Experiment with splitting triangle block coverage from block drawing

src/r_poly_triangle.cpp

@@ -92,6 +92,7 @@ void PolyTriangleDrawer::draw_arrays(const PolyDrawArgs &drawargs, TriDrawVarian
 	switch (variant)
 	{
 	default:
+	//case TriDrawVariant::DrawNormal: drawfunc = &ScreenTriangle::DrawFunc; break;
 	case TriDrawVariant::DrawNormal: drawfunc = dest_bgra ? llvm->TriDrawNormal32[bmode] : llvm->TriDrawNormal8[bmode]; break;
 	case TriDrawVariant::FillNormal: drawfunc = dest_bgra ? llvm->TriFillNormal32[bmode] : llvm->TriFillNormal8[bmode]; break;
 	case TriDrawVariant::DrawSubsector: drawfunc = dest_bgra ? llvm->TriDrawSubsector32[bmode] : llvm->TriDrawSubsector8[bmode]; break;
@@ -569,3 +570,263 @@ void PolyVertexBuffer::Clear()
 {
 	NextBufferVertex = 0;
 }
+
+/////////////////////////////////////////////////////////////////////////////
+
+void ScreenTriangle::Setup(const TriDrawTriangleArgs *args, WorkerThreadData *thread)
+{
+	const TriVertex &v1 = *args->v1;
+	const TriVertex &v2 = *args->v2;
+	const TriVertex &v3 = *args->v3;
+	int clipright = args->clipright;
+	int clipbottom = args->clipbottom;
+	
+	ScreenTriangleFullSpan *span = FullSpans;
+	ScreenTrianglePartialBlock *partial = PartialBlocks;
+	int curSpan = 0;
+	int curPartial = 0;
+	span->Length = 0;
+	
+	// 28.4 fixed-point coordinates
+	const int Y1 = (int)round(16.0f * v1.y);
+	const int Y2 = (int)round(16.0f * v2.y);
+	const int Y3 = (int)round(16.0f * v3.y);
+	
+	const int X1 = (int)round(16.0f * v1.x);
+	const int X2 = (int)round(16.0f * v2.x);
+	const int X3 = (int)round(16.0f * v3.x);
+	
+	// Deltas
+	const int DX12 = X1 - X2;
+	const int DX23 = X2 - X3;
+	const int DX31 = X3 - X1;
+	
+	const int DY12 = Y1 - Y2;
+	const int DY23 = Y2 - Y3;
+	const int DY31 = Y3 - Y1;
+	
+	// Fixed-point deltas
+	const int FDX12 = DX12 << 4;
+	const int FDX23 = DX23 << 4;
+	const int FDX31 = DX31 << 4;
+	
+	const int FDY12 = DY12 << 4;
+	const int FDY23 = DY23 << 4;
+	const int FDY31 = DY31 << 4;
+	
+	// Bounding rectangle
+	int minx = MAX((MIN(MIN(X1, X2), X3) + 0xF) >> 4, 0);
+	int maxx = MIN((MAX(MAX(X1, X2), X3) + 0xF) >> 4, clipright - 1);
+	int miny = MAX((MIN(MIN(Y1, Y2), Y3) + 0xF) >> 4, 0);
+	int maxy = MIN((MAX(MAX(Y1, Y2), Y3) + 0xF) >> 4, clipbottom - 1);
+	if (minx >= maxx || miny >= maxy)
+		return;
+	
+	// Block size, standard 8x8 (must be power of two)
+	const int q = 8;
+	
+	// Start in corner of 8x8 block
+	minx &= ~(q - 1);
+	miny &= ~(q - 1);
+	
+	// Half-edge constants
+	int C1 = DY12 * X1 - DX12 * Y1;
+	int C2 = DY23 * X2 - DX23 * Y2;
+	int C3 = DY31 * X3 - DX31 * Y3;
+	
+	// Correct for fill convention
+	if (DY12 < 0 || (DY12 == 0 && DX12 > 0)) C1++;
+	if (DY23 < 0 || (DY23 == 0 && DX23 > 0)) C2++;
+	if (DY31 < 0 || (DY31 == 0 && DX31 > 0)) C3++;
+	
+	// First block line for this thread
+	int core = thread->core;
+	int num_cores = thread->num_cores;
+	int core_skip = (num_cores - ((miny / q) - core) % num_cores) % num_cores;
+	miny += core_skip * q;
+	
+	// Loop through blocks
+	for (int y = miny; y < maxy; y += q * num_cores)
+	{
+		for (int x = minx; x < maxx; x += q)
+		{
+			// Corners of block
+			int x0 = x << 4;
+			int x1 = (x + q - 1) << 4;
+			int y0 = y << 4;
+			int y1 = (y + q - 1) << 4;
+			
+			// Evaluate half-space functions
+			bool a00 = C1 + DX12 * y0 - DY12 * x0 > 0;
+			bool a10 = C1 + DX12 * y0 - DY12 * x1 > 0;
+			bool a01 = C1 + DX12 * y1 - DY12 * x0 > 0;
+			bool a11 = C1 + DX12 * y1 - DY12 * x1 > 0;
+			int a = (a00 << 0) | (a10 << 1) | (a01 << 2) | (a11 << 3);
+			
+			bool b00 = C2 + DX23 * y0 - DY23 * x0 > 0;
+			bool b10 = C2 + DX23 * y0 - DY23 * x1 > 0;
+			bool b01 = C2 + DX23 * y1 - DY23 * x0 > 0;
+			bool b11 = C2 + DX23 * y1 - DY23 * x1 > 0;
+			int b = (b00 << 0) | (b10 << 1) | (b01 << 2) | (b11 << 3);
+			
+			bool c00 = C3 + DX31 * y0 - DY31 * x0 > 0;
+			bool c10 = C3 + DX31 * y0 - DY31 * x1 > 0;
+			bool c01 = C3 + DX31 * y1 - DY31 * x0 > 0;
+			bool c11 = C3 + DX31 * y1 - DY31 * x1 > 0;
+			int c = (c00 << 0) | (c10 << 1) | (c01 << 2) | (c11 << 3);
+			
+			// Skip block when outside an edge
+			if (a == 0x0 || b == 0x0 || c == 0x0) continue;
+			
+			// Accept whole block when totally covered
+			if (a == 0xF && b == 0xF && c == 0xF && x + q <= clipright && y + q <= clipbottom)
+			{
+				if (span->Length != 0)
+				{
+					span->Length++;
+				}
+				else
+				{
+					span->X = x;
+					span->Y = y;
+					span->Length = 1;
+				}
+			}
+			else // Partially covered block
+			{
+				int CY1 = C1 + DX12 * y0 - DY12 * x0;
+				int CY2 = C2 + DX23 * y0 - DY23 * x0;
+				int CY3 = C3 + DX31 * y0 - DY31 * x0;
+				
+				uint64_t mask = 0;
+				
+				for (int iy = 0; iy < q; iy++)
+				{
+					int CX1 = CY1;
+					int CX2 = CY2;
+					int CX3 = CY3;
+					
+					for (int ix = x; ix < x + q; ix++)
+					{
+						bool covered = (CX1 > 0 && CX2 > 0 && CX3 > 0 && ix < clipright && iy < clipbottom);
+						mask <<= 1;
+						mask |= covered;
+						
+						CX1 -= FDY12;
+						CX2 -= FDY23;
+						CX3 -= FDY31;
+					}
+					
+					CY1 += FDX12;
+					CY2 += FDX23;
+					CY3 += FDX31;
+				}
+				
+				if (mask != 0xffffffffffffffffLL)
+				{
+					if (span->Length > 0)
+					{
+						curSpan++;
+						span++;
+						span->Length = 0;
+					}
+					
+					partial->X = x;
+					partial->Y = y;
+					partial->Mask0 = (uint32_t)(mask >> 32);
+					partial->Mask1 = (uint32_t)mask;
+					partial++;
+					curPartial++;
+				}
+				else if (span->Length != 0)
+				{
+					span->Length++;
+				}
+				else
+				{
+					span->X = x;
+					span->Y = y;
+					span->Length = 1;
+				}
+			}
+		}
+		
+		if (span->Length > 0)
+		{
+			curSpan++;
+			span++;
+			span->Length = 0;
+		}
+	}
+	
+	NumFullSpans = curSpan;
+	NumPartialBlocks = curPartial;
+}
+
+void ScreenTriangle::Draw(const TriDrawTriangleArgs *args)
+{
+	for (int i = 0; i < NumFullSpans; i++)
+	{
+		const auto &span = FullSpans[i];
+		
+		uint32_t *dest = (uint32_t*)args->dest + span.X + span.Y * args->pitch;
+		int pitch = args->pitch;
+		int width = span.Length * 8;
+		int height = 8;
+		for (int y = 0; y < height; y++)
+		{
+			for (int x = 0; x < width; x++)
+			{
+				dest[x] = 0xffffffff;
+			}
+		
+			dest += pitch;
+		}
+	}
+	
+	for (int i = 0; i < NumPartialBlocks; i++)
+	{
+		const auto &block = PartialBlocks[i];
+		
+		uint32_t *dest = (uint32_t*)args->dest + block.X + block.Y * args->pitch;
+		int pitch = args->pitch;
+		uint32_t mask0 = block.Mask0;
+		uint32_t mask1 = block.Mask1;
+		for (int y = 0; y < 4; y++)
+		{
+			for (int x = 0; x < 8; x++)
+			{
+				if (mask0 & (1<<31))
+					dest[x] = 0xffff0000;
+				mask0 <<= 1;
+			}
+			dest += pitch;
+		}
+		for (int y = 4; y < 8; y++)
+		{
+			for (int x = 0; x < 8; x++)
+			{
+				if (mask1 & (1<<31))
+					dest[x] = 0xffff0000;
+				mask1 <<= 1;
+			}
+			dest += pitch;
+		}
+	}
+}
+
+void ScreenTriangle::DrawFunc(const TriDrawTriangleArgs *args, WorkerThreadData *thread)
+{
+	static ScreenTriangle triangle[8];
+	
+	triangle[thread->core].Setup(args, thread);
+	triangle[thread->core].Draw(args);
+}
+
+ScreenTriangle::ScreenTriangle()
+{
+	FullSpansBuffer.resize(MAXWIDTH / 8 * (MAXHEIGHT / 8));
+	PartialBlocksBuffer.resize(MAXWIDTH / 8 * (MAXHEIGHT / 8));
+	FullSpans = FullSpansBuffer.data();
+	PartialBlocks = PartialBlocksBuffer.data();
+}

src/r_poly_triangle.h

@@ -264,3 +264,49 @@ public:
 	static TriVertex *GetVertices(int count);
 	static void Clear();
 };
+
+struct ScreenTriangleBlock
+{
+	float W;
+	float Varying[TriVertex::NumVarying];
+};
+
+struct ScreenTriangleFullSpan
+{
+	uint16_t X;
+	uint16_t Y;
+	uint32_t Length;
+};
+
+struct ScreenTrianglePartialBlock
+{
+	uint16_t X;
+	uint16_t Y;
+	uint32_t Mask0;
+	uint32_t Mask1;
+};
+
+class ScreenTriangle
+{
+public:
+	static void DrawFunc(const TriDrawTriangleArgs *args, WorkerThreadData *thread);
+	
+	ScreenTriangle();
+	
+	void Setup(const TriDrawTriangleArgs *args, WorkerThreadData *thread);
+	void Draw(const TriDrawTriangleArgs *args);
+
+	ScreenTriangleFullSpan *FullSpans;
+	ScreenTrianglePartialBlock *PartialBlocks;
+	int NumFullSpans;
+	int NumPartialBlocks;
+	int StartX;
+	int StartY;
+	ScreenTriangleBlock Start;
+	ScreenTriangleBlock GradientX;
+	ScreenTriangleBlock GradientY;
+	
+private:
+	std::vector<ScreenTriangleFullSpan> FullSpansBuffer;
+	std::vector<ScreenTrianglePartialBlock> PartialBlocksBuffer;
+};