rsx: Refactor rsx_utils header into smaller pieces

rpcs3/Emu/CMakeLists.txt

@@ -580,8 +580,8 @@ target_sources(rpcs3_emu PRIVATE
     RSX/RSXThread.cpp
     RSX/RSXZCULL.cpp
     RSX/rsx_methods.cpp
-    RSX/rsx_utils.cpp
     RSX/rsx_vertex_data.cpp
+    RSX/Utils/rsx_utils.cpp
 )
 
 if(NOT ANDROID AND NOT APPLE)

rpcs3/Emu/Cell/Modules/cellAvconfExt.cpp

@@ -2,7 +2,7 @@
 #include "Emu/system_config.h"
 #include "Emu/Cell/PPUModule.h"
 #include "Emu/IdManager.h"
-#include "Emu/RSX/rsx_utils.h"
+#include "Emu/RSX/Utils/rsx_utils.h"
 #include "Utilities/StrUtil.h"
 
 #include "cellMic.h"

rpcs3/Emu/Cell/Modules/cellVideoOut.cpp

@@ -3,7 +3,7 @@
 #include "Emu/Cell/ErrorCodes.h"
 #include "Emu/Cell/PPUModule.h"
 #include "Emu/IdManager.h"
-#include "Emu/RSX/rsx_utils.h"
+#include "Emu/RSX/Utils/rsx_utils.h"
 #include "Emu/RSX/RSXThread.h"
 
 #include "cellVideoOut.h"

rpcs3/Emu/Memory/vm_locking.h

@@ -1,7 +1,7 @@
 #pragma once
 
 #include "vm.h"
-#include "Emu/RSX/rsx_utils.h"
+#include "Emu/RSX/Utils/rsx_utils.h"
 
 class cpu_thread;
 class shared_mutex;

rpcs3/Emu/RSX/Common/TextureUtils.cpp

@@ -2,8 +2,9 @@
 #include "Emu/Memory/vm.h"
 #include "TextureUtils.h"
 #include "../RSXThread.h"
-#include "../rsx_utils.h"
-#include "../color_utils.h"
+#include "../Utils/rsx_utils.h"
+#include "../Utils/color_utils.hpp"
+#include "../Utils/image_utils.hpp"
 
 #include "3rdparty/bcdec/bcdec.hpp"
 

rpcs3/Emu/RSX/Common/TextureUtils.h

@@ -2,7 +2,7 @@
 
 #include "io_buffer.h"
 #include "simple_array.hpp"
-#include "../color_utils.h"
+#include "../Utils/color_utils.hpp"
 #include "../RSXTexture.h"
 
 #include <vector>

rpcs3/Emu/RSX/Common/surface_store.h

@@ -5,7 +5,7 @@
 #include "ranged_map.hpp"
 #include "surface_cache_dma.hpp"
 #include "../gcm_enums.h"
-#include "../rsx_utils.h"
+#include "../Utils/rsx_utils.h"
 #include <list>
 
 #include "util/asm.hpp"

rpcs3/Emu/RSX/Common/surface_utils.h

@@ -3,7 +3,7 @@
 #include "util/types.hpp"
 #include "Utilities/geometry.h"
 #include "TextureUtils.h"
-#include "../rsx_utils.h"
+#include "../Utils/rsx_utils.h"
 #include "Emu/Memory/vm.h"
 
 #define ENABLE_SURFACE_CACHE_DEBUG 0

rpcs3/Emu/RSX/Common/texture_cache.h

@@ -3,6 +3,7 @@
 #include "Emu/RSX/Common/simple_array.hpp"
 #include "Emu/RSX/Core/RSXContext.h"
 #include "Emu/RSX/RSXThread.h"
+#include "Emu/RSX/Utils/algorithm.hpp"
 
 #include "texture_cache_utils.h"
 #include "texture_cache_predictor.h"

rpcs3/Emu/RSX/Common/texture_cache_checker.h

@@ -2,7 +2,7 @@
 
 #ifdef TEXTURE_CACHE_DEBUG
 
-#include "../rsx_utils.h"
+#include "../Utils/rsx_utils.h"
 
 namespace rsx {
 

rpcs3/Emu/RSX/Common/texture_cache_helpers.h

@@ -1,8 +1,8 @@
 #pragma once
 
-#include "../rsx_utils.h"
 #include "simple_array.hpp"
 #include "TextureUtils.h"
+#include "../Utils/rsx_utils.h"
 
 namespace rsx
 {

rpcs3/Emu/RSX/Common/texture_cache_predictor.h

@@ -1,7 +1,7 @@
 #pragma once
 
-#include "../rsx_utils.h"
 #include "TextureUtils.h"
+#include "../Utils/rsx_utils.h"
 
 #include <unordered_map>
 

rpcs3/Emu/RSX/Core/RSXContext.cpp

@@ -1,5 +1,5 @@
 #include "stdafx.h"
-#include "Emu/RSX/rsx_utils.h"
+#include "Emu/RSX/Utils/algorithm.hpp"
 #include "RSXContext.h"
 
 namespace rsx

rpcs3/Emu/RSX/Core/RSXContext.h

@@ -3,7 +3,7 @@
 #include <util/types.hpp>
 #include "Emu/Cell/lv2/sys_rsx.h"
 #include "Emu/RSX/GCM.h"
-#include "Emu/RSX/rsx_utils.h"
+#include "Emu/RSX/Utils/rsx_utils.h"
 #include "RSXIOMap.hpp"
 
 namespace rsx

rpcs3/Emu/RSX/Core/RSXDisplay.cpp

@@ -2,7 +2,7 @@
 #include "RSXDisplay.h"
 
 #include "../Common/simple_array.hpp"
-#include "../rsx_utils.h"
+#include "../Utils/rsx_utils.h"
 
 namespace rsx
 {

rpcs3/Emu/RSX/GL/GLCompute.h

@@ -1,9 +1,11 @@
 #pragma once
 
-#include "Emu/IdManager.h"
 #include "GLHelpers.h"
 #include "glutils/program.h"
-#include "../rsx_utils.h"
+
+#include "Emu/IdManager.h"
+#include "Emu/RSX/Utils/algorithm.hpp"
+#include "Utilities/StrUtil.h"
 
 #include <unordered_map>
 

rpcs3/Emu/RSX/GL/GLProgramBuffer.h

@@ -3,7 +3,7 @@
 #include "GLFragmentProgram.h"
 #include "GLPipelineCompiler.h"
 #include "../Program/ProgramStateCache.h"
-#include "../rsx_utils.h"
+#include "../Utils/rsx_utils.h"
 
 struct GLTraits
 {

rpcs3/Emu/RSX/GL/GLRenderTargets.h

@@ -1,6 +1,6 @@
 #pragma once
 #include "../Common/surface_store.h"
-#include "../rsx_utils.h"
+#include "../Utils/rsx_utils.h"
 
 #include "glutils/fbo.h"
 

rpcs3/Emu/RSX/GL/GLTextureCache.cpp

@@ -2,6 +2,7 @@
 #include "GLTexture.h"
 #include "GLTextureCache.h"
 #include "../Common/BufferUtils.h"
+#include "../Utils/image_utils.hpp"
 
 #include "util/asm.hpp"
 

rpcs3/Emu/RSX/GL/upscalers/fsr1/fsr_pass.cpp

@@ -26,6 +26,8 @@
 #pragma clang diagnostic pop
 #endif
 
+#include "Emu/system_config.h"
+
 namespace gl
 {
 	namespace FidelityFX

rpcs3/Emu/RSX/Host/MM.cpp

@@ -2,6 +2,7 @@
 #include "MM.h"
 #include <Emu/RSX/Common/simple_array.hpp>
 #include <Emu/RSX/RSXOffload.h>
+#include <Emu/RSX/Utils/rsx_utils.h>
 
 #include <Emu/Memory/vm.h>
 #include <Emu/IdManager.h>

rpcs3/Emu/RSX/NV47/FW/draw_call.cpp

@@ -2,7 +2,7 @@
 #include "draw_call.hpp"
 
 #include "Emu/RSX/rsx_methods.h" // FIXME
-#include "Emu/RSX/rsx_utils.h"
+#include "Emu/RSX/Utils/rsx_utils.h"
 #include "Emu/RSX/RSXThread.h"
 #include "Emu/RSX/Common/BufferUtils.h"
 #include "Emu/RSX/NV47/HW/context.h"

rpcs3/Emu/RSX/NV47/HW/nv3089.cpp

@@ -5,6 +5,7 @@
 #include "Emu/RSX/Core/RSXReservationLock.hpp"
 #include "Emu/RSX/Common/tiled_dma_copy.hpp"
 #include "Emu/RSX/Host/MM.h"
+#include "Emu/RSX/Utils/image_utils.hpp"
 
 #include "context_accessors.define.h"
 

rpcs3/Emu/RSX/Program/GLSLCommon.cpp

@@ -3,7 +3,7 @@
 #include "GLSLCommon.h"
 #include "RSXFragmentProgram.h"
 
-#include "Emu/RSX/color_utils.h"
+#include "Emu/RSX/Utils/color_utils.hpp"
 #include "Emu/RSX/gcm_enums.h"
 #include "Utilities/StrFmt.h"
 

rpcs3/Emu/RSX/Program/GLSLCommon.h

@@ -4,7 +4,7 @@
 
 #include "GLSLTypes.h"
 #include "ShaderParam.h"
-#include "../color_utils.h"
+#include "../Utils/color_utils.hpp"
 
 struct RSXFragmentProgram;
 

rpcs3/Emu/RSX/RSXTexture.cpp

@@ -1,9 +1,10 @@
 #include "stdafx.h"
 #include "RSXTexture.h"
 
-#include "rsx_utils.h"
 #include "Common/TextureUtils.h"
 #include "Program/GLSLCommon.h"
+#include "Utils/algorithm.hpp"
+#include "Utils/rsx_utils.h"
 
 #include "Emu/system_config.h"
 #include "util/simd.hpp"

rpcs3/Emu/RSX/RSXTexture.h

@@ -1,6 +1,6 @@
 #pragma once
 #include "gcm_enums.h"
-#include "color_utils.h"
+#include "Utils/color_utils.hpp"
 
 namespace rsx
 {

rpcs3/Emu/RSX/RSXThread.h

@@ -6,13 +6,13 @@
 #include "RSXFIFO.h"
 #include "RSXOffload.h"
 #include "RSXZCULL.h"
-#include "rsx_utils.h"
 #include "Common/bitfield.hpp"
 #include "Common/profiling_timer.hpp"
 #include "Common/texture_cache_types.h"
 #include "Common/TextureUtils.h"
 #include "Program/RSXVertexProgram.h"
 #include "Program/RSXFragmentProgram.h"
+#include "Utils/rsx_utils.h"
 
 #include "Utilities/Thread.h"
 #include "Utilities/geometry.h"

rpcs3/Emu/RSX/RSXZCULL.h

@@ -6,7 +6,7 @@
 
 #include "Utilities/mutex.h"
 
-#include "rsx_utils.h"
+#include "Utils/rsx_utils.h"
 
 #include <vector>
 #include <stack>

rpcs3/Emu/RSX/Utils/algorithm.hpp

@@ -0,0 +1,54 @@
+#pragma once
+
+#include <cmath>
+#include <util/types.hpp>
+
+namespace rsx
+{
+	static constexpr u32 floor_log2(u32 value)
+	{
+		return value <= 1 ? 0 : std::countl_zero(value) ^ 31;
+	}
+
+	static constexpr u32 ceil_log2(u32 value)
+	{
+		return floor_log2(value) + u32{!!(value & (value - 1))};
+	}
+
+	static constexpr u32 next_pow2(u32 x)
+	{
+		if (x <= 2) return x;
+
+		return static_cast<u32>((1ULL << 32) >> std::countl_zero(x - 1));
+	}
+
+	static inline bool fcmp(float a, float b, float epsilon = 0.000001f)
+	{
+		return fabsf(a - b) < epsilon;
+	}
+
+	// General purpose alignment without power-of-2 constraint
+	template <typename T, typename U>
+	static inline T align2(T value, U alignment)
+	{
+		return ((value + alignment - 1) / alignment) * alignment;
+	}
+
+	// General purpose downward alignment without power-of-2 constraint
+	template <typename T, typename U>
+	static inline T align_down2(T value, U alignment)
+	{
+		return (value / alignment) * alignment;
+	}
+
+	// Copy memory in inverse direction from source
+	// Used to scale negatively x axis while transfering image data
+	template <typename Ts = u8, typename Td = Ts>
+	static void memcpy_r(void* dst, void* src, usz size)
+	{
+		for (u32 i = 0; i < size; i++)
+		{
+			*(static_cast<Td*>(dst) + i) = *(static_cast<Ts*>(src) - i);
+		}
+	}
+}

rpcs3/Emu/RSX/Utils/color_utils.hpp

@@ -2,7 +2,7 @@
 
 #include <util/types.hpp>
 #include <Utilities/geometry.h>
-#include "gcm_enums.h"
+#include "../gcm_enums.h"
 #include "Utilities/StrFmt.h"
 
 namespace rsx

rpcs3/Emu/RSX/Utils/image_utils.hpp

@@ -0,0 +1,188 @@
+#pragma once
+
+#include <util/types.hpp>
+#include "algorithm.hpp"
+
+extern "C"
+{
+#include <libavutil/pixfmt.h>
+}
+
+namespace rsx
+{
+	template <typename T>
+	void pad_texture(const void* input_pixels, void* output_pixels, u16 input_width, u16 input_height, u16 output_width, u16 /*output_height*/)
+	{
+		const T *src = static_cast<const T*>(input_pixels);
+		T *dst = static_cast<T*>(output_pixels);
+
+		for (u16 h = 0; h < input_height; ++h)
+		{
+			const u32 padded_pos = h * output_width;
+			const u32 pos = h * input_width;
+			for (u16 w = 0; w < input_width; ++w)
+			{
+				dst[padded_pos + w] = src[pos + w];
+			}
+		}
+	}
+
+	// Returns interleaved bits of X|Y|Z used as Z-order curve indices
+	static inline u32 calculate_z_index(u32 x, u32 y, u32 z, u32 log2_width, u32 log2_height, u32 log2_depth)
+	{
+		AUDIT(x < (1u << log2_width) && y < (1u << log2_height) && z < (1u << log2_depth));
+
+		// offset = X' | Y' | Z' which are x,y,z bits interleaved
+		u32 offset = 0;
+		u32 shift_count = 0;
+		do
+		{
+			if (log2_width)
+			{
+				offset |= (x & 0x1) << shift_count++;
+				x >>= 1;
+				log2_width--;
+			}
+
+			if (log2_height)
+			{
+				offset |= (y & 0x1) << shift_count++;
+				y >>= 1;
+				log2_height--;
+			}
+
+			if (log2_depth)
+			{
+				offset |= (z & 0x1) << shift_count++;
+				z >>= 1;
+				log2_depth--;
+			}
+		}
+		while (x | y | z);
+
+		return offset;
+	}
+
+	/*   Note: What the ps3 calls swizzling in this case is actually z-ordering / morton ordering of pixels
+	*       - Input can be swizzled or linear, bool flag handles conversion to and from
+	*       - It will handle any width and height that are a power of 2, square or non square
+	*    Restriction: It has mixed results if the height or width is not a power of 2
+	*    Restriction: Only works with 2D surfaces
+	*/
+	template <typename T, bool input_is_swizzled>
+	void convert_linear_swizzle(const void* input_pixels, void* output_pixels, u16 width, u16 height, u32 pitch)
+	{
+		const u32 log2width = ceil_log2(width);
+		const u32 log2height = ceil_log2(height);
+
+		// Max mask possible for square texture
+		u32 x_mask = 0x55555555;
+		u32 y_mask = 0xAAAAAAAA;
+
+		// We have to limit the masks to the lower of the two dimensions to allow for non-square textures
+		u32 limit_mask = (log2width < log2height) ? log2width : log2height;
+		// double the limit mask to account for bits in both x and y
+		limit_mask = 1 << (limit_mask << 1);
+
+		//x_mask, bits above limit are 1's for x-carry
+		x_mask = (x_mask | ~(limit_mask - 1));
+		//y_mask. bits above limit are 0'd, as we use a different method for y-carry over
+		y_mask = (y_mask & (limit_mask - 1));
+
+		u32 offs_y = 0;
+		u32 offs_x = 0;
+		u32 offs_x0 = 0; //total y-carry offset for x
+		const u32 y_incr = limit_mask;
+
+		// NOTE: The swizzled area is always a POT region and we must scan all of it to fill in the linear.
+		// It is assumed that there is no padding on the linear side for simplicity - backend upload/download will crop as needed.
+		// Remember, in cases of swizzling (and also tiled addressing) it is possible for tiled pixels to fall outside of their linear memory region.
+		const u32 pitch_in_blocks = pitch / sizeof(T);
+		u32 row_offset = 0;
+
+		if constexpr (!input_is_swizzled)
+		{
+			for (int y = 0; y < height; ++y, row_offset += pitch_in_blocks)
+			{
+				auto src = static_cast<const T*>(input_pixels) + row_offset;
+				auto dst = static_cast<T*>(output_pixels) + offs_y;
+				offs_x = offs_x0;
+
+				for (int x = 0; x < width; ++x)
+				{
+					dst[offs_x] = src[x];
+					offs_x = (offs_x - x_mask) & x_mask;
+				}
+
+				offs_y = (offs_y - y_mask) & y_mask;
+
+				if (offs_y == 0)
+				{
+					offs_x0 += y_incr;
+				}
+			}
+		}
+		else
+		{
+			for (int y = 0; y < height; ++y, row_offset += pitch_in_blocks)
+			{
+				auto src = static_cast<const T*>(input_pixels) + offs_y;
+				auto dst = static_cast<T*>(output_pixels) + row_offset;
+				offs_x = offs_x0;
+
+				for (int x = 0; x < width; ++x)
+				{
+					dst[x] = src[offs_x];
+					offs_x = (offs_x - x_mask) & x_mask;
+				}
+
+				offs_y = (offs_y - y_mask) & y_mask;
+
+				if (offs_y == 0)
+				{
+					offs_x0 += y_incr;
+				}
+			}
+		}
+	}
+
+	/**
+	 * Write swizzled data to linear memory with support for 3 dimensions
+	 * Z ordering is done in all 3 planes independently with a unit being a 2x2 block per-plane
+	 * A unit in 3d textures is a group of 2x2x2 texels advancing towards depth in units of 2x2x1 blocks
+	 * i.e 32 texels per "unit"
+	 */
+	template <typename T>
+	void convert_linear_swizzle_3d(const void* input_pixels, void* output_pixels, u16 width, u16 height, u16 depth)
+	{
+		if (depth == 1)
+		{
+			convert_linear_swizzle<T, true>(input_pixels, output_pixels, width, height, width * sizeof(T));
+			return;
+		}
+
+		auto src = static_cast<const T*>(input_pixels);
+		auto dst = static_cast<T*>(output_pixels);
+
+		const u32 log2_w = ceil_log2(width);
+		const u32 log2_h = ceil_log2(height);
+		const u32 log2_d = ceil_log2(depth);
+
+		for (u32 z = 0; z < depth; ++z)
+		{
+			for (u32 y = 0; y < height; ++y)
+			{
+				for (u32 x = 0; x < width; ++x)
+				{
+					*dst++ = src[calculate_z_index(x, y, z, log2_w, log2_h, log2_d)];
+				}
+			}
+		}
+	}
+
+	void convert_scale_image(u8 *dst, AVPixelFormat dst_format, int dst_width, int dst_height, int dst_pitch,
+		const u8 *src, AVPixelFormat src_format, int src_width, int src_height, int src_pitch, int src_slice_h, bool bilinear);
+
+	void clip_image(u8 *dst, const u8 *src, int clip_x, int clip_y, int clip_w, int clip_h, int bpp, int src_pitch, int dst_pitch);
+	void clip_image_may_overlap(u8 *dst, const u8 *src, int clip_x, int clip_y, int clip_w, int clip_h, int bpp, int src_pitch, int dst_pitch, u8* buffer);
+}

rpcs3/Emu/RSX/Utils/rsx_utils.cpp

@@ -1,6 +1,6 @@
 #include "stdafx.h"
 #include "rsx_utils.h"
-#include "rsx_methods.h"
+#include "Emu/RSX/rsx_methods.h"
 #include "Emu/Cell/Modules/cellVideoOut.h"
 
 #ifdef _MSC_VER

rpcs3/Emu/RSX/Utils/rsx_utils.h

@@ -1,14 +1,9 @@
 #pragma once
 
-#include "../system_config.h"
+#include "../../system_config.h"
 #include "Utilities/address_range.h"
 #include "Utilities/geometry.h"
-#include "gcm_enums.h"
-
-extern "C"
-{
-#include <libavutil/pixfmt.h>
-}
+#include "../gcm_enums.h"
 
 #define RSX_SURFACE_DIMENSION_IGNORED 1
 
@@ -228,45 +223,6 @@ namespace rsx
 		}
 	};
 
-	template <typename T>
-	void pad_texture(const void* input_pixels, void* output_pixels, u16 input_width, u16 input_height, u16 output_width, u16 /*output_height*/)
-	{
-		const T *src = static_cast<const T*>(input_pixels);
-		T *dst = static_cast<T*>(output_pixels);
-
-		for (u16 h = 0; h < input_height; ++h)
-		{
-			const u32 padded_pos = h * output_width;
-			const u32 pos = h * input_width;
-			for (u16 w = 0; w < input_width; ++w)
-			{
-				dst[padded_pos + w] = src[pos + w];
-			}
-		}
-	}
-
-	static constexpr u32 floor_log2(u32 value)
-	{
-		return value <= 1 ? 0 : std::countl_zero(value) ^ 31;
-	}
-
-	static constexpr u32 ceil_log2(u32 value)
-	{
-		return floor_log2(value) + u32{!!(value & (value - 1))};
-	}
-
-	static constexpr u32 next_pow2(u32 x)
-	{
-		if (x <= 2) return x;
-
-		return static_cast<u32>((1ULL << 32) >> std::countl_zero(x - 1));
-	}
-
-	static inline bool fcmp(float a, float b, float epsilon = 0.000001f)
-	{
-		return fabsf(a - b) < epsilon;
-	}
-
 	// Returns an ever-increasing tag value
 	static inline u64 get_shared_tag()
 	{
@@ -282,190 +238,6 @@ namespace rsx
 			CELL_GCM_LOCATION_MAIN;
 	}
 
-	// General purpose alignment without power-of-2 constraint
-	template <typename T, typename U>
-	static inline T align2(T value, U alignment)
-	{
-		return ((value + alignment - 1) / alignment) * alignment;
-	}
-
-	// General purpose downward alignment without power-of-2 constraint
-	template <typename T, typename U>
-	static inline T align_down2(T value, U alignment)
-	{
-		return (value / alignment) * alignment;
-	}
-
-	// Copy memory in inverse direction from source
-	// Used to scale negatively x axis while transfering image data
-	template <typename Ts = u8, typename Td = Ts>
-	static void memcpy_r(void* dst, void* src, usz size)
-	{
-		for (u32 i = 0; i < size; i++)
-		{
-			*(static_cast<Td*>(dst) + i) = *(static_cast<Ts*>(src) - i);
-		}
-	}
-
-	// Returns interleaved bits of X|Y|Z used as Z-order curve indices
-	static inline u32 calculate_z_index(u32 x, u32 y, u32 z, u32 log2_width, u32 log2_height, u32 log2_depth)
-	{
-		AUDIT(x < (1u << log2_width) && y < (1u << log2_height) && z < (1u << log2_depth));
-
-		// offset = X' | Y' | Z' which are x,y,z bits interleaved
-		u32 offset = 0;
-		u32 shift_count = 0;
-		do
-		{
-			if (log2_width)
-			{
-				offset |= (x & 0x1) << shift_count++;
-				x >>= 1;
-				log2_width--;
-			}
-
-			if (log2_height)
-			{
-				offset |= (y & 0x1) << shift_count++;
-				y >>= 1;
-				log2_height--;
-			}
-
-			if (log2_depth)
-			{
-				offset |= (z & 0x1) << shift_count++;
-				z >>= 1;
-				log2_depth--;
-			}
-		}
-		while (x | y | z);
-
-		return offset;
-	}
-
-	/*   Note: What the ps3 calls swizzling in this case is actually z-ordering / morton ordering of pixels
-	*       - Input can be swizzled or linear, bool flag handles conversion to and from
-	*       - It will handle any width and height that are a power of 2, square or non square
-	*    Restriction: It has mixed results if the height or width is not a power of 2
-	*    Restriction: Only works with 2D surfaces
-	*/
-	template <typename T, bool input_is_swizzled>
-	void convert_linear_swizzle(const void* input_pixels, void* output_pixels, u16 width, u16 height, u32 pitch)
-	{
-		const u32 log2width = ceil_log2(width);
-		const u32 log2height = ceil_log2(height);
-
-		// Max mask possible for square texture
-		u32 x_mask = 0x55555555;
-		u32 y_mask = 0xAAAAAAAA;
-
-		// We have to limit the masks to the lower of the two dimensions to allow for non-square textures
-		u32 limit_mask = (log2width < log2height) ? log2width : log2height;
-		// double the limit mask to account for bits in both x and y
-		limit_mask = 1 << (limit_mask << 1);
-
-		//x_mask, bits above limit are 1's for x-carry
-		x_mask = (x_mask | ~(limit_mask - 1));
-		//y_mask. bits above limit are 0'd, as we use a different method for y-carry over
-		y_mask = (y_mask & (limit_mask - 1));
-
-		u32 offs_y = 0;
-		u32 offs_x = 0;
-		u32 offs_x0 = 0; //total y-carry offset for x
-		const u32 y_incr = limit_mask;
-
-		// NOTE: The swizzled area is always a POT region and we must scan all of it to fill in the linear.
-		// It is assumed that there is no padding on the linear side for simplicity - backend upload/download will crop as needed.
-		// Remember, in cases of swizzling (and also tiled addressing) it is possible for tiled pixels to fall outside of their linear memory region.
-		const u32 pitch_in_blocks = pitch / sizeof(T);
-		u32 row_offset = 0;
-
-		if constexpr (!input_is_swizzled)
-		{
-			for (int y = 0; y < height; ++y, row_offset += pitch_in_blocks)
-			{
-				auto src = static_cast<const T*>(input_pixels) + row_offset;
-				auto dst = static_cast<T*>(output_pixels) + offs_y;
-				offs_x = offs_x0;
-
-				for (int x = 0; x < width; ++x)
-				{
-					dst[offs_x] = src[x];
-					offs_x = (offs_x - x_mask) & x_mask;
-				}
-
-				offs_y = (offs_y - y_mask) & y_mask;
-
-				if (offs_y == 0)
-				{
-					offs_x0 += y_incr;
-				}
-			}
-		}
-		else
-		{
-			for (int y = 0; y < height; ++y, row_offset += pitch_in_blocks)
-			{
-				auto src = static_cast<const T*>(input_pixels) + offs_y;
-				auto dst = static_cast<T*>(output_pixels) + row_offset;
-				offs_x = offs_x0;
-
-				for (int x = 0; x < width; ++x)
-				{
-					dst[x] = src[offs_x];
-					offs_x = (offs_x - x_mask) & x_mask;
-				}
-
-				offs_y = (offs_y - y_mask) & y_mask;
-
-				if (offs_y == 0)
-				{
-					offs_x0 += y_incr;
-				}
-			}
-		}
-	}
-
-	/**
-	 * Write swizzled data to linear memory with support for 3 dimensions
-	 * Z ordering is done in all 3 planes independently with a unit being a 2x2 block per-plane
-	 * A unit in 3d textures is a group of 2x2x2 texels advancing towards depth in units of 2x2x1 blocks
-	 * i.e 32 texels per "unit"
-	 */
-	template <typename T>
-	void convert_linear_swizzle_3d(const void* input_pixels, void* output_pixels, u16 width, u16 height, u16 depth)
-	{
-		if (depth == 1)
-		{
-			convert_linear_swizzle<T, true>(input_pixels, output_pixels, width, height, width * sizeof(T));
-			return;
-		}
-
-		auto src = static_cast<const T*>(input_pixels);
-		auto dst = static_cast<T*>(output_pixels);
-
-		const u32 log2_w = ceil_log2(width);
-		const u32 log2_h = ceil_log2(height);
-		const u32 log2_d = ceil_log2(depth);
-
-		for (u32 z = 0; z < depth; ++z)
-		{
-			for (u32 y = 0; y < height; ++y)
-			{
-				for (u32 x = 0; x < width; ++x)
-				{
-					*dst++ = src[calculate_z_index(x, y, z, log2_w, log2_h, log2_d)];
-				}
-			}
-		}
-	}
-
-	void convert_scale_image(u8 *dst, AVPixelFormat dst_format, int dst_width, int dst_height, int dst_pitch,
-		const u8 *src, AVPixelFormat src_format, int src_width, int src_height, int src_pitch, int src_slice_h, bool bilinear);
-
-	void clip_image(u8 *dst, const u8 *src, int clip_x, int clip_y, int clip_w, int clip_h, int bpp, int src_pitch, int dst_pitch);
-	void clip_image_may_overlap(u8 *dst, const u8 *src, int clip_x, int clip_y, int clip_w, int clip_h, int bpp, int src_pitch, int dst_pitch, u8* buffer);
-
 	std::array<float, 4> get_constant_blend_colors();
 
 	/**

rpcs3/Emu/RSX/VK/VKCompute.h

@@ -1,9 +1,10 @@
 #pragma once
-#include "Emu/RSX/VK/VKProgramPipeline.h"
+#include "VKProgramPipeline.h"
 #include "vkutils/descriptors.h"
 #include "vkutils/buffer_object.h"
 
 #include "Emu/IdManager.h"
+#include "Emu/RSX/Utils/algorithm.hpp"
 
 #include "Utilities/StrUtil.h"
 #include "util/asm.hpp"

rpcs3/Emu/RSX/VK/VKGSRenderTypes.hpp

@@ -6,7 +6,7 @@
 #include "VKResourceManager.h"
 
 #include "Emu/RSX/Common/simple_array.hpp"
-#include "Emu/RSX/rsx_utils.h"
+#include "Emu/RSX/Utils/rsx_utils.h"
 #include "Emu/RSX/rsx_cache.h"
 #include "Utilities/mutex.h"
 #include "util/asm.hpp"

rpcs3/Emu/RSX/VK/VKHelpers.h

@@ -6,7 +6,7 @@
 #include "VulkanAPI.h"
 #include "Utilities/geometry.h"
 #include "Emu/RSX/Common/TextureUtils.h"
-#include "Emu/RSX/rsx_utils.h"
+#include "Emu/RSX/Utils/rsx_utils.h"
 
 #define OCCLUSION_MAX_POOL_SIZE   DESCRIPTOR_MAX_DRAW_CALLS
 

rpcs3/Emu/RSX/VK/VKTexture.cpp

@@ -11,7 +11,7 @@
 #include "VKGSRender.h"
 
 #include "../GCM.h"
-#include "../rsx_utils.h"
+#include "../Utils/rsx_utils.h"
 #include "Utilities/deferred_op.hpp"
 
 #include "util/asm.hpp"

rpcs3/Emu/RSX/VK/VKTextureCache.h

@@ -8,6 +8,7 @@
 
 #include "../Common/texture_cache.h"
 #include "../Common/tiled_dma_copy.hpp"
+#include "../Utils/image_utils.hpp"
 
 #include <memory>
 #include <vector>

rpcs3/Emu/RSX/VK/vkutils/image_helpers.cpp

@@ -2,7 +2,7 @@
 #include "image_helpers.h"
 #include "image.h"
 #include "../VKRenderPass.h"
-#include "../../color_utils.h"
+#include "../../Utils/color_utils.hpp"
 
 namespace vk
 {

rpcs3/Emu/RSX/VK/vkutils/memory.cpp

@@ -1,15 +1,7 @@
 #include "device.h"
 #include "memory.h"
 
-namespace
-{
-	// Copied from rsx_utils.h. Move to a more convenient location
-	template<typename T, typename U>
-	static inline T align2(T value, U alignment)
-	{
-		return ((value + alignment - 1) / alignment) * alignment;
-	}
-}
+#include "Emu/RSX/Utils/algorithm.hpp"
 
 namespace vk
 {
@@ -236,7 +228,7 @@ namespace vk
 			for (const auto& memory_type_index : *request.memory_type)
 			{
 				mem_req.memoryTypeBits = 1u << memory_type_index;
-				mem_req.size = ::align2(request.size, request.alignment);
+				mem_req.size = rsx::align2(request.size, request.alignment);
 				mem_req.alignment = request.alignment;
 				create_info.memoryTypeBits = 1u << memory_type_index;
 				create_info.flags = m_allocation_flags;

rpcs3/Emu/RSX/VK/vkutils/memory.h

@@ -1,9 +1,9 @@
 #pragma once
 
-#include "../VulkanAPI.h"
-#include "../../rsx_utils.h"
 #include "shared.h"
 
+#include "Emu/RSX/Utils/rsx_utils.h"
+
 #ifdef __clang__
 #pragma clang diagnostic push
 #pragma clang diagnostic ignored "-Wnullability-completeness"

rpcs3/Emu/RSX/VK/vkutils/query_pool.hpp

@@ -1,8 +1,7 @@
 #pragma once
 
-#include "../VulkanAPI.h"
-#include "../../rsx_utils.h"
 #include "shared.h"
+#include "../../Utils/rsx_utils.h"
 
 namespace vk
 {

rpcs3/Emu/RSX/VK/vkutils/sampler.cpp

@@ -1,7 +1,7 @@
 #include "memory.h"
 #include "sampler.h"
-#include "../../color_utils.h"
-#include "../../rsx_utils.h"
+#include "../../Utils/color_utils.hpp"
+#include "../../Utils/algorithm.hpp"
 
 namespace vk
 {

rpcs3/Emu/RSX/rsx_decode.h

@@ -6,7 +6,7 @@
 #include <tuple>
 #include <algorithm>
 #include "gcm_enums.h"
-#include "rsx_utils.h"
+#include "Utils/rsx_utils.h"
 
 namespace rsx
 {

rpcs3/emucore.vcxproj

@@ -522,7 +522,7 @@
     <ClCompile Include="Emu\RSX\RSXFIFO.cpp" />
     <ClCompile Include="Emu\RSX\RSXOffload.cpp" />
     <ClCompile Include="Emu\RSX\rsx_methods.cpp" />
-    <ClCompile Include="Emu\RSX\rsx_utils.cpp" />
+    <ClCompile Include="Emu\RSX\Utils\rsx_utils.cpp" />
     <ClCompile Include="Crypto\aes.cpp">
       <PrecompiledHeader>NotUsing</PrecompiledHeader>
     </ClCompile>
@@ -682,7 +682,6 @@
     <ClInclude Include="Emu\NP\rpcn_config.h" />
     <ClInclude Include="Emu\NP\ip_address.h" />
     <ClInclude Include="Emu\perf_monitor.hpp" />
-    <ClInclude Include="Emu\RSX\color_utils.h" />
     <ClInclude Include="Emu\RSX\Common\aligned_malloc.hpp" />
     <ClInclude Include="Emu\RSX\Common\bitfield.hpp" />
     <ClInclude Include="Emu\RSX\Common\buffer_stream.hpp" />
@@ -1073,7 +1072,10 @@
     <ClInclude Include="Emu\Memory\vm_reservation.h" />
     <ClInclude Include="Emu\Memory\vm_var.h" />
     <ClInclude Include="Emu\RSX\rsx_methods.h" />
-    <ClInclude Include="Emu\RSX\rsx_utils.h" />
+    <ClInclude Include="Emu\RSX\Utils\algorithm.hpp" />
+    <ClInclude Include="Emu\RSX\Utils\color_utils.hpp" />
+    <ClInclude Include="Emu\RSX\Utils\image_utils.hpp" />
+    <ClInclude Include="Emu\RSX\Utils\rsx_utils.h" />
     <ClInclude Include="Emu\System.h" />
     <ClInclude Include="Emu\perf_meter.hpp" />
     <ClInclude Include="Emu\GDB.h" />

rpcs3/emucore.vcxproj.filters

@@ -1333,7 +1333,7 @@
     <ClCompile Include="Emu\RSX\gcm_printing.cpp">
       <Filter>Emu\GPU\RSX\Utils</Filter>
     </ClCompile>
-    <ClCompile Include="Emu\RSX\rsx_utils.cpp">
+    <ClCompile Include="Emu\RSX\Utils\rsx_utils.cpp">
       <Filter>Emu\GPU\RSX\Utils</Filter>
     </ClCompile>
     <ClCompile Include="Emu\RSX\rsx_vertex_data.cpp">
@@ -2800,9 +2800,6 @@
     <ClInclude Include="Emu\RSX\gcm_printing.h">
       <Filter>Emu\GPU\RSX\Utils</Filter>
     </ClInclude>
-    <ClInclude Include="Emu\RSX\rsx_utils.h">
-      <Filter>Emu\GPU\RSX\Utils</Filter>
-    </ClInclude>
     <ClInclude Include="Emu\RSX\rsx_vertex_data.h">
       <Filter>Emu\GPU\RSX\Host Mini-Driver</Filter>
     </ClInclude>
@@ -2830,7 +2827,16 @@
     <ClInclude Include="Emu\RSX\display.h">
       <Filter>Emu\GPU\RSX\Utils</Filter>
     </ClInclude>
-    <ClInclude Include="Emu\RSX\color_utils.h">
+    <ClInclude Include="Emu\RSX\Utils\algorithm.hpp">
+      <Filter>Emu\GPU\RSX\Utils</Filter>
+    </ClInclude>
+    <ClInclude Include="Emu\RSX\Utils\color_utils.hpp">
+      <Filter>Emu\GPU\RSX\Utils</Filter>
+    </ClInclude>
+    <ClInclude Include="Emu\RSX\Utils\image_utils.hpp">
+      <Filter>Emu\GPU\RSX\Utils</Filter>
+    </ClInclude>
+    <ClInclude Include="Emu\RSX\Utils\rsx_utils.h">
       <Filter>Emu\GPU\RSX\Utils</Filter>
     </ClInclude>
     <ClInclude Include="Emu\RSX\Host\RSXDMAWriter.h">

rpcs3/rpcs3qt/gs_frame.cpp

@@ -16,7 +16,8 @@
 #include "Emu/Cell/Modules/cellAudio.h"
 #include "Emu/Cell/lv2/sys_rsxaudio.h"
 #include "Emu/RSX/RSXThread.h"
-#include "Emu/RSX/rsx_utils.h"
+#include "Emu/RSX/Utils/image_utils.hpp"
+#include "Emu/RSX/Utils/rsx_utils.h"
 #include "Emu/RSX/Overlays/overlay_message.h"
 #include "Emu/Io/interception.h"
 #include "Emu/Io/recording_config.h"

rpcs3/rpcs3qt/memory_viewer_panel.cpp

@@ -7,7 +7,7 @@
 #include "Emu/Cell/SPUThread.h"
 #include "Emu/CPU/CPUDisAsm.h"
 #include "Emu/RSX/RSXThread.h"
-#include "Emu/RSX/rsx_utils.h"
+#include "Emu/RSX/Utils/rsx_utils.h"
 #include "Emu/IdManager.h"
 #include "Emu/System.h"
 #include <QVBoxLayout>