From 414b0741c445a7960f9ad1ee4a5672f8af4760db Mon Sep 17 00:00:00 2001 From: Subv Date: Sun, 8 Mar 2015 12:05:17 -0500 Subject: GPU: Implemented more depth buffer formats. This fixes the horizontal lines in Picross E, Cubic Ninja, Cave Story 3D and possibly others --- src/video_core/color.h | 57 ++++++++++++++++++++++++++++++++++++++++ src/video_core/pica.h | 7 +++++ src/video_core/rasterizer.cpp | 60 ++++++++++++++++++++++++++++++++++++------- 3 files changed, 115 insertions(+), 9 deletions(-) (limited to 'src/video_core') diff --git a/src/video_core/color.h b/src/video_core/color.h index 35da901f2..35b56efc0 100644 --- a/src/video_core/color.h +++ b/src/video_core/color.h @@ -100,6 +100,33 @@ inline const Math::Vec4 DecodeRGBA4(const u8* bytes) { Convert4To8((pixel >> 4) & 0xF), Convert4To8(pixel & 0xF) }; } +/** + * Decode a depth value stored in D16 format + * @param bytes Pointer to encoded source value + * @return Depth value as an u32 + */ +inline const u32 DecodeD16(const u8* bytes) { + return *reinterpret_cast(bytes); +} + +/** + * Decode a depth value stored in D24 format + * @param bytes Pointer to encoded source value + * @return Depth value as an u32 + */ +inline const u32 DecodeD24(const u8* bytes) { + return (bytes[2] << 16) | (bytes[1] << 8) | bytes[0]; +} + +/** + * Decode a depth value and a stencil value stored in D24S8 format + * @param bytes Pointer to encoded source values + * @return Resulting values stored as a Math::Vec2 + */ +inline const Math::Vec2 DecodeD24S8(const u8* bytes) { + return { (bytes[2] << 16) | (bytes[1] << 8) | bytes[0], bytes[3] }; +} + /** * Encode a color as RGBA8 format * @param color Source color to encode @@ -153,4 +180,34 @@ inline void EncodeRGBA4(const Math::Vec4& color, u8* bytes) { (Convert8To4(color.g()) << 8) | (Convert8To4(color.b()) << 4) | Convert8To4(color.a()); } +/** + * Encode a depth value as D16 format + * @param value Source depth value to encode + * @param bytes Pointer where to store the encoded value + */ +inline void EncodeD16(u32 value, u8* bytes) { + *reinterpret_cast(bytes) = value & 0xFFFF; +} + +/** + * Encode a depth value as D24 format + * @param value Source depth value to encode + * @param bytes Pointer where to store the encoded value + */ +inline void EncodeD24(u32 value, u8* bytes) { + bytes[0] = value & 0xFF; + bytes[1] = (value >> 8) & 0xFF; + bytes[2] = (value >> 16) & 0xFF; +} + +/** + * Encode depth and stencil values as D24S8 format + * @param depth Source depth values to encode + * @param stencil Source stencil value to encode + * @param bytes Pointer where to store the encoded value + */ +inline void EncodeD24S8(u32 depth, u8 stencil, u8* bytes) { + *reinterpret_cast(bytes) = (stencil << 24) | depth; +} + } // namespace diff --git a/src/video_core/pica.h b/src/video_core/pica.h index b14de9278..6549693f5 100644 --- a/src/video_core/pica.h +++ b/src/video_core/pica.h @@ -418,6 +418,13 @@ struct Regs { RGBA4 = 4, }; + enum DepthFormat : u32 { + D16 = 0, + + D24 = 2, + D24S8 = 3 + }; + INSERT_PADDING_WORDS(0x6); u32 depth_format; diff --git a/src/video_core/rasterizer.cpp b/src/video_core/rasterizer.cpp index 5861c1926..dc32128c6 100644 --- a/src/video_core/rasterizer.cpp +++ b/src/video_core/rasterizer.cpp @@ -91,7 +91,7 @@ static const Math::Vec4 GetPixel(int x, int y) { } return {}; - } +} static u32 GetDepth(int x, int y) { const PAddr addr = registers.framebuffer.GetDepthBufferPhysicalAddress(); @@ -100,23 +100,65 @@ static u32 GetDepth(int x, int y) { y = (registers.framebuffer.height - y); const u32 coarse_y = y & ~7; - u32 stride = registers.framebuffer.width * 2; - // Assuming 16-bit depth buffer format until actual format handling is implemented - return *(u16*)(depth_buffer + VideoCore::GetMortonOffset(x, y, 2) + coarse_y * stride); + switch (registers.framebuffer.depth_format) { + case registers.framebuffer.D16: + { + u32 stride = registers.framebuffer.width * 2; + return Color::DecodeD16(depth_buffer + VideoCore::GetMortonOffset(x, y, 2) + coarse_y * stride); + } + case registers.framebuffer.D24: + { + u32 stride = registers.framebuffer.width * 3; + u8* address = depth_buffer + VideoCore::GetMortonOffset(x, y, 3) + coarse_y * stride; + return Color::DecodeD24(address); + } + case registers.framebuffer.D24S8: + { + u32 stride = registers.framebuffer.width * 4; + return Color::DecodeD24S8(depth_buffer + VideoCore::GetMortonOffset(x, y, 4) + coarse_y * stride).x; + } + default: + LOG_CRITICAL(HW_GPU, "Unimplemented depth format %u", registers.framebuffer.depth_format); + UNIMPLEMENTED(); + return 0; + } } -static void SetDepth(int x, int y, u16 value) { +static void SetDepth(int x, int y, u32 value) { const PAddr addr = registers.framebuffer.GetDepthBufferPhysicalAddress(); u8* depth_buffer = Memory::GetPointer(PAddrToVAddr(addr)); y = (registers.framebuffer.height - y); const u32 coarse_y = y & ~7; - u32 stride = registers.framebuffer.width * 2; - // Assuming 16-bit depth buffer format until actual format handling is implemented - *(u16*)(depth_buffer + VideoCore::GetMortonOffset(x, y, 2) + coarse_y * stride) = value; + switch (registers.framebuffer.depth_format) { + case registers.framebuffer.D16: + { + u32 stride = registers.framebuffer.width * 2; + Color::EncodeD16(value, depth_buffer + VideoCore::GetMortonOffset(x, y, 2) + coarse_y * stride); + break; + } + case registers.framebuffer.D24: + { + u32 stride = registers.framebuffer.width * 3; + u8* address = depth_buffer + VideoCore::GetMortonOffset(x, y, 3) + coarse_y * stride; + Color::EncodeD24(value, address); + break; + } + case registers.framebuffer.D24S8: + { + u32 stride = registers.framebuffer.width * 4; + // TODO(Subv): Implement the stencil buffer + Color::EncodeD24S8(value, 0, depth_buffer + VideoCore::GetMortonOffset(x, y, 4) + coarse_y * stride); + break; + } + default: + LOG_CRITICAL(HW_GPU, "Unimplemented depth format %u", registers.framebuffer.depth_format); + UNIMPLEMENTED(); + break; + } } // NOTE: Assuming that rasterizer coordinates are 12.4 fixed-point values @@ -595,7 +637,7 @@ static void ProcessTriangleInternal(const VertexShader::OutputVertex& v0, u16 z = (u16)((v0.screenpos[2].ToFloat32() * w0 + v1.screenpos[2].ToFloat32() * w1 + v2.screenpos[2].ToFloat32() * w2) * 65535.f / wsum); - u16 ref_z = GetDepth(x >> 4, y >> 4); + u32 ref_z = GetDepth(x >> 4, y >> 4); bool pass = false; -- cgit v1.2.3 From 1248e291f0c9a29734b0f5175df8fa675cce930c Mon Sep 17 00:00:00 2001 From: Subv Date: Sun, 8 Mar 2015 21:31:15 -0500 Subject: GPU: Added the stencil test structure to the Pica Regs struct. --- src/video_core/color.h | 18 +++++++-------- src/video_core/pica.h | 43 ++++++++++++++++++++++++++-------- src/video_core/rasterizer.cpp | 54 ++++++++++++++++++------------------------- 3 files changed, 65 insertions(+), 50 deletions(-) (limited to 'src/video_core') diff --git a/src/video_core/color.h b/src/video_core/color.h index 35b56efc0..14ade74f2 100644 --- a/src/video_core/color.h +++ b/src/video_core/color.h @@ -105,7 +105,7 @@ inline const Math::Vec4 DecodeRGBA4(const u8* bytes) { * @param bytes Pointer to encoded source value * @return Depth value as an u32 */ -inline const u32 DecodeD16(const u8* bytes) { +inline u32 DecodeD16(const u8* bytes) { return *reinterpret_cast(bytes); } @@ -114,7 +114,7 @@ inline const u32 DecodeD16(const u8* bytes) { * @param bytes Pointer to encoded source value * @return Depth value as an u32 */ -inline const u32 DecodeD24(const u8* bytes) { +inline u32 DecodeD24(const u8* bytes) { return (bytes[2] << 16) | (bytes[1] << 8) | bytes[0]; } @@ -181,8 +181,8 @@ inline void EncodeRGBA4(const Math::Vec4& color, u8* bytes) { } /** - * Encode a depth value as D16 format - * @param value Source depth value to encode + * Encode a 16 bit depth value as D16 format + * @param value 16 bit source depth value to encode * @param bytes Pointer where to store the encoded value */ inline void EncodeD16(u32 value, u8* bytes) { @@ -190,8 +190,8 @@ inline void EncodeD16(u32 value, u8* bytes) { } /** - * Encode a depth value as D24 format - * @param value Source depth value to encode + * Encode a 24 bit depth value as D24 format + * @param value 24 bit source depth value to encode * @param bytes Pointer where to store the encoded value */ inline void EncodeD24(u32 value, u8* bytes) { @@ -201,9 +201,9 @@ inline void EncodeD24(u32 value, u8* bytes) { } /** - * Encode depth and stencil values as D24S8 format - * @param depth Source depth values to encode - * @param stencil Source stencil value to encode + * Encode a 24 bit depth and 8 bit stencil values as D24S8 format + * @param depth 24 bit source depth value to encode + * @param stencil 8 bit source stencil value to encode * @param bytes Pointer where to store the encoded value */ inline void EncodeD24S8(u32 depth, u8 stencil, u8* bytes) { diff --git a/src/video_core/pica.h b/src/video_core/pica.h index 6549693f5..fe20cd77d 100644 --- a/src/video_core/pica.h +++ b/src/video_core/pica.h @@ -393,7 +393,15 @@ struct Regs { BitField< 8, 8, u32> ref; } alpha_test; - INSERT_PADDING_WORDS(0x2); + union { + BitField< 0, 1, u32> stencil_test_enable; + BitField< 4, 3, CompareFunc> stencil_test_func; + BitField< 8, 8, u32> stencil_replacement_value; + BitField<16, 8, u32> stencil_reference_value; + BitField<24, 8, u32> stencil_mask; + } stencil_test; + + INSERT_PADDING_WORDS(0x1); union { BitField< 0, 1, u32> depth_test_enable; @@ -408,6 +416,30 @@ struct Regs { INSERT_PADDING_WORDS(0x8); } output_merger; + enum DepthFormat : u32 { + D16 = 0, + + D24 = 2, + D24S8 = 3 + }; + + /* + * Returns the number of bytes in the specified depth format + */ + static u32 BytesPerDepthPixel(DepthFormat format) { + switch (format) { + case DepthFormat::D16: + return 2; + case DepthFormat::D24: + return 3; + case DepthFormat::D24S8: + return 4; + default: + LOG_CRITICAL(HW_GPU, "Unknown depth format %u", format); + UNIMPLEMENTED(); + } + } + struct { // Components are laid out in reverse byte order, most significant bits first. enum ColorFormat : u32 { @@ -418,16 +450,9 @@ struct Regs { RGBA4 = 4, }; - enum DepthFormat : u32 { - D16 = 0, - - D24 = 2, - D24S8 = 3 - }; - INSERT_PADDING_WORDS(0x6); - u32 depth_format; + DepthFormat depth_format; BitField<16, 3, u32> color_format; INSERT_PADDING_WORDS(0x4); diff --git a/src/video_core/rasterizer.cpp b/src/video_core/rasterizer.cpp index dc32128c6..dd46f0ec3 100644 --- a/src/video_core/rasterizer.cpp +++ b/src/video_core/rasterizer.cpp @@ -100,24 +100,19 @@ static u32 GetDepth(int x, int y) { y = (registers.framebuffer.height - y); const u32 coarse_y = y & ~7; + u32 bytes_per_pixel = Pica::Regs::BytesPerDepthPixel(registers.framebuffer.depth_format); + u32 stride = registers.framebuffer.width * bytes_per_pixel; + + u32 src_offset = VideoCore::GetMortonOffset(x, y, bytes_per_pixel) + coarse_y * stride; + u8* src_pixel = depth_buffer + src_offset; switch (registers.framebuffer.depth_format) { - case registers.framebuffer.D16: - { - u32 stride = registers.framebuffer.width * 2; - return Color::DecodeD16(depth_buffer + VideoCore::GetMortonOffset(x, y, 2) + coarse_y * stride); - } - case registers.framebuffer.D24: - { - u32 stride = registers.framebuffer.width * 3; - u8* address = depth_buffer + VideoCore::GetMortonOffset(x, y, 3) + coarse_y * stride; - return Color::DecodeD24(address); - } - case registers.framebuffer.D24S8: - { - u32 stride = registers.framebuffer.width * 4; - return Color::DecodeD24S8(depth_buffer + VideoCore::GetMortonOffset(x, y, 4) + coarse_y * stride).x; - } + case Pica::Regs::DepthFormat::D16: + return Color::DecodeD16(src_pixel); + case Pica::Regs::DepthFormat::D24: + return Color::DecodeD24(src_pixel); + case Pica::Regs::DepthFormat::D24S8: + return Color::DecodeD24S8(src_pixel).x; default: LOG_CRITICAL(HW_GPU, "Unimplemented depth format %u", registers.framebuffer.depth_format); UNIMPLEMENTED(); @@ -132,28 +127,23 @@ static void SetDepth(int x, int y, u32 value) { y = (registers.framebuffer.height - y); const u32 coarse_y = y & ~7; + u32 bytes_per_pixel = Pica::Regs::BytesPerDepthPixel(registers.framebuffer.depth_format); + u32 stride = registers.framebuffer.width * bytes_per_pixel; + + u32 dst_offset = VideoCore::GetMortonOffset(x, y, bytes_per_pixel) + coarse_y * stride; + u8* dst_pixel = depth_buffer + dst_offset; switch (registers.framebuffer.depth_format) { - case registers.framebuffer.D16: - { - u32 stride = registers.framebuffer.width * 2; - Color::EncodeD16(value, depth_buffer + VideoCore::GetMortonOffset(x, y, 2) + coarse_y * stride); + case Pica::Regs::DepthFormat::D16: + Color::EncodeD16(value, dst_pixel); break; - } - case registers.framebuffer.D24: - { - u32 stride = registers.framebuffer.width * 3; - u8* address = depth_buffer + VideoCore::GetMortonOffset(x, y, 3) + coarse_y * stride; - Color::EncodeD24(value, address); + case Pica::Regs::DepthFormat::D24: + Color::EncodeD24(value, dst_pixel); break; - } - case registers.framebuffer.D24S8: - { - u32 stride = registers.framebuffer.width * 4; + case Pica::Regs::DepthFormat::D24S8: // TODO(Subv): Implement the stencil buffer - Color::EncodeD24S8(value, 0, depth_buffer + VideoCore::GetMortonOffset(x, y, 4) + coarse_y * stride); + Color::EncodeD24S8(value, 0, dst_pixel); break; - } default: LOG_CRITICAL(HW_GPU, "Unimplemented depth format %u", registers.framebuffer.depth_format); UNIMPLEMENTED(); -- cgit v1.2.3