diff --git a/Source/Core/Common/MathUtil.h b/Source/Core/Common/MathUtil.h index da12822f5d..14b9309fc6 100644 --- a/Source/Core/Common/MathUtil.h +++ b/Source/Core/Common/MathUtil.h @@ -150,7 +150,7 @@ float MathFloatVectorSum(const std::vector&); #define ROUND_DOWN(x, a) ((x) & ~((a) - 1)) // Rounds down. 0 -> undefined -inline u64 Log2(u64 val) +inline int Log2(u64 val) { #if defined(__GNUC__) return 63 - __builtin_clzll(val); @@ -161,7 +161,7 @@ inline u64 Log2(u64 val) return result; #else - u64 result = -1; + int result = -1; while (val != 0) { val >>= 1; diff --git a/Source/Core/DolphinWX/GameListCtrl.cpp b/Source/Core/DolphinWX/GameListCtrl.cpp index bd420f8156..6029c741a4 100644 --- a/Source/Core/DolphinWX/GameListCtrl.cpp +++ b/Source/Core/DolphinWX/GameListCtrl.cpp @@ -413,12 +413,12 @@ wxString NiceSizeFormat(u64 _size) { const char* const unit_symbols[] = {"B", "KiB", "MiB", "GiB", "TiB", "PiB", "EiB", "ZiB", "YiB"}; - auto const unit = Log2(std::max(_size, 1)) / 10; - auto const unit_size = (1 << (unit * 10)); + const u64 unit = Log2(std::max(_size, 1)) / 10; + const u64 unit_size = (1 << (unit * 10)); // ugly rounding integer math - auto const value = (_size + unit_size / 2) / unit_size; - auto const frac = (_size % unit_size * 10 + unit_size / 2) / unit_size % 10; + const u64 value = (_size + unit_size / 2) / unit_size; + const u64 frac = (_size % unit_size * 10 + unit_size / 2) / unit_size % 10; return StrToWxStr(StringFromFormat("%" PRIu64 ".%" PRIu64 " %s", value, frac, unit_symbols[unit])); } diff --git a/Source/Core/VideoBackends/OGL/ProgramShaderCache.cpp b/Source/Core/VideoBackends/OGL/ProgramShaderCache.cpp index 79826860a7..0de9034f51 100644 --- a/Source/Core/VideoBackends/OGL/ProgramShaderCache.cpp +++ b/Source/Core/VideoBackends/OGL/ProgramShaderCache.cpp @@ -452,12 +452,6 @@ void ProgramShaderCache::CreateHeader ( void ) // Precision defines for GLSLES3 "%s\n" - "\n"// A few required defines and ones that will make our lives a lot easier - "#define ATTRIN in\n" - "#define ATTROUT out\n" - "#define VARYIN %s\n" - "#define VARYOUT %s\n" - // Silly differences "#define float2 vec2\n" "#define float3 vec3\n" @@ -470,8 +464,9 @@ void ProgramShaderCache::CreateHeader ( void ) "#define frac fract\n" "#define lerp mix\n" - // Terrible hack, look at DriverDetails.h - "%s\n" + // Terrible hacks, look at DriverDetails.h + "%s\n" // replace textureSize as constant + "%s\n" // wipe out all centroid usages , v==GLSLES3 ? "#version 300 es" : v==GLSL_130 ? "#version 130" : v==GLSL_140 ? "#version 140" : "#version 150" , v>1, ts.y-uv.y-1), 0);\n" " float y = mix(c0.b, c0.r, (uv.x & 1) == 1);\n" " float yComp = 1.164 * (y - 0.0625);\n" " float uComp = c0.g - 0.5;\n" diff --git a/Source/Core/VideoCommon/PixelShaderGen.cpp b/Source/Core/VideoCommon/PixelShaderGen.cpp index 66965139b8..e51294e55c 100644 --- a/Source/Core/VideoCommon/PixelShaderGen.cpp +++ b/Source/Core/VideoCommon/PixelShaderGen.cpp @@ -311,19 +311,19 @@ static inline void GeneratePixelShader(T& out, DSTALPHA_MODE dstAlphaMode, API_T if (per_pixel_depth) out.Write("#define depth gl_FragDepth\n"); - out.Write("VARYIN float4 colors_02;\n"); - out.Write("VARYIN float4 colors_12;\n"); + out.Write("centroid in float4 colors_02;\n"); + out.Write("centroid in float4 colors_12;\n"); // compute window position if needed because binding semantic WPOS is not widely supported // Let's set up attributes for (unsigned int i = 0; i < xfregs.numTexGen.numTexGens; ++i) { - out.Write("VARYIN float3 uv%d_2;\n", i); + out.Write("centroid in float3 uv%d_2;\n", i); } - out.Write("VARYIN float4 clipPos_2;\n"); + out.Write("centroid in float4 clipPos_2;\n"); if (g_ActiveConfig.bEnablePixelLighting && g_ActiveConfig.backend_info.bSupportsPixelLighting) { - out.Write("VARYIN float4 Normal_2;\n"); + out.Write("centroid in float4 Normal_2;\n"); } if (forced_early_z) diff --git a/Source/Core/VideoCommon/TextureConversionShader.cpp b/Source/Core/VideoCommon/TextureConversionShader.cpp index a7db33418f..47f9df21ba 100644 --- a/Source/Core/VideoCommon/TextureConversionShader.cpp +++ b/Source/Core/VideoCommon/TextureConversionShader.cpp @@ -14,6 +14,7 @@ #include "VideoCommon/TextureConversionShader.h" #include "VideoCommon/TextureDecoder.h" #include "VideoCommon/VideoConfig.h" +#include "Common/MathUtil.h" #define WRITE p+=sprintf @@ -65,8 +66,7 @@ void WriteSwizzler(char*& p, u32 format, API_TYPE ApiType) int blkW = TexDecoder_GetBlockWidthInTexels(format); int blkH = TexDecoder_GetBlockHeightInTexels(format); int samples = GetEncodedSampleCount(format); - // 32 bit textures (RGBA8 and Z24) are store in 2 cache line increments - int factor = samples == 1 ? 2 : 1; + if (ApiType == API_OPENGL) { WRITE(p, "#define samp0 samp9\n"); @@ -87,37 +87,41 @@ void WriteSwizzler(char*& p, u32 format, API_TYPE ApiType) WRITE(p, "{\n" " int2 sampleUv;\n" " int2 uv1 = int2(gl_FragCoord.xy);\n" - " float2 uv0 = float2(0.0, 0.0);\n" ); - WRITE(p, " uv1.x = uv1.x * %d;\n", samples); + WRITE(p, " int y_block_position = uv1.y & %d;\n", ~(blkH - 1)); + WRITE(p, " int y_offset_in_block = uv1.y & %d;\n", blkH - 1); + WRITE(p, " int x_virtual_position = (uv1.x << %d) + y_offset_in_block * position.z;\n", Log2(samples)); + WRITE(p, " int x_block_position = (x_virtual_position >> %d) & %d;\n", Log2(blkH), ~(blkW - 1)); + if (samples == 1) + { + // 32 bit textures (RGBA8 and Z24) are stored in 2 cache line increments + WRITE(p, " bool first = 0 == (x_virtual_position & %d);\n", 8 * samples); // first cache line, used in the encoders + WRITE(p, " x_virtual_position = x_virtual_position << 1;\n"); + } + WRITE(p, " int x_offset_in_block = x_virtual_position & %d;\n", blkW - 1); + WRITE(p, " int y_offset = (x_virtual_position >> %d) & %d;\n", Log2(blkW), blkH - 1); - WRITE(p, " int yl = uv1.y / %d;\n", blkH); - WRITE(p, " int yb = yl * %d;\n", blkH); - WRITE(p, " int yoff = uv1.y - yb;\n"); - WRITE(p, " int xp = uv1.x + yoff * position.z;\n"); - WRITE(p, " int xel = xp / %d;\n", samples == 1 ? factor : blkW); - WRITE(p, " int xb = xel / %d;\n", blkH); - WRITE(p, " int xoff = xel - xb * %d;\n", blkH); - WRITE(p, " int xl = uv1.x * %d / %d;\n", factor, blkW); - WRITE(p, " int xib = uv1.x * %d - xl * %d;\n", factor, blkW); - WRITE(p, " int halfxb = xb / %d;\n", factor); + WRITE(p, " sampleUv.x = x_offset_in_block + x_block_position;\n"); + WRITE(p, " sampleUv.y = y_block_position + y_offset;\n"); - WRITE(p, " sampleUv.x = xib + halfxb * %d;\n", blkW); - WRITE(p, " sampleUv.y = yb + xoff;\n"); + WRITE(p, " float2 uv0 = float2(sampleUv);\n"); // sampleUv is the sample position in (int)gx_coords + WRITE(p, " uv0 += float2(0.5, 0.5);\n"); // move to center of pixel + WRITE(p, " uv0 *= float(position.w);\n"); // scale by two if needed (also move to pixel borders so that linear filtering will average adjacent pixel) + WRITE(p, " uv0 += float2(position.xy);\n"); // move to copied rect + WRITE(p, " uv0 /= float2(%d, %d);\n", EFB_WIDTH, EFB_HEIGHT); // normalize to [0:1] + if (ApiType == API_OPENGL) // ogl has to flip up and down + { + WRITE(p, " uv0.y = 1.0-uv0.y;\n"); + } + + WRITE(p, " float sample_offset = position.w / float(%d);\n", EFB_WIDTH); } void WriteSampleColor(char*& p, const char* colorComp, const char* dest, int xoffset, API_TYPE ApiType) { - WRITE(p, // sampleUv is the sample position in (int)gx_coords - "uv0 = float2(sampleUv + int2(%d, 0));\n" // pixel offset (if more than one pixel is samped) - "uv0 += float2(0.5, 0.5);\n" // move to center of pixel - "uv0 *= float(position.w);\n" // scale by two if needed (this will move to pixels border to filter linear) - "uv0 += float2(position.xy);\n" // move to copyed rect - "uv0 /= float2(%d, %d);\n" // normlize to [0:1] - "uv0.y = 1.0-uv0.y;\n" // ogl foo (disable this line for d3d) - "%s = texture(samp0, uv0).%s;\n", - xoffset, EFB_WIDTH, EFB_HEIGHT, dest, colorComp + WRITE(p, " %s = texture(samp0, uv0 + float2(%d, 0) * sample_offset).%s;\n", + dest, xoffset, colorComp ); } @@ -373,8 +377,6 @@ void WriteRGBA8Encoder(char* p,API_TYPE ApiType) { WriteSwizzler(p, GX_TF_RGBA8, ApiType); - WRITE(p, " bool first = xb == (halfxb * 2);\n"); - WRITE(p, " float4 texSample;\n"); WRITE(p, " float4 color0;\n"); WRITE(p, " float4 color1;\n"); @@ -563,8 +565,6 @@ void WriteZ24Encoder(char* p, API_TYPE ApiType) { WriteSwizzler(p, GX_TF_Z24X8, ApiType); - WRITE(p, " bool first = xb == (halfxb * 2);\n"); - WRITE(p, " float depth0;\n"); WRITE(p, " float depth1;\n"); WRITE(p, " float3 expanded0;\n"); diff --git a/Source/Core/VideoCommon/VertexShaderGen.cpp b/Source/Core/VideoCommon/VertexShaderGen.cpp index 67cb4c1231..3e2e02eecd 100644 --- a/Source/Core/VideoCommon/VertexShaderGen.cpp +++ b/Source/Core/VideoCommon/VertexShaderGen.cpp @@ -106,26 +106,26 @@ static inline void GenerateVertexShader(T& out, u32 components, API_TYPE api_typ if(api_type == API_OPENGL) { - out.Write("ATTRIN float4 rawpos; // ATTR%d,\n", SHADER_POSITION_ATTRIB); + out.Write("in float4 rawpos; // ATTR%d,\n", SHADER_POSITION_ATTRIB); if (components & VB_HAS_POSMTXIDX) - out.Write("ATTRIN float fposmtx; // ATTR%d,\n", SHADER_POSMTX_ATTRIB); + out.Write("in float fposmtx; // ATTR%d,\n", SHADER_POSMTX_ATTRIB); if (components & VB_HAS_NRM0) - out.Write("ATTRIN float3 rawnorm0; // ATTR%d,\n", SHADER_NORM0_ATTRIB); + out.Write("in float3 rawnorm0; // ATTR%d,\n", SHADER_NORM0_ATTRIB); if (components & VB_HAS_NRM1) - out.Write("ATTRIN float3 rawnorm1; // ATTR%d,\n", SHADER_NORM1_ATTRIB); + out.Write("in float3 rawnorm1; // ATTR%d,\n", SHADER_NORM1_ATTRIB); if (components & VB_HAS_NRM2) - out.Write("ATTRIN float3 rawnorm2; // ATTR%d,\n", SHADER_NORM2_ATTRIB); + out.Write("in float3 rawnorm2; // ATTR%d,\n", SHADER_NORM2_ATTRIB); if (components & VB_HAS_COL0) - out.Write("ATTRIN float4 color0; // ATTR%d,\n", SHADER_COLOR0_ATTRIB); + out.Write("in float4 color0; // ATTR%d,\n", SHADER_COLOR0_ATTRIB); if (components & VB_HAS_COL1) - out.Write("ATTRIN float4 color1; // ATTR%d,\n", SHADER_COLOR1_ATTRIB); + out.Write("in float4 color1; // ATTR%d,\n", SHADER_COLOR1_ATTRIB); for (int i = 0; i < 8; ++i) { u32 hastexmtx = (components & (VB_HAS_TEXMTXIDX0<