dst0, u32* dst1, const u32* src0, const u32* src1,
 			c[8] = c[7];
 		}
 
-		mask = 0;
-
 		if (c[0] != c[4])
 			mask |= 1 << 0;
 		if (c[1] != c[4])
@@ -380,14 +376,12 @@ static void lq2x_32_def(u32* dst0, u32* dst1, const u32* src0, const u32* src1,
 	}
 }
 
-static void lq2xS_32_def(u32* dst0, u32* dst1, const u32* src0, const u32* src1, const u32* src2, unsigned count)
+static void lq2xS_32_def(u32* dst0, u32* dst1, const u32* src0, const u32* src1, const u32* src2, size_t count)
 {
-   unsigned i;
-
-   for(i=0;i<count;++i) {
-      unsigned char mask = 0;
-
-      u32 c[9];
+	for (size_t i = 0; i < count; ++i)
+	{
+		u8 mask = 0;
+		u32 c[9];
 
       c[1] = src0[0];
       c[4] = src1[0];
@@ -429,7 +423,7 @@ static void lq2xS_32_def(u32* dst0, u32* dst1, const u32* src0, const u32* src1,
 
 		brightArray[j] = bright;
 	}
-	int diffBright = (maxBright - minBright) >> 4;
+	u32 diffBright = (u32)(maxBright - minBright) >> 4;
 	if(diffBright > 1)
 	{
 		const int centerBright = brightArray[4];

desmume/src/filter/videofilter.cpp

@@ -306,7 +306,7 @@ bool VideoFilter::SetSourceSize(const size_t width, const size_t height)
 	}
 	memset(newPixBuffer, 0, width * (height + 8) * sizeof(uint32_t));
 	
-	if (this->__vfSrcSurface.Surface == NULL || this->__vfSrcSurface.Width != width || this->__vfSrcSurface.Height != height)
+	if ( (this->__vfSrcSurface.Surface == NULL) || ((size_t)this->__vfSrcSurface.Width != width) || ((size_t)this->__vfSrcSurface.Height != height) )
 	{
 		sizeChanged = true;
 	}

desmume/src/frontend/cocoa/userinterface/MacOGLDisplayView.mm

@@ -468,7 +468,7 @@ void MacOGLDisplayPresenter::Init()
 	GLint profileVersion = 0;
 	CGLDescribePixelFormat(this->_pixelFormat, 0, kCGLPFAOpenGLProfile, &profileVersion);
 	
-	if (profileVersion == kCGLOGLPVersion_3_2_Core)
+	if ( (profileVersion == kCGLOGLPVersion_3_2_Core) || (profileVersion == 0x4100) )
 	{
 		this->_contextInfo = new OGLContextInfo_3_2;
 	}

desmume/src/rasterize.cpp

@@ -1,5 +1,5 @@
 /*
-	Copyright (C) 2009-2024 DeSmuME team
+	Copyright (C) 2009-2025 DeSmuME team
 
 	This file is free software: you can redistribute it and/or modify
 	it under the terms of the GNU General Public License as published by
@@ -880,7 +880,7 @@ FORCEINLINE void RasterizerUnit<RENDERER>::_drawscanline(const POLYGON_ATTR poly
 
 	//CONSIDER: in case some other math is wrong (shouldve been clipped OK), we might go out of bounds here.
 	//better check the Y value.
-	if ( (pLeft.y < 0) || (pLeft.y >= framebufferHeight) )
+	if ( (pLeft.y < 0) || (pLeft.y >= (s32)framebufferHeight) )
 	{
 		const float gpuScalingFactorHeight = (float)framebufferHeight / (float)GPU_FRAMEBUFFER_NATIVE_HEIGHT;
 		printf("rasterizer rendering at y=%d! oops! (x%.1f)\n", pLeft.y, gpuScalingFactorHeight);
@@ -924,7 +924,7 @@ FORCEINLINE void RasterizerUnit<RENDERER>::_drawscanline(const POLYGON_ATTR poly
 		x = 0;
 	}
 	
-	if (x+rasterWidth > framebufferWidth)
+	if (x+rasterWidth > (s32)framebufferWidth)
 	{
 		if (RENDERER && !USELINEHACK)
 		{
@@ -1036,7 +1036,7 @@ void RasterizerUnit<RENDERER>::_runscanlines(const POLYGON_ATTR polyAttr, const
 	bool first = true;
 
 	//HACK: special handling for horizontal line poly
-	if ( USELINEHACK && (left.height == 0) && (right.height == 0) && (left.y < framebufferHeight) && (left.y >= 0) )
+	if ( USELINEHACK && (left.height == 0) && (right.height == 0) && (left.y < (s32)framebufferHeight) && (left.y >= 0) )
 	{
 		const bool draw = ( !SLI || ((left.y >= this->_SLI_startLine) && (left.y < this->_SLI_endLine)) );
 		if (draw)
@@ -1265,11 +1265,11 @@ void RasterizerUnit<RENDERER>::_shape_engine(const POLYGON_ATTR polyAttr, const
 }
 
 template<bool RENDERER>
-void RasterizerUnit<RENDERER>::SetSLI(u32 startLine, u32 endLine, bool debug)
+void RasterizerUnit<RENDERER>::SetSLI(const size_t startLine, const size_t endLine, bool debug)
 {
 	this->_debug_thisPoly = debug;
-	this->_SLI_startLine = startLine;
-	this->_SLI_endLine = endLine;
+	this->_SLI_startLine = (s32)startLine;
+	this->_SLI_endLine = (s32)endLine;
 }
 
 template<bool RENDERER>
@@ -1340,22 +1340,22 @@ FORCEINLINE void RasterizerUnit<RENDERER>::Render()
 			{
 				if (useLineHack)
 				{
-					this->_shape_engine<SLI, true, true, true>(polyAttr, isPolyTranslucent, dstColor, dstWidth, dstHeight, vertCount);
+					this->_shape_engine<SLI, true, true, true>(polyAttr, isPolyTranslucent, dstColor, dstWidth, dstHeight, (int)vertCount);
 				}
 				else
 				{
-					this->_shape_engine<SLI, true, true, false>(polyAttr, isPolyTranslucent, dstColor, dstWidth, dstHeight, vertCount);
+					this->_shape_engine<SLI, true, true, false>(polyAttr, isPolyTranslucent, dstColor, dstWidth, dstHeight, (int)vertCount);
 				}
 			}
 			else
 			{
 				if (useLineHack)
 				{
-					this->_shape_engine<SLI, true, false, true>(polyAttr, isPolyTranslucent, dstColor, dstWidth, dstHeight, vertCount);
+					this->_shape_engine<SLI, true, false, true>(polyAttr, isPolyTranslucent, dstColor, dstWidth, dstHeight, (int)vertCount);
 				}
 				else
 				{
-					this->_shape_engine<SLI, true, false, false>(polyAttr, isPolyTranslucent, dstColor, dstWidth, dstHeight, vertCount);
+					this->_shape_engine<SLI, true, false, false>(polyAttr, isPolyTranslucent, dstColor, dstWidth, dstHeight, (int)vertCount);
 				}
 			}
 		}
@@ -1365,22 +1365,22 @@ FORCEINLINE void RasterizerUnit<RENDERER>::Render()
 			{
 				if (useLineHack)
 				{
-					this->_shape_engine<SLI, false, true, true>(polyAttr, isPolyTranslucent, dstColor, dstWidth, dstHeight, vertCount);
+					this->_shape_engine<SLI, false, true, true>(polyAttr, isPolyTranslucent, dstColor, dstWidth, dstHeight, (int)vertCount);
 				}
 				else
 				{
-					this->_shape_engine<SLI, false, true, false>(polyAttr, isPolyTranslucent, dstColor, dstWidth, dstHeight, vertCount);
+					this->_shape_engine<SLI, false, true, false>(polyAttr, isPolyTranslucent, dstColor, dstWidth, dstHeight, (int)vertCount);
 				}
 			}
 			else
 			{
 				if (useLineHack)
 				{
-					this->_shape_engine<SLI, false, false, true>(polyAttr, isPolyTranslucent, dstColor, dstWidth, dstHeight, vertCount);
+					this->_shape_engine<SLI, false, false, true>(polyAttr, isPolyTranslucent, dstColor, dstWidth, dstHeight, (int)vertCount);
 				}
 				else
 				{
-					this->_shape_engine<SLI, false, false, false>(polyAttr, isPolyTranslucent, dstColor, dstWidth, dstHeight, vertCount);
+					this->_shape_engine<SLI, false, false, false>(polyAttr, isPolyTranslucent, dstColor, dstWidth, dstHeight, (int)vertCount);
 				}
 			}
 		}
@@ -1770,7 +1770,7 @@ SoftRasterizerRenderer::SoftRasterizerRenderer()
 	_enableLineHack = CommonSettings.GFX3D_LineHack;
 	_enableFragmentSamplingHack = CommonSettings.GFX3D_TXTHack;
 	
-	_HACK_viewer_rasterizerUnit.SetSLI(0, (u32)_framebufferHeight, false);
+	_HACK_viewer_rasterizerUnit.SetSLI(0, _framebufferHeight, false);
 	
 	const size_t coreCount = CommonSettings.num_cores;
 	_threadCount = coreCount;
@@ -1801,7 +1801,7 @@ SoftRasterizerRenderer::SoftRasterizerRenderer()
 		_threadClearParam[0].startPixel = 0;
 		_threadClearParam[0].endPixel = _framebufferPixCount;
 		
-		_rasterizerUnit[0].SetSLI((u32)_threadPostprocessParam[0].startLine, (u32)_threadPostprocessParam[0].endLine, false);
+		_rasterizerUnit[0].SetSLI(_threadPostprocessParam[0].startLine, _threadPostprocessParam[0].endLine, false);
 		_rasterizerUnit[0].SetRenderer(this);
 	}
 	else
@@ -1827,7 +1827,7 @@ SoftRasterizerRenderer::SoftRasterizerRenderer()
 			_threadClearParam[i].startPixel = i * _customPixelsPerThread;
 			_threadClearParam[i].endPixel = (i < _threadCount - 1) ? (i + 1) * _customPixelsPerThread : _framebufferPixCount;
 			
-			_rasterizerUnit[i].SetSLI((u32)_threadPostprocessParam[i].startLine, (u32)_threadPostprocessParam[i].endLine, false);
+			_rasterizerUnit[i].SetSLI(_threadPostprocessParam[i].startLine, _threadPostprocessParam[i].endLine, false);
 			_rasterizerUnit[i].SetRenderer(this);
 			
 			char name[16];
@@ -2473,7 +2473,7 @@ Render3DError SoftRasterizerRenderer::SetFramebufferSize(size_t w, size_t h)
 		this->_threadClearParam[0].startPixel = 0;
 		this->_threadClearParam[0].endPixel = pixCount;
 		
-		this->_rasterizerUnit[0].SetSLI((u32)this->_threadPostprocessParam[0].startLine, (u32)this->_threadPostprocessParam[0].endLine, false);
+		this->_rasterizerUnit[0].SetSLI(this->_threadPostprocessParam[0].startLine, this->_threadPostprocessParam[0].endLine, false);
 	}
 	else
 	{
@@ -2488,7 +2488,7 @@ Render3DError SoftRasterizerRenderer::SetFramebufferSize(size_t w, size_t h)
 			this->_threadClearParam[i].startPixel = i * this->_customPixelsPerThread;
 			this->_threadClearParam[i].endPixel = (i < this->_threadCount - 1) ? (i + 1) * this->_customPixelsPerThread : pixCount;
 			
-			this->_rasterizerUnit[i].SetSLI((u32)this->_threadPostprocessParam[i].startLine, (u32)this->_threadPostprocessParam[i].endLine, false);
+			this->_rasterizerUnit[i].SetSLI(this->_threadPostprocessParam[i].startLine, this->_threadPostprocessParam[i].endLine, false);
 		}
 	}
 	

desmume/src/rasterize.h

@@ -1,5 +1,5 @@
 /*
-	Copyright (C) 2009-2023 DeSmuME team
+	Copyright (C) 2009-2025 DeSmuME team
 
 	This file is free software: you can redistribute it and/or modify
 	it under the terms of the GNU General Public License as published by
@@ -116,8 +116,8 @@ class RasterizerUnit
 {
 protected:
 	bool _debug_thisPoly;
-	u32 _SLI_startLine;
-	u32 _SLI_endLine;
+	s32 _SLI_startLine;
+	s32 _SLI_endLine;
 	
 	SoftRasterizerRenderer *_softRender;
 	SoftRasterizerTexture *_currentTexture;
@@ -139,7 +139,7 @@ protected:
 	template<bool SLI, bool ISFRONTFACING, bool ISSHADOWPOLYGON, bool USELINEHACK> void _shape_engine(const POLYGON_ATTR polyAttr, const bool isPolyTranslucent, Color4u8 *dstColor, const size_t framebufferWidth, const size_t framebufferHeight, int type);
 	
 public:
-	void SetSLI(u32 startLine, u32 endLine, bool debug);
+	void SetSLI(const size_t startLine, const size_t endLine, bool debug);
 	void SetRenderer(SoftRasterizerRenderer *theRenderer);
 	template<bool SLI, bool USELINEHACK> FORCEINLINE void Render();
 };

desmume/src/render3D.cpp

@@ -886,7 +886,7 @@ Render3DError Render3D_SIMD<SIMDBYTES>::SetFramebufferSize(size_t w, size_t h)
 		return RENDER3DERROR_NOERR;
 	}
 	
-	this->_framebufferSIMDPixCount = (SIMDBYTES > 0) ? this->_framebufferPixCount - (this->_framebufferPixCount % SIMDBYTES) : _framebufferPixCount;
+	this->_framebufferSIMDPixCount = (SIMDBYTES > 0) ? this->_framebufferPixCount - (this->_framebufferPixCount % SIMDBYTES) : this->_framebufferPixCount;
 	
 	return error;
 }

desmume/src/render3D.h

@@ -1,6 +1,6 @@
 /*
 	Copyright (C) 2006-2007 shash
-	Copyright (C) 2007-2024 DeSmuME team
+	Copyright (C) 2007-2025 DeSmuME team
 
 	This file is free software: you can redistribute it and/or modify
 	it under the terms of the GNU General Public License as published by
@@ -266,7 +266,7 @@ public:
 	static void* operator new(size_t size);
 	static void operator delete(void *p);
 	Render3D();
-	~Render3D();
+	virtual ~Render3D();
 	
 	const Render3DDeviceInfo& GetDeviceInfo();
 	RendererID GetRenderID();
@@ -332,6 +332,7 @@ class Render3D_SIMD : public Render3D
 {
 public:
 	Render3D_SIMD();
+	virtual ~Render3D_SIMD() {}
 	
 	virtual Render3DError SetFramebufferSize(size_t w, size_t h);
 };
@@ -341,6 +342,7 @@ public:
 class Render3D_AVX2 : public Render3D_SIMD<32>
 {
 public:
+	virtual ~Render3D_AVX2() {}
 	virtual void _ClearImageBaseLoop(const u16 *__restrict inColor16, const u16 *__restrict inDepth16, u16 *__restrict outColor16, u32 *__restrict outDepth24, u8 *__restrict outFog);
 };
 
@@ -349,6 +351,7 @@ public:
 class Render3D_SSE2 : public Render3D_SIMD<16>
 {
 public:
+	virtual ~Render3D_SSE2() {}
 	virtual void _ClearImageBaseLoop(const u16 *__restrict inColor16, const u16 *__restrict inDepth16, u16 *__restrict outColor16, u32 *__restrict outDepth24, u8 *__restrict outFog);
 };
 
@@ -357,6 +360,7 @@ public:
 class Render3D_NEON : public Render3D_SIMD<16>
 {
 public:
+	virtual ~Render3D_NEON() {}
 	virtual void _ClearImageBaseLoop(const u16 *__restrict inColor16, const u16 *__restrict inDepth16, u16 *__restrict outColor16, u32 *__restrict outDepth24, u8 *__restrict outFog);
 };
 
@@ -365,6 +369,7 @@ public:
 class Render3D_AltiVec : public Render3D_SIMD<16>
 {
 public:
+	virtual ~Render3D_AltiVec() {}
 	virtual void _ClearImageBaseLoop(const u16 *__restrict inColor16, const u16 *__restrict inDepth16, u16 *__restrict outColor16, u32 *__restrict outDepth24, u8 *__restrict outFog);
 };