Mirrors/yuzu
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Sources: Run clang-format on everything.

Browse source
This commit is contained in:
Emmanuel Gil Peyrot 2016-09-18 09:38:01 +09:00
parent fe948af095
commit dc8479928c
386 changed files with 19560 additions and 18080 deletions
Download patch file Download diff file Expand all files Collapse all files

234
src/video_core/renderer_opengl/gl_rasterizer.cpp
View file

@ -32,8 +32,7 @@ static bool IsPassThroughTevStage(const Pica::Regs::TevStageConfig& stage) {
stage.alpha_source1 == Pica::Regs::TevStageConfig::Source::Previous &&
stage.color_modifier1 == Pica::Regs::TevStageConfig::ColorModifier::SourceColor &&
stage.alpha_modifier1 == Pica::Regs::TevStageConfig::AlphaModifier::SourceAlpha &&
stage.GetColorMultiplier() == 1 &&
stage.GetAlphaMultiplier() == 1);
stage.GetColorMultiplier() == 1 && stage.GetAlphaMultiplier() == 1);
}
RasterizerOpenGL::RasterizerOpenGL() : shader_dirty(true) {
@ -65,26 +64,34 @@ RasterizerOpenGL::RasterizerOpenGL() : shader_dirty(true) {
uniform_block_data.fog_lut_dirty = true;
// Set vertex attributes
glVertexAttribPointer(GLShader::ATTRIBUTE_POSITION, 4, GL_FLOAT, GL_FALSE, sizeof(HardwareVertex), (GLvoid*)offsetof(HardwareVertex, position));
glVertexAttribPointer(GLShader::ATTRIBUTE_POSITION, 4, GL_FLOAT, GL_FALSE,
sizeof(HardwareVertex), (GLvoid*)offsetof(HardwareVertex, position));
glEnableVertexAttribArray(GLShader::ATTRIBUTE_POSITION);
glVertexAttribPointer(GLShader::ATTRIBUTE_COLOR, 4, GL_FLOAT, GL_FALSE, sizeof(HardwareVertex), (GLvoid*)offsetof(HardwareVertex, color));
glVertexAttribPointer(GLShader::ATTRIBUTE_COLOR, 4, GL_FLOAT, GL_FALSE, sizeof(HardwareVertex),
(GLvoid*)offsetof(HardwareVertex, color));
glEnableVertexAttribArray(GLShader::ATTRIBUTE_COLOR);
glVertexAttribPointer(GLShader::ATTRIBUTE_TEXCOORD0, 2, GL_FLOAT, GL_FALSE, sizeof(HardwareVertex), (GLvoid*)offsetof(HardwareVertex, tex_coord0));
glVertexAttribPointer(GLShader::ATTRIBUTE_TEXCOORD1, 2, GL_FLOAT, GL_FALSE, sizeof(HardwareVertex), (GLvoid*)offsetof(HardwareVertex, tex_coord1));
glVertexAttribPointer(GLShader::ATTRIBUTE_TEXCOORD2, 2, GL_FLOAT, GL_FALSE, sizeof(HardwareVertex), (GLvoid*)offsetof(HardwareVertex, tex_coord2));
glVertexAttribPointer(GLShader::ATTRIBUTE_TEXCOORD0, 2, GL_FLOAT, GL_FALSE,
sizeof(HardwareVertex), (GLvoid*)offsetof(HardwareVertex, tex_coord0));
glVertexAttribPointer(GLShader::ATTRIBUTE_TEXCOORD1, 2, GL_FLOAT, GL_FALSE,
sizeof(HardwareVertex), (GLvoid*)offsetof(HardwareVertex, tex_coord1));
glVertexAttribPointer(GLShader::ATTRIBUTE_TEXCOORD2, 2, GL_FLOAT, GL_FALSE,
sizeof(HardwareVertex), (GLvoid*)offsetof(HardwareVertex, tex_coord2));
glEnableVertexAttribArray(GLShader::ATTRIBUTE_TEXCOORD0);
glEnableVertexAttribArray(GLShader::ATTRIBUTE_TEXCOORD1);
glEnableVertexAttribArray(GLShader::ATTRIBUTE_TEXCOORD2);
glVertexAttribPointer(GLShader::ATTRIBUTE_TEXCOORD0_W, 1, GL_FLOAT, GL_FALSE, sizeof(HardwareVertex), (GLvoid*)offsetof(HardwareVertex, tex_coord0_w));
glVertexAttribPointer(GLShader::ATTRIBUTE_TEXCOORD0_W, 1, GL_FLOAT, GL_FALSE,
sizeof(HardwareVertex), (GLvoid*)offsetof(HardwareVertex, tex_coord0_w));
glEnableVertexAttribArray(GLShader::ATTRIBUTE_TEXCOORD0_W);
glVertexAttribPointer(GLShader::ATTRIBUTE_NORMQUAT, 4, GL_FLOAT, GL_FALSE, sizeof(HardwareVertex), (GLvoid*)offsetof(HardwareVertex, normquat));
glVertexAttribPointer(GLShader::ATTRIBUTE_NORMQUAT, 4, GL_FLOAT, GL_FALSE,
sizeof(HardwareVertex), (GLvoid*)offsetof(HardwareVertex, normquat));
glEnableVertexAttribArray(GLShader::ATTRIBUTE_NORMQUAT);
glVertexAttribPointer(GLShader::ATTRIBUTE_VIEW, 3, GL_FLOAT, GL_FALSE, sizeof(HardwareVertex), (GLvoid*)offsetof(HardwareVertex, view));
glVertexAttribPointer(GLShader::ATTRIBUTE_VIEW, 3, GL_FLOAT, GL_FALSE, sizeof(HardwareVertex),
(GLvoid*)offsetof(HardwareVertex, view));
glEnableVertexAttribArray(GLShader::ATTRIBUTE_VIEW);
// Create render framebuffer
@ -130,7 +137,6 @@ RasterizerOpenGL::RasterizerOpenGL() : shader_dirty(true) {
}
RasterizerOpenGL::~RasterizerOpenGL() {
}
/**
@ -149,8 +155,8 @@ RasterizerOpenGL::~RasterizerOpenGL() {
* manually using two Lerps, and doing this correction before each Lerp.
*/
static bool AreQuaternionsOpposite(Math::Vec4<Pica::float24> qa, Math::Vec4<Pica::float24> qb) {
Math::Vec4f a{ qa.x.ToFloat32(), qa.y.ToFloat32(), qa.z.ToFloat32(), qa.w.ToFloat32() };
Math::Vec4f b{ qb.x.ToFloat32(), qb.y.ToFloat32(), qb.z.ToFloat32(), qb.w.ToFloat32() };
Math::Vec4f a{qa.x.ToFloat32(), qa.y.ToFloat32(), qa.z.ToFloat32(), qa.w.ToFloat32()};
Math::Vec4f b{qb.x.ToFloat32(), qb.y.ToFloat32(), qb.z.ToFloat32(), qb.w.ToFloat32()};
return (Math::Dot(a, b) < 0.f);
}
@ -173,15 +179,20 @@ void RasterizerOpenGL::DrawTriangles() {
CachedSurface* color_surface;
CachedSurface* depth_surface;
MathUtil::Rectangle<int> rect;
std::tie(color_surface, depth_surface, rect) = res_cache.GetFramebufferSurfaces(regs.framebuffer);
std::tie(color_surface, depth_surface, rect) =
res_cache.GetFramebufferSurfaces(regs.framebuffer);
state.draw.draw_framebuffer = framebuffer.handle;
state.Apply();
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, color_surface != nullptr ? color_surface->texture.handle : 0, 0);
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_ATTACHMENT, GL_TEXTURE_2D, depth_surface != nullptr ? depth_surface->texture.handle : 0, 0);
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D,
color_surface != nullptr ? color_surface->texture.handle : 0, 0);
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_ATTACHMENT, GL_TEXTURE_2D,
depth_surface != nullptr ? depth_surface->texture.handle : 0, 0);
bool has_stencil = regs.framebuffer.depth_format == Pica::Regs::DepthFormat::D24S8;
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_STENCIL_ATTACHMENT, GL_TEXTURE_2D, (has_stencil && depth_surface != nullptr) ? depth_surface->texture.handle : 0, 0);
glFramebufferTexture2D(
GL_DRAW_FRAMEBUFFER, GL_STENCIL_ATTACHMENT, GL_TEXTURE_2D,
(has_stencil && depth_surface != nullptr) ? depth_surface->texture.handle : 0, 0);
if (OpenGLState::CheckFBStatus(GL_DRAW_FRAMEBUFFER) != GL_FRAMEBUFFER_COMPLETE) {
return;
@ -194,7 +205,8 @@ void RasterizerOpenGL::DrawTriangles() {
glViewport((GLint)(rect.left + regs.viewport_corner.x * color_surface->res_scale_width),
(GLint)(rect.bottom + regs.viewport_corner.y * color_surface->res_scale_height),
(GLsizei)(viewport_width * color_surface->res_scale_width), (GLsizei)(viewport_height * color_surface->res_scale_height));
(GLsizei)(viewport_width * color_surface->res_scale_width),
(GLsizei)(viewport_height * color_surface->res_scale_height));
if (uniform_block_data.data.framebuffer_scale[0] != color_surface->res_scale_width ||
uniform_block_data.data.framebuffer_scale[1] != color_surface->res_scale_height) {
@ -245,14 +257,16 @@ void RasterizerOpenGL::DrawTriangles() {
// Sync the uniform data
if (uniform_block_data.dirty) {
glBufferData(GL_UNIFORM_BUFFER, sizeof(UniformData), &uniform_block_data.data, GL_STATIC_DRAW);
glBufferData(GL_UNIFORM_BUFFER, sizeof(UniformData), &uniform_block_data.data,
GL_STATIC_DRAW);
uniform_block_data.dirty = false;
}
state.Apply();
// Draw the vertex batch
glBufferData(GL_ARRAY_BUFFER, vertex_batch.size() * sizeof(HardwareVertex), vertex_batch.data(), GL_STREAM_DRAW);
glBufferData(GL_ARRAY_BUFFER, vertex_batch.size() * sizeof(HardwareVertex), vertex_batch.data(),
GL_STREAM_DRAW);
glDrawArrays(GL_TRIANGLES, 0, (GLsizei)vertex_batch.size());
// Mark framebuffer surfaces as dirty
@ -278,7 +292,7 @@ void RasterizerOpenGL::DrawTriangles() {
void RasterizerOpenGL::NotifyPicaRegisterChanged(u32 id) {
const auto& regs = Pica::g_state.regs;
switch(id) {
switch (id) {
// Culling
case PICA_REG_INDEX(cull_mode):
SyncCullMode();
@ -548,7 +562,7 @@ void RasterizerOpenGL::NotifyPicaRegisterChanged(u32 id) {
SyncLightAmbient(7);
break;
// Fragment lighting position
// Fragment lighting position
case PICA_REG_INDEX_WORKAROUND(lighting.light[0].x, 0x144 + 0 * 0x10):
case PICA_REG_INDEX_WORKAROUND(lighting.light[0].z, 0x145 + 0 * 0x10):
SyncLightPosition(0);
@ -659,13 +673,11 @@ void RasterizerOpenGL::NotifyPicaRegisterChanged(u32 id) {
case PICA_REG_INDEX_WORKAROUND(lighting.lut_data[4], 0x1cc):
case PICA_REG_INDEX_WORKAROUND(lighting.lut_data[5], 0x1cd):
case PICA_REG_INDEX_WORKAROUND(lighting.lut_data[6], 0x1ce):
case PICA_REG_INDEX_WORKAROUND(lighting.lut_data[7], 0x1cf):
{
case PICA_REG_INDEX_WORKAROUND(lighting.lut_data[7], 0x1cf): {
auto& lut_config = regs.lighting.lut_config;
uniform_block_data.lut_dirty[lut_config.type / 4] = true;
break;
}
}
}
@ -699,8 +711,10 @@ bool RasterizerOpenGL::AccelerateDisplayTransfer(const GPU::Regs::DisplayTransfe
CachedSurface dst_params;
dst_params.addr = config.GetPhysicalOutputAddress();
dst_params.width = config.scaling != config.NoScale ? config.output_width / 2 : config.output_width.Value();
dst_params.height = config.scaling == config.ScaleXY ? config.output_height / 2 : config.output_height.Value();
dst_params.width =
config.scaling != config.NoScale ? config.output_width / 2 : config.output_width.Value();
dst_params.height =
config.scaling == config.ScaleXY ? config.output_height / 2 : config.output_height.Value();
dst_params.is_tiled = config.input_linear != config.dont_swizzle;
dst_params.pixel_format = CachedSurface::PixelFormatFromGPUPixelFormat(config.output_format);
@ -735,7 +749,8 @@ bool RasterizerOpenGL::AccelerateDisplayTransfer(const GPU::Regs::DisplayTransfe
return false;
}
u32 dst_size = dst_params.width * dst_params.height * CachedSurface::GetFormatBpp(dst_params.pixel_format) / 8;
u32 dst_size = dst_params.width * dst_params.height *
CachedSurface::GetFormatBpp(dst_params.pixel_format) / 8;
dst_surface->dirty = true;
res_cache.FlushRegion(config.GetPhysicalOutputAddress(), dst_size, dst_surface, true);
return true;
@ -757,12 +772,15 @@ bool RasterizerOpenGL::AccelerateFill(const GPU::Regs::MemoryFillConfig& config)
GLuint old_fb = cur_state.draw.draw_framebuffer;
cur_state.draw.draw_framebuffer = framebuffer.handle;
// TODO: When scissor test is implemented, need to disable scissor test in cur_state here so Clear call isn't affected
// TODO: When scissor test is implemented, need to disable scissor test in cur_state here so
// Clear call isn't affected
cur_state.Apply();
if (dst_type == SurfaceType::Color || dst_type == SurfaceType::Texture) {
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, dst_surface->texture.handle, 0);
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0, 0);
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D,
dst_surface->texture.handle, 0);
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0,
0);
if (OpenGLState::CheckFBStatus(GL_DRAW_FRAMEBUFFER) != GL_FRAMEBUFFER_COMPLETE) {
return false;
@ -770,8 +788,10 @@ bool RasterizerOpenGL::AccelerateFill(const GPU::Regs::MemoryFillConfig& config)
GLfloat color_values[4] = {0.0f, 0.0f, 0.0f, 0.0f};
// TODO: Handle additional pixel format and fill value size combinations to accelerate more cases
// For instance, checking if fill value's bytes/bits repeat to allow filling I8/A8/I4/A4/...
// TODO: Handle additional pixel format and fill value size combinations to accelerate more
// cases
// For instance, checking if fill value's bytes/bits repeat to allow filling
// I8/A8/I4/A4/...
// Currently only handles formats that are multiples of the fill value size
if (config.fill_24bit) {
@ -846,7 +866,8 @@ bool RasterizerOpenGL::AccelerateFill(const GPU::Regs::MemoryFillConfig& config)
glClearBufferfv(GL_COLOR, 0, color_values);
} else if (dst_type == SurfaceType::Depth) {
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, 0, 0);
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_ATTACHMENT, GL_TEXTURE_2D, dst_surface->texture.handle, 0);
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_ATTACHMENT, GL_TEXTURE_2D,
dst_surface->texture.handle, 0);
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0, 0);
if (OpenGLState::CheckFBStatus(GL_DRAW_FRAMEBUFFER) != GL_FRAMEBUFFER_COMPLETE) {
@ -865,7 +886,8 @@ bool RasterizerOpenGL::AccelerateFill(const GPU::Regs::MemoryFillConfig& config)
glClearBufferfv(GL_DEPTH, 0, &value_float);
} else if (dst_type == SurfaceType::DepthStencil) {
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, 0, 0);
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, dst_surface->texture.handle, 0);
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D,
dst_surface->texture.handle, 0);
if (OpenGLState::CheckFBStatus(GL_DRAW_FRAMEBUFFER) != GL_FRAMEBUFFER_COMPLETE) {
return false;
@ -889,7 +911,9 @@ bool RasterizerOpenGL::AccelerateFill(const GPU::Regs::MemoryFillConfig& config)
return true;
}
bool RasterizerOpenGL::AccelerateDisplay(const GPU::Regs::FramebufferConfig& config, PAddr framebuffer_addr, u32 pixel_stride, ScreenInfo& screen_info) {
bool RasterizerOpenGL::AccelerateDisplay(const GPU::Regs::FramebufferConfig& config,
PAddr framebuffer_addr, u32 pixel_stride,
ScreenInfo& screen_info) {
if (framebuffer_addr == 0) {
return false;
}
@ -912,10 +936,9 @@ bool RasterizerOpenGL::AccelerateDisplay(const GPU::Regs::FramebufferConfig& con
u32 scaled_width = src_surface->GetScaledWidth();
u32 scaled_height = src_surface->GetScaledHeight();
screen_info.display_texcoords = MathUtil::Rectangle<float>((float)src_rect.top / (float)scaled_height,
(float)src_rect.left / (float)scaled_width,
(float)src_rect.bottom / (float)scaled_height,
(float)src_rect.right / (float)scaled_width);
screen_info.display_texcoords = MathUtil::Rectangle<float>(
(float)src_rect.top / (float)scaled_height, (float)src_rect.left / (float)scaled_width,
(float)src_rect.bottom / (float)scaled_height, (float)src_rect.right / (float)scaled_width);
screen_info.display_texture = src_surface->texture.handle;
@ -928,7 +951,8 @@ void RasterizerOpenGL::SamplerInfo::Create() {
wrap_s = wrap_t = TextureConfig::Repeat;
border_color = 0;
glSamplerParameteri(sampler.handle, GL_TEXTURE_MIN_FILTER, GL_LINEAR); // default is GL_LINEAR_MIPMAP_LINEAR
glSamplerParameteri(sampler.handle, GL_TEXTURE_MIN_FILTER,
GL_LINEAR); // default is GL_LINEAR_MIPMAP_LINEAR
// Other attributes have correct defaults
}
@ -976,41 +1000,64 @@ void RasterizerOpenGL::SetShader() {
} else {
LOG_DEBUG(Render_OpenGL, "Creating new shader");
shader->shader.Create(GLShader::GenerateVertexShader().c_str(), GLShader::GenerateFragmentShader(config).c_str());
shader->shader.Create(GLShader::GenerateVertexShader().c_str(),
GLShader::GenerateFragmentShader(config).c_str());
state.draw.shader_program = shader->shader.handle;
state.Apply();
// Set the texture samplers to correspond to different texture units
GLuint uniform_tex = glGetUniformLocation(shader->shader.handle, "tex[0]");
if (uniform_tex != -1) { glUniform1i(uniform_tex, 0); }
if (uniform_tex != -1) {
glUniform1i(uniform_tex, 0);
}
uniform_tex = glGetUniformLocation(shader->shader.handle, "tex[1]");
if (uniform_tex != -1) { glUniform1i(uniform_tex, 1); }
if (uniform_tex != -1) {
glUniform1i(uniform_tex, 1);
}
uniform_tex = glGetUniformLocation(shader->shader.handle, "tex[2]");
if (uniform_tex != -1) { glUniform1i(uniform_tex, 2); }
if (uniform_tex != -1) {
glUniform1i(uniform_tex, 2);
}
// Set the texture samplers to correspond to different lookup table texture units
GLuint uniform_lut = glGetUniformLocation(shader->shader.handle, "lut[0]");
if (uniform_lut != -1) { glUniform1i(uniform_lut, 3); }
if (uniform_lut != -1) {
glUniform1i(uniform_lut, 3);
}
uniform_lut = glGetUniformLocation(shader->shader.handle, "lut[1]");
if (uniform_lut != -1) { glUniform1i(uniform_lut, 4); }
if (uniform_lut != -1) {
glUniform1i(uniform_lut, 4);
}
uniform_lut = glGetUniformLocation(shader->shader.handle, "lut[2]");
if (uniform_lut != -1) { glUniform1i(uniform_lut, 5); }
if (uniform_lut != -1) {
glUniform1i(uniform_lut, 5);
}
uniform_lut = glGetUniformLocation(shader->shader.handle, "lut[3]");
if (uniform_lut != -1) { glUniform1i(uniform_lut, 6); }
if (uniform_lut != -1) {
glUniform1i(uniform_lut, 6);
}
uniform_lut = glGetUniformLocation(shader->shader.handle, "lut[4]");
if (uniform_lut != -1) { glUniform1i(uniform_lut, 7); }
if (uniform_lut != -1) {
glUniform1i(uniform_lut, 7);
}
uniform_lut = glGetUniformLocation(shader->shader.handle, "lut[5]");
if (uniform_lut != -1) { glUniform1i(uniform_lut, 8); }
if (uniform_lut != -1) {
glUniform1i(uniform_lut, 8);
}
GLuint uniform_fog_lut = glGetUniformLocation(shader->shader.handle, "fog_lut");
if (uniform_fog_lut != -1) { glUniform1i(uniform_fog_lut, 9); }
if (uniform_fog_lut != -1) {
glUniform1i(uniform_fog_lut, 9);
}
current_shader = shader_cache.emplace(config, std::move(shader)).first->second.get();
unsigned int block_index = glGetUniformBlockIndex(current_shader->shader.handle, "shader_data");
unsigned int block_index =
glGetUniformBlockIndex(current_shader->shader.handle, "shader_data");
GLint block_size;
glGetActiveUniformBlockiv(current_shader->shader.handle, block_index, GL_UNIFORM_BLOCK_DATA_SIZE, &block_size);
glGetActiveUniformBlockiv(current_shader->shader.handle, block_index,
GL_UNIFORM_BLOCK_DATA_SIZE, &block_size);
ASSERT_MSG(block_size == sizeof(UniformData), "Uniform block size did not match!");
glUniformBlockBinding(current_shader->shader.handle, block_index, 0);
@ -1073,7 +1120,8 @@ void RasterizerOpenGL::SyncDepthScale() {
}
void RasterizerOpenGL::SyncDepthOffset() {
float depth_offset = Pica::float24::FromRaw(Pica::g_state.regs.viewport_depth_near_plane).ToFloat32();
float depth_offset =
Pica::float24::FromRaw(Pica::g_state.regs.viewport_depth_near_plane).ToFloat32();
if (depth_offset != uniform_block_data.data.depth_offset) {
uniform_block_data.data.depth_offset = depth_offset;
uniform_block_data.dirty = true;
@ -1086,10 +1134,14 @@ void RasterizerOpenGL::SyncBlendEnabled() {
void RasterizerOpenGL::SyncBlendFuncs() {
const auto& regs = Pica::g_state.regs;
state.blend.rgb_equation = PicaToGL::BlendEquation(regs.output_merger.alpha_blending.blend_equation_rgb);
state.blend.a_equation = PicaToGL::BlendEquation(regs.output_merger.alpha_blending.blend_equation_a);
state.blend.src_rgb_func = PicaToGL::BlendFunc(regs.output_merger.alpha_blending.factor_source_rgb);
state.blend.dst_rgb_func = PicaToGL::BlendFunc(regs.output_merger.alpha_blending.factor_dest_rgb);
state.blend.rgb_equation =
PicaToGL::BlendEquation(regs.output_merger.alpha_blending.blend_equation_rgb);
state.blend.a_equation =
PicaToGL::BlendEquation(regs.output_merger.alpha_blending.blend_equation_a);
state.blend.src_rgb_func =
PicaToGL::BlendFunc(regs.output_merger.alpha_blending.factor_source_rgb);
state.blend.dst_rgb_func =
PicaToGL::BlendFunc(regs.output_merger.alpha_blending.factor_dest_rgb);
state.blend.src_a_func = PicaToGL::BlendFunc(regs.output_merger.alpha_blending.factor_source_a);
state.blend.dst_a_func = PicaToGL::BlendFunc(regs.output_merger.alpha_blending.factor_dest_a);
}
@ -1104,25 +1156,23 @@ void RasterizerOpenGL::SyncBlendColor() {
void RasterizerOpenGL::SyncFogColor() {
const auto& regs = Pica::g_state.regs;
uniform_block_data.data.fog_color = {
regs.fog_color.r.Value() / 255.0f,
regs.fog_color.g.Value() / 255.0f,
regs.fog_color.b.Value() / 255.0f
};
uniform_block_data.data.fog_color = {regs.fog_color.r.Value() / 255.0f,
regs.fog_color.g.Value() / 255.0f,
regs.fog_color.b.Value() / 255.0f};
uniform_block_data.dirty = true;
}
void RasterizerOpenGL::SyncFogLUT() {
std::array<GLuint, 128> new_data;
std::transform(Pica::g_state.fog.lut.begin(), Pica::g_state.fog.lut.end(), new_data.begin(), [](const auto& entry) {
return entry.raw;
});
std::transform(Pica::g_state.fog.lut.begin(), Pica::g_state.fog.lut.end(), new_data.begin(),
[](const auto& entry) { return entry.raw; });
if (new_data != fog_lut_data) {
fog_lut_data = new_data;
glActiveTexture(GL_TEXTURE9);
glTexSubImage1D(GL_TEXTURE_1D, 0, 0, 128, GL_RED_INTEGER, GL_UNSIGNED_INT, fog_lut_data.data());
glTexSubImage1D(GL_TEXTURE_1D, 0, 0, 128, GL_RED_INTEGER, GL_UNSIGNED_INT,
fog_lut_data.data());
}
}
@ -1154,34 +1204,40 @@ void RasterizerOpenGL::SyncColorWriteMask() {
void RasterizerOpenGL::SyncStencilWriteMask() {
const auto& regs = Pica::g_state.regs;
state.stencil.write_mask = (regs.framebuffer.allow_depth_stencil_write != 0)
? static_cast<GLuint>(regs.output_merger.stencil_test.write_mask)
: 0;
? static_cast<GLuint>(regs.output_merger.stencil_test.write_mask)
: 0;
}
void RasterizerOpenGL::SyncDepthWriteMask() {
const auto& regs = Pica::g_state.regs;
state.depth.write_mask = (regs.framebuffer.allow_depth_stencil_write != 0 && regs.output_merger.depth_write_enable)
? GL_TRUE
: GL_FALSE;
state.depth.write_mask =
(regs.framebuffer.allow_depth_stencil_write != 0 && regs.output_merger.depth_write_enable)
? GL_TRUE
: GL_FALSE;
}
void RasterizerOpenGL::SyncStencilTest() {
const auto& regs = Pica::g_state.regs;
state.stencil.test_enabled = regs.output_merger.stencil_test.enable && regs.framebuffer.depth_format == Pica::Regs::DepthFormat::D24S8;
state.stencil.test_enabled = regs.output_merger.stencil_test.enable &&
regs.framebuffer.depth_format == Pica::Regs::DepthFormat::D24S8;
state.stencil.test_func = PicaToGL::CompareFunc(regs.output_merger.stencil_test.func);
state.stencil.test_ref = regs.output_merger.stencil_test.reference_value;
state.stencil.test_mask = regs.output_merger.stencil_test.input_mask;
state.stencil.action_stencil_fail = PicaToGL::StencilOp(regs.output_merger.stencil_test.action_stencil_fail);
state.stencil.action_depth_fail = PicaToGL::StencilOp(regs.output_merger.stencil_test.action_depth_fail);
state.stencil.action_depth_pass = PicaToGL::StencilOp(regs.output_merger.stencil_test.action_depth_pass);
state.stencil.action_stencil_fail =
PicaToGL::StencilOp(regs.output_merger.stencil_test.action_stencil_fail);
state.stencil.action_depth_fail =
PicaToGL::StencilOp(regs.output_merger.stencil_test.action_depth_fail);
state.stencil.action_depth_pass =
PicaToGL::StencilOp(regs.output_merger.stencil_test.action_depth_pass);
}
void RasterizerOpenGL::SyncDepthTest() {
const auto& regs = Pica::g_state.regs;
state.depth.test_enabled = regs.output_merger.depth_test_enable == 1 ||
regs.output_merger.depth_write_enable == 1;
state.depth.test_func = regs.output_merger.depth_test_enable == 1 ?
PicaToGL::CompareFunc(regs.output_merger.depth_test_func) : GL_ALWAYS;
state.depth.test_enabled =
regs.output_merger.depth_test_enable == 1 || regs.output_merger.depth_write_enable == 1;
state.depth.test_func = regs.output_merger.depth_test_enable == 1
? PicaToGL::CompareFunc(regs.output_merger.depth_test_func)
: GL_ALWAYS;
}
void RasterizerOpenGL::SyncScissorTest() {
@ -1208,7 +1264,8 @@ void RasterizerOpenGL::SyncCombinerColor() {
}
}
void RasterizerOpenGL::SyncTevConstColor(int stage_index, const Pica::Regs::TevStageConfig& tev_stage) {
void RasterizerOpenGL::SyncTevConstColor(int stage_index,
const Pica::Regs::TevStageConfig& tev_stage) {
auto const_color = PicaToGL::ColorRGBA8(tev_stage.const_color);
if (const_color != uniform_block_data.data.const_color[stage_index]) {
uniform_block_data.data.const_color[stage_index] = const_color;
@ -1237,7 +1294,8 @@ void RasterizerOpenGL::SyncLightingLUT(unsigned lut_index) {
if (new_data != lighting_lut_data[lut_index]) {
lighting_lut_data[lut_index] = new_data;
glActiveTexture(GL_TEXTURE3 + lut_index);
glTexSubImage1D(GL_TEXTURE_1D, 0, 0, 256, GL_RGBA, GL_FLOAT, lighting_lut_data[lut_index].data());
glTexSubImage1D(GL_TEXTURE_1D, 0, 0, 256, GL_RGBA, GL_FLOAT,
lighting_lut_data[lut_index].data());
}
}
@ -1277,7 +1335,7 @@ void RasterizerOpenGL::SyncLightPosition(int light_index) {
GLvec3 position = {
Pica::float16::FromRaw(Pica::g_state.regs.lighting.light[light_index].x).ToFloat32(),
Pica::float16::FromRaw(Pica::g_state.regs.lighting.light[light_index].y).ToFloat32(),
Pica::float16::FromRaw(Pica::g_state.regs.lighting.light[light_index].z).ToFloat32() };
Pica::float16::FromRaw(Pica::g_state.regs.lighting.light[light_index].z).ToFloat32()};
if (position != uniform_block_data.data.light_src[light_index].position) {
uniform_block_data.data.light_src[light_index].position = position;
@ -1286,7 +1344,9 @@ void RasterizerOpenGL::SyncLightPosition(int light_index) {
}
void RasterizerOpenGL::SyncLightDistanceAttenuationBias(int light_index) {
GLfloat dist_atten_bias = Pica::float20::FromRaw(Pica::g_state.regs.lighting.light[light_index].dist_atten_bias).ToFloat32();
GLfloat dist_atten_bias =
Pica::float20::FromRaw(Pica::g_state.regs.lighting.light[light_index].dist_atten_bias)
.ToFloat32();
if (dist_atten_bias != uniform_block_data.data.light_src[light_index].dist_atten_bias) {
uniform_block_data.data.light_src[light_index].dist_atten_bias = dist_atten_bias;
@ -1295,7 +1355,9 @@ void RasterizerOpenGL::SyncLightDistanceAttenuationBias(int light_index) {
}
void RasterizerOpenGL::SyncLightDistanceAttenuationScale(int light_index) {
GLfloat dist_atten_scale = Pica::float20::FromRaw(Pica::g_state.regs.lighting.light[light_index].dist_atten_scale).ToFloat32();
GLfloat dist_atten_scale =
Pica::float20::FromRaw(Pica::g_state.regs.lighting.light[light_index].dist_atten_scale)
.ToFloat32();
if (dist_atten_scale != uniform_block_data.data.light_src[light_index].dist_atten_scale) {
uniform_block_data.data.light_src[light_index].dist_atten_scale = dist_atten_scale;

50
src/video_core/renderer_opengl/gl_rasterizer.h
View file

@ -8,8 +8,8 @@
#include <cstddef>
#include <cstring>
#include <memory>
#include <vector>
#include <unordered_map>
#include <vector>
#include <glad/glad.h>
@ -40,10 +40,13 @@ struct ScreenInfo;
* Pica state is not being captured in the shader cache key, thereby resulting in (what should be)
* two separate shaders sharing the same key.
*
* We use a union because "implicitly-defined copy/move constructor for a union X copies the object representation of X."
* and "implicitly-defined copy assignment operator for a union X copies the object representation (3.9) of X."
* We use a union because "implicitly-defined copy/move constructor for a union X copies the object
* representation of X."
* and "implicitly-defined copy assignment operator for a union X copies the object representation
* (3.9) of X."
* = Bytewise copy instead of memberwise copy.
* This is important because the padding bytes are included in the hash and comparison between objects.
* This is important because the padding bytes are included in the hash and comparison between
* objects.
*/
union PicaShaderConfig {
@ -60,8 +63,9 @@ union PicaShaderConfig {
state.depthmap_enable = regs.depthmap_enable;
state.alpha_test_func = regs.output_merger.alpha_test.enable ?
regs.output_merger.alpha_test.func.Value() : Pica::Regs::CompareFunc::Always;
state.alpha_test_func = regs.output_merger.alpha_test.enable
? regs.output_merger.alpha_test.func.Value()
: Pica::Regs::CompareFunc::Always;
state.texture0_type = regs.texture0.type;
@ -81,9 +85,8 @@ union PicaShaderConfig {
state.fog_mode = regs.fog_mode;
state.fog_flip = regs.fog_flip;
state.combiner_buffer_input =
regs.tev_combiner_buffer_input.update_mask_rgb.Value() |
regs.tev_combiner_buffer_input.update_mask_a.Value() << 4;
state.combiner_buffer_input = regs.tev_combiner_buffer_input.update_mask_rgb.Value() |
regs.tev_combiner_buffer_input.update_mask_a.Value() << 4;
// Fragment lighting
@ -95,8 +98,10 @@ union PicaShaderConfig {
const auto& light = regs.lighting.light[num];
state.lighting.light[light_index].num = num;
state.lighting.light[light_index].directional = light.config.directional != 0;
state.lighting.light[light_index].two_sided_diffuse = light.config.two_sided_diffuse != 0;
state.lighting.light[light_index].dist_atten_enable = !regs.lighting.IsDistAttenDisabled(num);
state.lighting.light[light_index].two_sided_diffuse =
light.config.two_sided_diffuse != 0;
state.lighting.light[light_index].dist_atten_enable =
!regs.lighting.IsDistAttenDisabled(num);
}
state.lighting.lut_d0.enable = regs.lighting.config1.disable_lut_d0 == 0;
@ -147,7 +152,7 @@ union PicaShaderConfig {
return (stage_index < 4) && ((state.combiner_buffer_input >> 4) & (1 << stage_index));
}
bool operator ==(const PicaShaderConfig& o) const {
bool operator==(const PicaShaderConfig& o) const {
return std::memcmp(&state, &o.state, sizeof(PicaShaderConfig::State)) == 0;
};
@ -212,7 +217,8 @@ union PicaShaderConfig {
} state;
};
#if (__GNUC__ >= 5) || defined(__clang__) || defined(_MSC_VER)
static_assert(std::is_trivially_copyable<PicaShaderConfig::State>::value, "PicaShaderConfig::State must be trivially copyable");
static_assert(std::is_trivially_copyable<PicaShaderConfig::State>::value,
"PicaShaderConfig::State must be trivially copyable");
#endif
namespace std {
@ -228,12 +234,10 @@ struct hash<PicaShaderConfig> {
class RasterizerOpenGL : public VideoCore::RasterizerInterface {
public:
RasterizerOpenGL();
~RasterizerOpenGL() override;
void AddTriangle(const Pica::Shader::OutputVertex& v0,
const Pica::Shader::OutputVertex& v1,
void AddTriangle(const Pica::Shader::OutputVertex& v0, const Pica::Shader::OutputVertex& v1,
const Pica::Shader::OutputVertex& v2) override;
void DrawTriangles() override;
void NotifyPicaRegisterChanged(u32 id) override;
@ -242,7 +246,8 @@ public:
void FlushAndInvalidateRegion(PAddr addr, u32 size) override;
bool AccelerateDisplayTransfer(const GPU::Regs::DisplayTransferConfig& config) override;
bool AccelerateFill(const GPU::Regs::MemoryFillConfig& config) override;
bool AccelerateDisplay(const GPU::Regs::FramebufferConfig& config, PAddr framebuffer_addr, u32 pixel_stride, ScreenInfo& screen_info) override;
bool AccelerateDisplay(const GPU::Regs::FramebufferConfig& config, PAddr framebuffer_addr,
u32 pixel_stride, ScreenInfo& screen_info) override;
/// OpenGL shader generated for a given Pica register state
struct PicaShader {
@ -251,13 +256,13 @@ public:
};
private:
struct SamplerInfo {
using TextureConfig = Pica::Regs::TextureConfig;
OGLSampler sampler;
/// Creates the sampler object, initializing its state so that it's in sync with the SamplerInfo struct.
/// Creates the sampler object, initializing its state so that it's in sync with the
/// SamplerInfo struct.
void Create();
/// Syncs the sampler object with the config, updating any necessary state.
void SyncWithConfig(const TextureConfig& config);
@ -343,8 +348,11 @@ private:
alignas(16) GLvec4 tev_combiner_buffer_color;
};
static_assert(sizeof(UniformData) == 0x3C0, "The size of the UniformData structure has changed, update the structure in the shader");
static_assert(sizeof(UniformData) < 16384, "UniformData structure must be less than 16kb as per the OpenGL spec");
static_assert(
sizeof(UniformData) == 0x3C0,
"The size of the UniformData structure has changed, update the structure in the shader");
static_assert(sizeof(UniformData) < 16384,
"UniformData structure must be less than 16kb as per the OpenGL spec");
/// Sets the OpenGL shader in accordance with the current PICA register state
void SetShader();

285
src/video_core/renderer_opengl/gl_rasterizer_cache.cpp
View file

@ -35,18 +35,18 @@ struct FormatTuple {
};
static const std::array<FormatTuple, 5> fb_format_tuples = {{
{ GL_RGBA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8 }, // RGBA8
{ GL_RGB8, GL_BGR, GL_UNSIGNED_BYTE }, // RGB8
{ GL_RGB5_A1, GL_RGBA, GL_UNSIGNED_SHORT_5_5_5_1 }, // RGB5A1
{ GL_RGB565, GL_RGB, GL_UNSIGNED_SHORT_5_6_5 }, // RGB565
{ GL_RGBA4, GL_RGBA, GL_UNSIGNED_SHORT_4_4_4_4 }, // RGBA4
{GL_RGBA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8}, // RGBA8
{GL_RGB8, GL_BGR, GL_UNSIGNED_BYTE}, // RGB8
{GL_RGB5_A1, GL_RGBA, GL_UNSIGNED_SHORT_5_5_5_1}, // RGB5A1
{GL_RGB565, GL_RGB, GL_UNSIGNED_SHORT_5_6_5}, // RGB565
{GL_RGBA4, GL_RGBA, GL_UNSIGNED_SHORT_4_4_4_4}, // RGBA4
}};
static const std::array<FormatTuple, 4> depth_format_tuples = {{
{ GL_DEPTH_COMPONENT16, GL_DEPTH_COMPONENT, GL_UNSIGNED_SHORT }, // D16
{GL_DEPTH_COMPONENT16, GL_DEPTH_COMPONENT, GL_UNSIGNED_SHORT}, // D16
{},
{ GL_DEPTH_COMPONENT24, GL_DEPTH_COMPONENT, GL_UNSIGNED_INT }, // D24
{ GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8 }, // D24S8
{GL_DEPTH_COMPONENT24, GL_DEPTH_COMPONENT, GL_UNSIGNED_INT}, // D24
{GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8}, // D24S8
}};
RasterizerCacheOpenGL::RasterizerCacheOpenGL() {
@ -58,7 +58,9 @@ RasterizerCacheOpenGL::~RasterizerCacheOpenGL() {
FlushAll();
}
static void MortonCopyPixels(CachedSurface::PixelFormat pixel_format, u32 width, u32 height, u32 bytes_per_pixel, u32 gl_bytes_per_pixel, u8* morton_data, u8* gl_data, bool morton_to_gl) {
static void MortonCopyPixels(CachedSurface::PixelFormat pixel_format, u32 width, u32 height,
u32 bytes_per_pixel, u32 gl_bytes_per_pixel, u8* morton_data,
u8* gl_data, bool morton_to_gl) {
using PixelFormat = CachedSurface::PixelFormat;
u8* data_ptrs[2];
@ -72,7 +74,8 @@ static void MortonCopyPixels(CachedSurface::PixelFormat pixel_format, u32 width,
for (unsigned y = 0; y < height; ++y) {
for (unsigned x = 0; x < width; ++x) {
const u32 coarse_y = y & ~7;
u32 morton_offset = VideoCore::GetMortonOffset(x, y, bytes_per_pixel) + coarse_y * width * bytes_per_pixel;
u32 morton_offset = VideoCore::GetMortonOffset(x, y, bytes_per_pixel) +
coarse_y * width * bytes_per_pixel;
u32 gl_pixel_index = (x + (height - 1 - y) * width) * gl_bytes_per_pixel;
data_ptrs[morton_to_gl] = morton_data + morton_offset;
@ -81,7 +84,8 @@ static void MortonCopyPixels(CachedSurface::PixelFormat pixel_format, u32 width,
// Swap depth and stencil value ordering since 3DS does not match OpenGL
u32 depth_stencil;
memcpy(&depth_stencil, data_ptrs[1], sizeof(u32));
depth_stencil = (depth_stencil << depth_stencil_shifts[0]) | (depth_stencil >> depth_stencil_shifts[1]);
depth_stencil = (depth_stencil << depth_stencil_shifts[0]) |
(depth_stencil >> depth_stencil_shifts[1]);
memcpy(data_ptrs[0], &depth_stencil, sizeof(u32));
}
@ -90,7 +94,8 @@ static void MortonCopyPixels(CachedSurface::PixelFormat pixel_format, u32 width,
for (unsigned y = 0; y < height; ++y) {
for (unsigned x = 0; x < width; ++x) {
const u32 coarse_y = y & ~7;
u32 morton_offset = VideoCore::GetMortonOffset(x, y, bytes_per_pixel) + coarse_y * width * bytes_per_pixel;
u32 morton_offset = VideoCore::GetMortonOffset(x, y, bytes_per_pixel) +
coarse_y * width * bytes_per_pixel;
u32 gl_pixel_index = (x + (height - 1 - y) * width) * gl_bytes_per_pixel;
data_ptrs[morton_to_gl] = morton_data + morton_offset;
@ -102,17 +107,21 @@ static void MortonCopyPixels(CachedSurface::PixelFormat pixel_format, u32 width,
}
}
bool RasterizerCacheOpenGL::BlitTextures(GLuint src_tex, GLuint dst_tex, CachedSurface::SurfaceType type, const MathUtil::Rectangle<int>& src_rect, const MathUtil::Rectangle<int>& dst_rect) {
bool RasterizerCacheOpenGL::BlitTextures(GLuint src_tex, GLuint dst_tex,
CachedSurface::SurfaceType type,
const MathUtil::Rectangle<int>& src_rect,
const MathUtil::Rectangle<int>& dst_rect) {
using SurfaceType = CachedSurface::SurfaceType;
OpenGLState cur_state = OpenGLState::GetCurState();
// Make sure textures aren't bound to texture units, since going to bind them to framebuffer components
// Make sure textures aren't bound to texture units, since going to bind them to framebuffer
// components
OpenGLState::ResetTexture(src_tex);
OpenGLState::ResetTexture(dst_tex);
// Keep track of previous framebuffer bindings
GLuint old_fbs[2] = { cur_state.draw.read_framebuffer, cur_state.draw.draw_framebuffer };
GLuint old_fbs[2] = {cur_state.draw.read_framebuffer, cur_state.draw.draw_framebuffer};
cur_state.draw.read_framebuffer = transfer_framebuffers[0].handle;
cur_state.draw.draw_framebuffer = transfer_framebuffers[1].handle;
cur_state.Apply();
@ -120,11 +129,15 @@ bool RasterizerCacheOpenGL::BlitTextures(GLuint src_tex, GLuint dst_tex, CachedS
u32 buffers = 0;
if (type == SurfaceType::Color || type == SurfaceType::Texture) {
glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, src_tex, 0);
glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0, 0);
glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, src_tex,
0);
glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0,
0);
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, dst_tex, 0);
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0, 0);
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, dst_tex,
0);
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0,
0);
buffers = GL_COLOR_BUFFER_BIT;
} else if (type == SurfaceType::Depth) {
@ -139,10 +152,12 @@ bool RasterizerCacheOpenGL::BlitTextures(GLuint src_tex, GLuint dst_tex, CachedS
buffers = GL_DEPTH_BUFFER_BIT;
} else if (type == SurfaceType::DepthStencil) {
glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, 0, 0);
glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, src_tex, 0);
glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D,
src_tex, 0);
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, 0, 0);
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, dst_tex, 0);
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D,
dst_tex, 0);
buffers = GL_DEPTH_BUFFER_BIT | GL_STENCIL_BUFFER_BIT;
}
@ -155,9 +170,9 @@ bool RasterizerCacheOpenGL::BlitTextures(GLuint src_tex, GLuint dst_tex, CachedS
return false;
}
glBlitFramebuffer(src_rect.left, src_rect.top, src_rect.right, src_rect.bottom,
dst_rect.left, dst_rect.top, dst_rect.right, dst_rect.bottom,
buffers, buffers == GL_COLOR_BUFFER_BIT ? GL_LINEAR : GL_NEAREST);
glBlitFramebuffer(src_rect.left, src_rect.top, src_rect.right, src_rect.bottom, dst_rect.left,
dst_rect.top, dst_rect.right, dst_rect.bottom, buffers,
buffers == GL_COLOR_BUFFER_BIT ? GL_LINEAR : GL_NEAREST);
// Restore previous framebuffer bindings
cur_state.draw.read_framebuffer = old_fbs[0];
@ -167,17 +182,24 @@ bool RasterizerCacheOpenGL::BlitTextures(GLuint src_tex, GLuint dst_tex, CachedS
return true;
}
bool RasterizerCacheOpenGL::TryBlitSurfaces(CachedSurface* src_surface, const MathUtil::Rectangle<int>& src_rect, CachedSurface* dst_surface, const MathUtil::Rectangle<int>& dst_rect) {
bool RasterizerCacheOpenGL::TryBlitSurfaces(CachedSurface* src_surface,
const MathUtil::Rectangle<int>& src_rect,
CachedSurface* dst_surface,
const MathUtil::Rectangle<int>& dst_rect) {
using SurfaceType = CachedSurface::SurfaceType;
if (!CachedSurface::CheckFormatsBlittable(src_surface->pixel_format, dst_surface->pixel_format)) {
if (!CachedSurface::CheckFormatsBlittable(src_surface->pixel_format,
dst_surface->pixel_format)) {
return false;
}
return BlitTextures(src_surface->texture.handle, dst_surface->texture.handle, CachedSurface::GetFormatType(src_surface->pixel_format), src_rect, dst_rect);
return BlitTextures(src_surface->texture.handle, dst_surface->texture.handle,
CachedSurface::GetFormatType(src_surface->pixel_format), src_rect,
dst_rect);
}
static void AllocateSurfaceTexture(GLuint texture, CachedSurface::PixelFormat pixel_format, u32 width, u32 height) {
static void AllocateSurfaceTexture(GLuint texture, CachedSurface::PixelFormat pixel_format,
u32 width, u32 height) {
// Allocate an uninitialized texture of appropriate size and format for the surface
using SurfaceType = CachedSurface::SurfaceType;
@ -200,11 +222,11 @@ static void AllocateSurfaceTexture(GLuint texture, CachedSurface::PixelFormat pi
ASSERT(tuple_idx < depth_format_tuples.size());
tuple = depth_format_tuples[tuple_idx];
} else {
tuple = { GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE };
tuple = {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE};
}
glTexImage2D(GL_TEXTURE_2D, 0, tuple.internal_format, width, height, 0,
tuple.format, tuple.type, nullptr);
glTexImage2D(GL_TEXTURE_2D, 0, tuple.internal_format, width, height, 0, tuple.format,
tuple.type, nullptr);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, 0);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
@ -217,7 +239,8 @@ static void AllocateSurfaceTexture(GLuint texture, CachedSurface::PixelFormat pi
}
MICROPROFILE_DEFINE(OpenGL_SurfaceUpload, "OpenGL", "Surface Upload", MP_RGB(128, 64, 192));
CachedSurface* RasterizerCacheOpenGL::GetSurface(const CachedSurface& params, bool match_res_scale, bool load_if_create) {
CachedSurface* RasterizerCacheOpenGL::GetSurface(const CachedSurface& params, bool match_res_scale,
bool load_if_create) {
using PixelFormat = CachedSurface::PixelFormat;
using SurfaceType = CachedSurface::SurfaceType;
@ -225,29 +248,31 @@ CachedSurface* RasterizerCacheOpenGL::GetSurface(const CachedSurface& params, bo
return nullptr;
}
u32 params_size = params.width * params.height * CachedSurface::GetFormatBpp(params.pixel_format) / 8;
u32 params_size =
params.width * params.height * CachedSurface::GetFormatBpp(params.pixel_format) / 8;
// Check for an exact match in existing surfaces
CachedSurface* best_exact_surface = nullptr;
float exact_surface_goodness = -1.f;
auto surface_interval = boost::icl::interval<PAddr>::right_open(params.addr, params.addr + params_size);
auto surface_interval =
boost::icl::interval<PAddr>::right_open(params.addr, params.addr + params_size);
auto range = surface_cache.equal_range(surface_interval);
for (auto it = range.first; it != range.second; ++it) {
for (auto it2 = it->second.begin(); it2 != it->second.end(); ++it2) {
CachedSurface* surface = it2->get();
// Check if the request matches the surface exactly
if (params.addr == surface->addr &&
params.width == surface->width && params.height == surface->height &&
params.pixel_format == surface->pixel_format)
{
if (params.addr == surface->addr && params.width == surface->width &&
params.height == surface->height && params.pixel_format == surface->pixel_format) {
// Make sure optional param-matching criteria are fulfilled
bool tiling_match = (params.is_tiled == surface->is_tiled);
bool res_scale_match = (params.res_scale_width == surface->res_scale_width && params.res_scale_height == surface->res_scale_height);
bool res_scale_match = (params.res_scale_width == surface->res_scale_width &&
params.res_scale_height == surface->res_scale_height);
if (!match_res_scale || res_scale_match) {
// Prioritize same-tiling and highest resolution surfaces
float match_goodness = (float)tiling_match + surface->res_scale_width * surface->res_scale_height;
float match_goodness =
(float)tiling_match + surface->res_scale_width * surface->res_scale_height;
if (match_goodness > exact_surface_goodness || surface->dirty) {
exact_surface_goodness = match_goodness;
best_exact_surface = surface;
@ -288,9 +313,11 @@ CachedSurface* RasterizerCacheOpenGL::GetSurface(const CachedSurface& params, bo
if (!load_if_create) {
// Don't load any data; just allocate the surface's texture
AllocateSurfaceTexture(new_surface->texture.handle, new_surface->pixel_format, new_surface->GetScaledWidth(), new_surface->GetScaledHeight());
AllocateSurfaceTexture(new_surface->texture.handle, new_surface->pixel_format,
new_surface->GetScaledWidth(), new_surface->GetScaledHeight());
} else {
// TODO: Consider attempting subrect match in existing surfaces and direct blit here instead of memory upload below if that's a common scenario in some game
// TODO: Consider attempting subrect match in existing surfaces and direct blit here instead
// of memory upload below if that's a common scenario in some game
Memory::RasterizerFlushRegion(params.addr, params_size);
@ -318,7 +345,7 @@ CachedSurface* RasterizerCacheOpenGL::GetSurface(const CachedSurface& params, bo
tuple = fb_format_tuples[(unsigned int)params.pixel_format];
} else {
// Texture
tuple = { GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE };
tuple = {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE};
}
std::vector<Math::Vec4<u8>> tex_buffer(params.width * params.height);
@ -326,19 +353,23 @@ CachedSurface* RasterizerCacheOpenGL::GetSurface(const CachedSurface& params, bo
Pica::DebugUtils::TextureInfo tex_info;
tex_info.width = params.width;
tex_info.height = params.height;
tex_info.stride = params.width * CachedSurface::GetFormatBpp(params.pixel_format) / 8;
tex_info.stride =
params.width * CachedSurface::GetFormatBpp(params.pixel_format) / 8;
tex_info.format = (Pica::Regs::TextureFormat)params.pixel_format;
tex_info.physical_address = params.addr;
for (unsigned y = 0; y < params.height; ++y) {
for (unsigned x = 0; x < params.width; ++x) {
tex_buffer[x + params.width * y] = Pica::DebugUtils::LookupTexture(texture_src_data, x, params.height - 1 - y, tex_info);
tex_buffer[x + params.width * y] = Pica::DebugUtils::LookupTexture(
texture_src_data, x, params.height - 1 - y, tex_info);
}
}
glTexImage2D(GL_TEXTURE_2D, 0, tuple.internal_format, params.width, params.height, 0, GL_RGBA, GL_UNSIGNED_BYTE, tex_buffer.data());
glTexImage2D(GL_TEXTURE_2D, 0, tuple.internal_format, params.width, params.height,
0, GL_RGBA, GL_UNSIGNED_BYTE, tex_buffer.data());
} else {
// Depth/Stencil formats need special treatment since they aren't sampleable using LookupTexture and can't use RGBA format
// Depth/Stencil formats need special treatment since they aren't sampleable using
// LookupTexture and can't use RGBA format
size_t tuple_idx = (size_t)params.pixel_format - 14;
ASSERT(tuple_idx < depth_format_tuples.size());
const FormatTuple& tuple = depth_format_tuples[tuple_idx];
@ -350,14 +381,18 @@ CachedSurface* RasterizerCacheOpenGL::GetSurface(const CachedSurface& params, bo
u32 gl_bytes_per_pixel = use_4bpp ? 4 : bytes_per_pixel;
std::vector<u8> temp_fb_depth_buffer(params.width * params.height * gl_bytes_per_pixel);
std::vector<u8> temp_fb_depth_buffer(params.width * params.height *
gl_bytes_per_pixel);
u8* temp_fb_depth_buffer_ptr = use_4bpp ? temp_fb_depth_buffer.data() + 1 : temp_fb_depth_buffer.data();
u8* temp_fb_depth_buffer_ptr =
use_4bpp ? temp_fb_depth_buffer.data() + 1 : temp_fb_depth_buffer.data();
MortonCopyPixels(params.pixel_format, params.width, params.height, bytes_per_pixel, gl_bytes_per_pixel, texture_src_data, temp_fb_depth_buffer_ptr, true);
MortonCopyPixels(params.pixel_format, params.width, params.height, bytes_per_pixel,
gl_bytes_per_pixel, texture_src_data, temp_fb_depth_buffer_ptr,
true);
glTexImage2D(GL_TEXTURE_2D, 0, tuple.internal_format, params.width, params.height, 0,
tuple.format, tuple.type, temp_fb_depth_buffer.data());
glTexImage2D(GL_TEXTURE_2D, 0, tuple.internal_format, params.width, params.height,
0, tuple.format, tuple.type, temp_fb_depth_buffer.data());
}
}
glPixelStorei(GL_UNPACK_ROW_LENGTH, 0);
@ -367,10 +402,13 @@ CachedSurface* RasterizerCacheOpenGL::GetSurface(const CachedSurface& params, bo
OGLTexture scaled_texture;
scaled_texture.Create();
AllocateSurfaceTexture(scaled_texture.handle, new_surface->pixel_format, new_surface->GetScaledWidth(), new_surface->GetScaledHeight());
BlitTextures(new_surface->texture.handle, scaled_texture.handle, CachedSurface::GetFormatType(new_surface->pixel_format),
MathUtil::Rectangle<int>(0, 0, new_surface->width, new_surface->height),
MathUtil::Rectangle<int>(0, 0, new_surface->GetScaledWidth(), new_surface->GetScaledHeight()));
AllocateSurfaceTexture(scaled_texture.handle, new_surface->pixel_format,
new_surface->GetScaledWidth(), new_surface->GetScaledHeight());
BlitTextures(new_surface->texture.handle, scaled_texture.handle,
CachedSurface::GetFormatType(new_surface->pixel_format),
MathUtil::Rectangle<int>(0, 0, new_surface->width, new_surface->height),
MathUtil::Rectangle<int>(0, 0, new_surface->GetScaledWidth(),
new_surface->GetScaledHeight()));
new_surface->texture.Release();
new_surface->texture.handle = scaled_texture.handle;
@ -389,11 +427,15 @@ CachedSurface* RasterizerCacheOpenGL::GetSurface(const CachedSurface& params, bo
}
Memory::RasterizerMarkRegionCached(new_surface->addr, new_surface->size, 1);
surface_cache.add(std::make_pair(boost::icl::interval<PAddr>::right_open(new_surface->addr, new_surface->addr + new_surface->size), std::set<std::shared_ptr<CachedSurface>>({ new_surface })));
surface_cache.add(std::make_pair(boost::icl::interval<PAddr>::right_open(
new_surface->addr, new_surface->addr + new_surface->size),
std::set<std::shared_ptr<CachedSurface>>({new_surface})));
return new_surface.get();
}
CachedSurface* RasterizerCacheOpenGL::GetSurfaceRect(const CachedSurface& params, bool match_res_scale, bool load_if_create, MathUtil::Rectangle<int>& out_rect) {
CachedSurface* RasterizerCacheOpenGL::GetSurfaceRect(const CachedSurface& params,
bool match_res_scale, bool load_if_create,
MathUtil::Rectangle<int>& out_rect) {
if (params.addr == 0) {
return nullptr;
}
@ -405,7 +447,8 @@ CachedSurface* RasterizerCacheOpenGL::GetSurfaceRect(const CachedSurface& params
CachedSurface* best_subrect_surface = nullptr;
float subrect_surface_goodness = -1.f;
auto surface_interval = boost::icl::interval<PAddr>::right_open(params.addr, params.addr + params_size);
auto surface_interval =
boost::icl::interval<PAddr>::right_open(params.addr, params.addr + params_size);
auto cache_upper_bound = surface_cache.upper_bound(surface_interval);
for (auto it = surface_cache.lower_bound(surface_interval); it != cache_upper_bound; ++it) {
for (auto it2 = it->second.begin(); it2 != it->second.end(); ++it2) {
@ -414,14 +457,15 @@ CachedSurface* RasterizerCacheOpenGL::GetSurfaceRect(const CachedSurface& params
// Check if the request is contained in the surface
if (params.addr >= surface->addr &&
params.addr + params_size - 1 <= surface->addr + surface->size - 1 &&
params.pixel_format == surface->pixel_format)
{
params.pixel_format == surface->pixel_format) {
// Make sure optional param-matching criteria are fulfilled
bool tiling_match = (params.is_tiled == surface->is_tiled);
bool res_scale_match = (params.res_scale_width == surface->res_scale_width && params.res_scale_height == surface->res_scale_height);
bool res_scale_match = (params.res_scale_width == surface->res_scale_width &&
params.res_scale_height == surface->res_scale_height);
if (!match_res_scale || res_scale_match) {
// Prioritize same-tiling and highest resolution surfaces
float match_goodness = (float)tiling_match + surface->res_scale_width * surface->res_scale_height;
float match_goodness =
(float)tiling_match + surface->res_scale_width * surface->res_scale_height;
if (match_goodness > subrect_surface_goodness || surface->dirty) {
subrect_surface_goodness = match_goodness;
best_subrect_surface = surface;
@ -433,7 +477,8 @@ CachedSurface* RasterizerCacheOpenGL::GetSurfaceRect(const CachedSurface& params
// Return the best subrect surface if found
if (best_subrect_surface != nullptr) {
unsigned int bytes_per_pixel = (CachedSurface::GetFormatBpp(best_subrect_surface->pixel_format) / 8);
unsigned int bytes_per_pixel =
(CachedSurface::GetFormatBpp(best_subrect_surface->pixel_format) / 8);
int x0, y0;
@ -452,7 +497,9 @@ CachedSurface* RasterizerCacheOpenGL::GetSurfaceRect(const CachedSurface& params
y0 = begin_tile_index / tiles_per_row * 8;
// Tiled surfaces are flipped vertically in the rasterizer vs. 3DS memory.
out_rect = MathUtil::Rectangle<int>(x0, best_subrect_surface->height - y0, x0 + params.width, best_subrect_surface->height - (y0 + params.height));
out_rect =
MathUtil::Rectangle<int>(x0, best_subrect_surface->height - y0, x0 + params.width,
best_subrect_surface->height - (y0 + params.height));
}
out_rect.left = (int)(out_rect.left * best_subrect_surface->res_scale_width);
@ -465,16 +512,20 @@ CachedSurface* RasterizerCacheOpenGL::GetSurfaceRect(const CachedSurface& params
// No subrect found - create and return a new surface
if (!params.is_tiled) {
out_rect = MathUtil::Rectangle<int>(0, 0, (int)(params.width * params.res_scale_width), (int)(params.height * params.res_scale_height));
out_rect = MathUtil::Rectangle<int>(0, 0, (int)(params.width * params.res_scale_width),
(int)(params.height * params.res_scale_height));
} else {
out_rect = MathUtil::Rectangle<int>(0, (int)(params.height * params.res_scale_height), (int)(params.width * params.res_scale_width), 0);
out_rect = MathUtil::Rectangle<int>(0, (int)(params.height * params.res_scale_height),
(int)(params.width * params.res_scale_width), 0);
}
return GetSurface(params, match_res_scale, load_if_create);
}
CachedSurface* RasterizerCacheOpenGL::GetTextureSurface(const Pica::Regs::FullTextureConfig& config) {
Pica::DebugUtils::TextureInfo info = Pica::DebugUtils::TextureInfo::FromPicaRegister(config.config, config.format);
CachedSurface*
RasterizerCacheOpenGL::GetTextureSurface(const Pica::Regs::FullTextureConfig& config) {
Pica::DebugUtils::TextureInfo info =
Pica::DebugUtils::TextureInfo::FromPicaRegister(config.config, config.format);
CachedSurface params;
params.addr = info.physical_address;
@ -485,20 +536,28 @@ CachedSurface* RasterizerCacheOpenGL::GetTextureSurface(const Pica::Regs::FullTe
return GetSurface(params, false, true);
}
std::tuple<CachedSurface*, CachedSurface*, MathUtil::Rectangle<int>> RasterizerCacheOpenGL::GetFramebufferSurfaces(const Pica::Regs::FramebufferConfig& config) {
std::tuple<CachedSurface*, CachedSurface*, MathUtil::Rectangle<int>>
RasterizerCacheOpenGL::GetFramebufferSurfaces(const Pica::Regs::FramebufferConfig& config) {
const auto& regs = Pica::g_state.regs;
// Make sur that framebuffers don't overlap if both color and depth are being used
u32 fb_area = config.GetWidth() * config.GetHeight();
bool framebuffers_overlap = config.GetColorBufferPhysicalAddress() != 0 &&
config.GetDepthBufferPhysicalAddress() != 0 &&
MathUtil::IntervalsIntersect(config.GetColorBufferPhysicalAddress(), fb_area * GPU::Regs::BytesPerPixel(GPU::Regs::PixelFormat(config.color_format.Value())),
config.GetDepthBufferPhysicalAddress(), fb_area * Pica::Regs::BytesPerDepthPixel(config.depth_format));
bool framebuffers_overlap =
config.GetColorBufferPhysicalAddress() != 0 &&
config.GetDepthBufferPhysicalAddress() != 0 &&
MathUtil::IntervalsIntersect(
config.GetColorBufferPhysicalAddress(),
fb_area * GPU::Regs::BytesPerPixel(GPU::Regs::PixelFormat(config.color_format.Value())),
config.GetDepthBufferPhysicalAddress(),
fb_area * Pica::Regs::BytesPerDepthPixel(config.depth_format));
bool using_color_fb = config.GetColorBufferPhysicalAddress() != 0;
bool using_depth_fb = config.GetDepthBufferPhysicalAddress() != 0 && (regs.output_merger.depth_test_enable || regs.output_merger.depth_write_enable || !framebuffers_overlap);
bool using_depth_fb = config.GetDepthBufferPhysicalAddress() != 0 &&
(regs.output_merger.depth_test_enable ||
regs.output_merger.depth_write_enable || !framebuffers_overlap);
if (framebuffers_overlap && using_color_fb && using_depth_fb) {
LOG_CRITICAL(Render_OpenGL, "Color and depth framebuffer memory regions overlap; overlapping framebuffers not supported!");
LOG_CRITICAL(Render_OpenGL, "Color and depth framebuffer memory regions overlap; "
"overlapping framebuffers not supported!");
using_depth_fb = false;
}
@ -512,8 +571,10 @@ std::tuple<CachedSurface*, CachedSurface*, MathUtil::Rectangle<int>> RasterizerC
auto layout = VideoCore::g_emu_window->GetFramebufferLayout();
// Assume same scaling factor for top and bottom screens
color_params.res_scale_width = depth_params.res_scale_width = (float)layout.top_screen.GetWidth() / VideoCore::kScreenTopWidth;
color_params.res_scale_height = depth_params.res_scale_height = (float)layout.top_screen.GetHeight() / VideoCore::kScreenTopHeight;
color_params.res_scale_width = depth_params.res_scale_width =
(float)layout.top_screen.GetWidth() / VideoCore::kScreenTopWidth;
color_params.res_scale_height = depth_params.res_scale_height =
(float)layout.top_screen.GetHeight() / VideoCore::kScreenTopHeight;
}
color_params.addr = config.GetColorBufferPhysicalAddress();
@ -523,22 +584,28 @@ std::tuple<CachedSurface*, CachedSurface*, MathUtil::Rectangle<int>> RasterizerC
depth_params.pixel_format = CachedSurface::PixelFormatFromDepthFormat(config.depth_format);
MathUtil::Rectangle<int> color_rect;
CachedSurface* color_surface = using_color_fb ? GetSurfaceRect(color_params, true, true, color_rect) : nullptr;
CachedSurface* color_surface =
using_color_fb ? GetSurfaceRect(color_params, true, true, color_rect) : nullptr;
MathUtil::Rectangle<int> depth_rect;
CachedSurface* depth_surface = using_depth_fb ? GetSurfaceRect(depth_params, true, true, depth_rect) : nullptr;
CachedSurface* depth_surface =
using_depth_fb ? GetSurfaceRect(depth_params, true, true, depth_rect) : nullptr;
// Sanity check to make sure found surfaces aren't the same
if (using_depth_fb && using_color_fb && color_surface == depth_surface) {
LOG_CRITICAL(Render_OpenGL, "Color and depth framebuffer surfaces overlap; overlapping surfaces not supported!");
LOG_CRITICAL(
Render_OpenGL,
"Color and depth framebuffer surfaces overlap; overlapping surfaces not supported!");
using_depth_fb = false;
depth_surface = nullptr;
}
MathUtil::Rectangle<int> rect;
if (color_surface != nullptr && depth_surface != nullptr && (depth_rect.left != color_rect.left || depth_rect.top != color_rect.top)) {
// Can't specify separate color and depth viewport offsets in OpenGL, so re-zero both if they don't match
if (color_surface != nullptr && depth_surface != nullptr &&
(depth_rect.left != color_rect.left || depth_rect.top != color_rect.top)) {
// Can't specify separate color and depth viewport offsets in OpenGL, so re-zero both if
// they don't match
if (color_rect.left != 0 || color_rect.top != 0) {
color_surface = GetSurface(color_params, true, true);
}
@ -548,9 +615,13 @@ std::tuple<CachedSurface*, CachedSurface*, MathUtil::Rectangle<int>> RasterizerC
}
if (!color_surface->is_tiled) {
rect = MathUtil::Rectangle<int>(0, 0, (int)(color_params.width * color_params.res_scale_width), (int)(color_params.height * color_params.res_scale_height));
rect = MathUtil::Rectangle<int>(
0, 0, (int)(color_params.width * color_params.res_scale_width),
(int)(color_params.height * color_params.res_scale_height));
} else {
rect = MathUtil::Rectangle<int>(0, (int)(color_params.height * color_params.res_scale_height), (int)(color_params.width * color_params.res_scale_width), 0);
rect = MathUtil::Rectangle<int>(
0, (int)(color_params.height * color_params.res_scale_height),
(int)(color_params.width * color_params.res_scale_width), 0);
}
} else if (color_surface != nullptr) {
rect = color_rect;
@ -564,7 +635,8 @@ std::tuple<CachedSurface*, CachedSurface*, MathUtil::Rectangle<int>> RasterizerC
}
CachedSurface* RasterizerCacheOpenGL::TryGetFillSurface(const GPU::Regs::MemoryFillConfig& config) {
auto surface_interval = boost::icl::interval<PAddr>::right_open(config.GetStartAddress(), config.GetEndAddress());
auto surface_interval =
boost::icl::interval<PAddr>::right_open(config.GetStartAddress(), config.GetEndAddress());
auto range = surface_cache.equal_range(surface_interval);
for (auto it = range.first; it != range.second; ++it) {
for (auto it2 = it->second.begin(); it2 != it->second.end(); ++it2) {
@ -581,8 +653,9 @@ CachedSurface* RasterizerCacheOpenGL::TryGetFillSurface(const GPU::Regs::MemoryF
if (surface->addr == config.GetStartAddress() &&
CachedSurface::GetFormatBpp(surface->pixel_format) == bits_per_value &&
(surface->width * surface->height * CachedSurface::GetFormatBpp(surface->pixel_format) / 8) == (config.GetEndAddress() - config.GetStartAddress()))
{
(surface->width * surface->height *
CachedSurface::GetFormatBpp(surface->pixel_format) / 8) ==
(config.GetEndAddress() - config.GetStartAddress())) {
return surface;
}
}
@ -617,8 +690,11 @@ void RasterizerCacheOpenGL::FlushSurface(CachedSurface* surface) {
if (surface->res_scale_width != 1.f || surface->res_scale_height != 1.f) {
unscaled_tex.Create();
AllocateSurfaceTexture(unscaled_tex.handle, surface->pixel_format, surface->width, surface->height);
BlitTextures(surface->texture.handle, unscaled_tex.handle, CachedSurface::GetFormatType(surface->pixel_format),
AllocateSurfaceTexture(unscaled_tex.handle, surface->pixel_format, surface->width,
surface->height);
BlitTextures(
surface->texture.handle, unscaled_tex.handle,
CachedSurface::GetFormatType(surface->pixel_format),
MathUtil::Rectangle<int>(0, 0, surface->GetScaledWidth(), surface->GetScaledHeight()),
MathUtil::Rectangle<int>(0, 0, surface->width, surface->height));
@ -648,10 +724,14 @@ void RasterizerCacheOpenGL::FlushSurface(CachedSurface* surface) {
glGetTexImage(GL_TEXTURE_2D, 0, tuple.format, tuple.type, temp_gl_buffer.data());
// Directly copy pixels. Internal OpenGL color formats are consistent so no conversion is necessary.
MortonCopyPixels(surface->pixel_format, surface->width, surface->height, bytes_per_pixel, bytes_per_pixel, dst_buffer, temp_gl_buffer.data(), false);
// Directly copy pixels. Internal OpenGL color formats are consistent so no conversion
// is necessary.
MortonCopyPixels(surface->pixel_format, surface->width, surface->height,
bytes_per_pixel, bytes_per_pixel, dst_buffer, temp_gl_buffer.data(),
false);
} else {
// Depth/Stencil formats need special treatment since they aren't sampleable using LookupTexture and can't use RGBA format
// Depth/Stencil formats need special treatment since they aren't sampleable using
// LookupTexture and can't use RGBA format
size_t tuple_idx = (size_t)surface->pixel_format - 14;
ASSERT(tuple_idx < depth_format_tuples.size());
const FormatTuple& tuple = depth_format_tuples[tuple_idx];
@ -669,7 +749,9 @@ void RasterizerCacheOpenGL::FlushSurface(CachedSurface* surface) {
u8* temp_gl_buffer_ptr = use_4bpp ? temp_gl_buffer.data() + 1 : temp_gl_buffer.data();
MortonCopyPixels(surface->pixel_format, surface->width, surface->height, bytes_per_pixel, gl_bytes_per_pixel, dst_buffer, temp_gl_buffer_ptr, false);
MortonCopyPixels(surface->pixel_format, surface->width, surface->height,
bytes_per_pixel, gl_bytes_per_pixel, dst_buffer, temp_gl_buffer_ptr,
false);
}
}
glPixelStorei(GL_PACK_ROW_LENGTH, 0);
@ -680,7 +762,8 @@ void RasterizerCacheOpenGL::FlushSurface(CachedSurface* surface) {
cur_state.Apply();
}
void RasterizerCacheOpenGL::FlushRegion(PAddr addr, u32 size, const CachedSurface* skip_surface, bool invalidate) {
void RasterizerCacheOpenGL::FlushRegion(PAddr addr, u32 size, const CachedSurface* skip_surface,
bool invalidate) {
if (size == 0) {
return;
}
@ -691,8 +774,11 @@ void RasterizerCacheOpenGL::FlushRegion(PAddr addr, u32 size, const CachedSurfac
auto surface_interval = boost::icl::interval<PAddr>::right_open(addr, addr + size);
auto cache_upper_bound = surface_cache.upper_bound(surface_interval);
for (auto it = surface_cache.lower_bound(surface_interval); it != cache_upper_bound; ++it) {
std::copy_if(it->second.begin(), it->second.end(), std::inserter(touching_surfaces, touching_surfaces.end()),
[skip_surface](std::shared_ptr<CachedSurface> surface) { return (surface.get() != skip_surface); });
std::copy_if(it->second.begin(), it->second.end(),
std::inserter(touching_surfaces, touching_surfaces.end()),
[skip_surface](std::shared_ptr<CachedSurface> surface) {
return (surface.get() != skip_surface);
});
}
// Flush and invalidate surfaces
@ -700,7 +786,10 @@ void RasterizerCacheOpenGL::FlushRegion(PAddr addr, u32 size, const CachedSurfac
FlushSurface(surface.get());
if (invalidate) {
Memory::RasterizerMarkRegionCached(surface->addr, surface->size, -1);
surface_cache.subtract(std::make_pair(boost::icl::interval<PAddr>::right_open(surface->addr, surface->addr + surface->size), std::set<std::shared_ptr<CachedSurface>>({ surface })));
surface_cache.subtract(
std::make_pair(boost::icl::interval<PAddr>::right_open(
surface->addr, surface->addr + surface->size),
std::set<std::shared_ptr<CachedSurface>>({surface})));
}
}
}

78
src/video_core/renderer_opengl/gl_rasterizer_cache.h
View file

@ -22,7 +22,8 @@
#include "video_core/renderer_opengl/gl_resource_manager.h"
namespace MathUtil {
template <class T> struct Rectangle;
template <class T>
struct Rectangle;
}
struct CachedSurface;
@ -32,38 +33,38 @@ using SurfaceCache = boost::icl::interval_map<PAddr, std::set<std::shared_ptr<Ca
struct CachedSurface {
enum class PixelFormat {
// First 5 formats are shared between textures and color buffers
RGBA8 = 0,
RGB8 = 1,
RGB5A1 = 2,
RGB565 = 3,
RGBA4 = 4,
RGBA8 = 0,
RGB8 = 1,
RGB5A1 = 2,
RGB565 = 3,
RGBA4 = 4,
// Texture-only formats
IA8 = 5,
RG8 = 6,
I8 = 7,
A8 = 8,
IA4 = 9,
I4 = 10,
A4 = 11,
ETC1 = 12,
ETC1A4 = 13,
IA8 = 5,
RG8 = 6,
I8 = 7,
A8 = 8,
IA4 = 9,
I4 = 10,
A4 = 11,
ETC1 = 12,
ETC1A4 = 13,
// Depth buffer-only formats
D16 = 14,
D16 = 14,
// gap
D24 = 16,
D24S8 = 17,
D24 = 16,
D24S8 = 17,
Invalid = 255,
Invalid = 255,
};
enum class SurfaceType {
Color = 0,
Texture = 1,
Depth = 2,
Color = 0,
Texture = 1,
Depth = 2,
DepthStencil = 3,
Invalid = 4,
Invalid = 4,
};
static unsigned int GetFormatBpp(CachedSurface::PixelFormat format) {
@ -101,7 +102,8 @@ struct CachedSurface {
}
static PixelFormat PixelFormatFromDepthFormat(Pica::Regs::DepthFormat format) {
return ((unsigned int)format < 4) ? (PixelFormat)((unsigned int)format + 14) : PixelFormat::Invalid;
return ((unsigned int)format < 4) ? (PixelFormat)((unsigned int)format + 14)
: PixelFormat::Invalid;
}
static PixelFormat PixelFormatFromGPUPixelFormat(GPU::Regs::PixelFormat format) {
@ -120,7 +122,8 @@ struct CachedSurface {
SurfaceType a_type = GetFormatType(pixel_format_a);
SurfaceType b_type = GetFormatType(pixel_format_b);
if ((a_type == SurfaceType::Color || a_type == SurfaceType::Texture) && (b_type == SurfaceType::Color || b_type == SurfaceType::Texture)) {
if ((a_type == SurfaceType::Color || a_type == SurfaceType::Texture) &&
(b_type == SurfaceType::Color || b_type == SurfaceType::Texture)) {
return true;
}
@ -187,22 +190,30 @@ public:
~RasterizerCacheOpenGL();
/// Blits one texture to another
bool BlitTextures(GLuint src_tex, GLuint dst_tex, CachedSurface::SurfaceType type, const MathUtil::Rectangle<int>& src_rect, const MathUtil::Rectangle<int>& dst_rect);
bool BlitTextures(GLuint src_tex, GLuint dst_tex, CachedSurface::SurfaceType type,
const MathUtil::Rectangle<int>& src_rect,
const MathUtil::Rectangle<int>& dst_rect);
/// Attempt to blit one surface's texture to another
bool TryBlitSurfaces(CachedSurface* src_surface, const MathUtil::Rectangle<int>& src_rect, CachedSurface* dst_surface, const MathUtil::Rectangle<int>& dst_rect);
bool TryBlitSurfaces(CachedSurface* src_surface, const MathUtil::Rectangle<int>& src_rect,
CachedSurface* dst_surface, const MathUtil::Rectangle<int>& dst_rect);
/// Loads a texture from 3DS memory to OpenGL and caches it (if not already cached)
CachedSurface* GetSurface(const CachedSurface& params, bool match_res_scale, bool load_if_create);
CachedSurface* GetSurface(const CachedSurface& params, bool match_res_scale,
bool load_if_create);
/// Attempt to find a subrect (resolution scaled) of a surface, otherwise loads a texture from 3DS memory to OpenGL and caches it (if not already cached)
CachedSurface* GetSurfaceRect(const CachedSurface& params, bool match_res_scale, bool load_if_create, MathUtil::Rectangle<int>& out_rect);
/// Attempt to find a subrect (resolution scaled) of a surface, otherwise loads a texture from
/// 3DS memory to OpenGL and caches it (if not already cached)
CachedSurface* GetSurfaceRect(const CachedSurface& params, bool match_res_scale,
bool load_if_create, MathUtil::Rectangle<int>& out_rect);
/// Gets a surface based on the texture configuration
CachedSurface* GetTextureSurface(const Pica::Regs::FullTextureConfig& config);
/// Gets the color and depth surfaces and rect (resolution scaled) based on the framebuffer configuration
std::tuple<CachedSurface*, CachedSurface*, MathUtil::Rectangle<int>> GetFramebufferSurfaces(const Pica::Regs::FramebufferConfig& config);
/// Gets the color and depth surfaces and rect (resolution scaled) based on the framebuffer
/// configuration
std::tuple<CachedSurface*, CachedSurface*, MathUtil::Rectangle<int>>
GetFramebufferSurfaces(const Pica::Regs::FramebufferConfig& config);
/// Attempt to get a surface that exactly matches the fill region and format
CachedSurface* TryGetFillSurface(const GPU::Regs::MemoryFillConfig& config);
@ -210,7 +221,8 @@ public:
/// Write the surface back to memory
void FlushSurface(CachedSurface* surface);
/// Write any cached resources overlapping the region back to memory (if dirty) and optionally invalidate them in the cache
/// Write any cached resources overlapping the region back to memory (if dirty) and optionally
/// invalidate them in the cache
void FlushRegion(PAddr addr, u32 size, const CachedSurface* skip_surface, bool invalidate);
/// Flush all cached resources tracked by this cache manager

114
src/video_core/renderer_opengl/gl_resource_manager.h
View file

@ -16,19 +16,28 @@
class OGLTexture : private NonCopyable {
public:
OGLTexture() = default;
OGLTexture(OGLTexture&& o) { std::swap(handle, o.handle); }
~OGLTexture() { Release(); }
OGLTexture& operator=(OGLTexture&& o) { std::swap(handle, o.handle); return *this; }
OGLTexture(OGLTexture&& o) {
std::swap(handle, o.handle);
}
~OGLTexture() {
Release();
}
OGLTexture& operator=(OGLTexture&& o) {
std::swap(handle, o.handle);
return *this;
}
/// Creates a new internal OpenGL resource and stores the handle
void Create() {
if (handle != 0) return;
if (handle != 0)
return;
glGenTextures(1, &handle);
}
/// Deletes the internal OpenGL resource
void Release() {
if (handle == 0) return;
if (handle == 0)
return;
glDeleteTextures(1, &handle);
OpenGLState::ResetTexture(handle);
handle = 0;
@ -40,19 +49,28 @@ public:
class OGLSampler : private NonCopyable {
public:
OGLSampler() = default;
OGLSampler(OGLSampler&& o) { std::swap(handle, o.handle); }
~OGLSampler() { Release(); }
OGLSampler& operator=(OGLSampler&& o) { std::swap(handle, o.handle); return *this; }
OGLSampler(OGLSampler&& o) {
std::swap(handle, o.handle);
}
~OGLSampler() {
Release();
}
OGLSampler& operator=(OGLSampler&& o) {
std::swap(handle, o.handle);
return *this;
}
/// Creates a new internal OpenGL resource and stores the handle
void Create() {
if (handle != 0) return;
if (handle != 0)
return;
glGenSamplers(1, &handle);
}
/// Deletes the internal OpenGL resource
void Release() {
if (handle == 0) return;
if (handle == 0)
return;
glDeleteSamplers(1, &handle);
OpenGLState::ResetSampler(handle);
handle = 0;
@ -64,19 +82,28 @@ public:
class OGLShader : private NonCopyable {
public:
OGLShader() = default;
OGLShader(OGLShader&& o) { std::swap(handle, o.handle); }
~OGLShader() { Release(); }
OGLShader& operator=(OGLShader&& o) { std::swap(handle, o.handle); return *this; }
OGLShader(OGLShader&& o) {
std::swap(handle, o.handle);
}
~OGLShader() {
Release();
}
OGLShader& operator=(OGLShader&& o) {
std::swap(handle, o.handle);
return *this;
}
/// Creates a new internal OpenGL resource and stores the handle
void Create(const char* vert_shader, const char* frag_shader) {
if (handle != 0) return;
if (handle != 0)
return;
handle = GLShader::LoadProgram(vert_shader, frag_shader);
}
/// Deletes the internal OpenGL resource
void Release() {
if (handle == 0) return;
if (handle == 0)
return;
glDeleteProgram(handle);
OpenGLState::ResetProgram(handle);
handle = 0;
@ -88,19 +115,28 @@ public:
class OGLBuffer : private NonCopyable {
public:
OGLBuffer() = default;
OGLBuffer(OGLBuffer&& o) { std::swap(handle, o.handle); }
~OGLBuffer() { Release(); }
OGLBuffer& operator=(OGLBuffer&& o) { std::swap(handle, o.handle); return *this; }
OGLBuffer(OGLBuffer&& o) {
std::swap(handle, o.handle);
}
~OGLBuffer() {
Release();
}
OGLBuffer& operator=(OGLBuffer&& o) {
std::swap(handle, o.handle);
return *this;
}
/// Creates a new internal OpenGL resource and stores the handle
void Create() {
if (handle != 0) return;
if (handle != 0)
return;
glGenBuffers(1, &handle);
}
/// Deletes the internal OpenGL resource
void Release() {
if (handle == 0) return;
if (handle == 0)
return;
glDeleteBuffers(1, &handle);
OpenGLState::ResetBuffer(handle);
handle = 0;
@ -112,19 +148,28 @@ public:
class OGLVertexArray : private NonCopyable {
public:
OGLVertexArray() = default;
OGLVertexArray(OGLVertexArray&& o) { std::swap(handle, o.handle); }
~OGLVertexArray() { Release(); }
OGLVertexArray& operator=(OGLVertexArray&& o) { std::swap(handle, o.handle); return *this; }
OGLVertexArray(OGLVertexArray&& o) {
std::swap(handle, o.handle);
}
~OGLVertexArray() {
Release();
}
OGLVertexArray& operator=(OGLVertexArray&& o) {
std::swap(handle, o.handle);
return *this;
}
/// Creates a new internal OpenGL resource and stores the handle
void Create() {
if (handle != 0) return;
if (handle != 0)
return;
glGenVertexArrays(1, &handle);
}
/// Deletes the internal OpenGL resource
void Release() {
if (handle == 0) return;
if (handle == 0)
return;
glDeleteVertexArrays(1, &handle);
OpenGLState::ResetVertexArray(handle);
handle = 0;
@ -136,19 +181,28 @@ public:
class OGLFramebuffer : private NonCopyable {
public:
OGLFramebuffer() = default;
OGLFramebuffer(OGLFramebuffer&& o) { std::swap(handle, o.handle); }
~OGLFramebuffer() { Release(); }
OGLFramebuffer& operator=(OGLFramebuffer&& o) { std::swap(handle, o.handle); return *this; }
OGLFramebuffer(OGLFramebuffer&& o) {
std::swap(handle, o.handle);
}
~OGLFramebuffer() {
Release();
}
OGLFramebuffer& operator=(OGLFramebuffer&& o) {
std::swap(handle, o.handle);
return *this;
}
/// Creates a new internal OpenGL resource and stores the handle
void Create() {
if (handle != 0) return;
if (handle != 0)
return;
glGenFramebuffers(1, &handle);
}
/// Deletes the internal OpenGL resource
void Release() {
if (handle == 0) return;
if (handle == 0)
return;
glDeleteFramebuffers(1, &handle);
OpenGLState::ResetFramebuffer(handle);
handle = 0;

213
src/video_core/renderer_opengl/gl_shader_gen.cpp
View file

@ -21,19 +21,18 @@ namespace GLShader {
/// Detects if a TEV stage is configured to be skipped (to avoid generating unnecessary code)
static bool IsPassThroughTevStage(const TevStageConfig& stage) {
return (stage.color_op == TevStageConfig::Operation::Replace &&
stage.alpha_op == TevStageConfig::Operation::Replace &&
stage.color_source1 == TevStageConfig::Source::Previous &&
stage.alpha_source1 == TevStageConfig::Source::Previous &&
stage.color_modifier1 == TevStageConfig::ColorModifier::SourceColor &&
stage.alpha_modifier1 == TevStageConfig::AlphaModifier::SourceAlpha &&
stage.GetColorMultiplier() == 1 &&
stage.GetAlphaMultiplier() == 1);
return (stage.color_op == TevStageConfig::Operation::Replace &&
stage.alpha_op == TevStageConfig::Operation::Replace &&
stage.color_source1 == TevStageConfig::Source::Previous &&
stage.alpha_source1 == TevStageConfig::Source::Previous &&
stage.color_modifier1 == TevStageConfig::ColorModifier::SourceColor &&
stage.alpha_modifier1 == TevStageConfig::AlphaModifier::SourceAlpha &&
stage.GetColorMultiplier() == 1 && stage.GetAlphaMultiplier() == 1);
}
/// Writes the specified TEV stage source component(s)
static void AppendSource(std::string& out, const PicaShaderConfig& config, TevStageConfig::Source source,
const std::string& index_name) {
static void AppendSource(std::string& out, const PicaShaderConfig& config,
TevStageConfig::Source source, const std::string& index_name) {
const auto& state = config.state;
using Source = TevStageConfig::Source;
switch (source) {
@ -48,7 +47,7 @@ static void AppendSource(std::string& out, const PicaShaderConfig& config, TevSt
break;
case Source::Texture0:
// Only unit 0 respects the texturing type (according to 3DBrew)
switch(state.texture0_type) {
switch (state.texture0_type) {
case Pica::Regs::TextureConfig::Texture2D:
out += "texture(tex[0], texcoord[0])";
break;
@ -57,7 +56,8 @@ static void AppendSource(std::string& out, const PicaShaderConfig& config, TevSt
break;
default:
out += "texture(tex[0], texcoord[0])";
LOG_CRITICAL(HW_GPU, "Unhandled texture type %x", static_cast<int>(state.texture0_type));
LOG_CRITICAL(HW_GPU, "Unhandled texture type %x",
static_cast<int>(state.texture0_type));
UNIMPLEMENTED();
break;
}
@ -85,8 +85,9 @@ static void AppendSource(std::string& out, const PicaShaderConfig& config, TevSt
}
/// Writes the color components to use for the specified TEV stage color modifier
static void AppendColorModifier(std::string& out, const PicaShaderConfig& config, TevStageConfig::ColorModifier modifier,
TevStageConfig::Source source, const std::string& index_name) {
static void AppendColorModifier(std::string& out, const PicaShaderConfig& config,
TevStageConfig::ColorModifier modifier,
TevStageConfig::Source source, const std::string& index_name) {
using ColorModifier = TevStageConfig::ColorModifier;
switch (modifier) {
case ColorModifier::SourceColor:
@ -142,8 +143,9 @@ static void AppendColorModifier(std::string& out, const PicaShaderConfig& config
}
/// Writes the alpha component to use for the specified TEV stage alpha modifier
static void AppendAlphaModifier(std::string& out, const PicaShaderConfig& config, TevStageConfig::AlphaModifier modifier,
TevStageConfig::Source source, const std::string& index_name) {
static void AppendAlphaModifier(std::string& out, const PicaShaderConfig& config,
TevStageConfig::AlphaModifier modifier,
TevStageConfig::Source source, const std::string& index_name) {
using AlphaModifier = TevStageConfig::AlphaModifier;
switch (modifier) {
case AlphaModifier::SourceAlpha:
@ -191,7 +193,7 @@ static void AppendAlphaModifier(std::string& out, const PicaShaderConfig& config
/// Writes the combiner function for the color components for the specified TEV stage operation
static void AppendColorCombiner(std::string& out, TevStageConfig::Operation operation,
const std::string& variable_name) {
const std::string& variable_name) {
out += "clamp(";
using Operation = TevStageConfig::Operation;
switch (operation) {
@ -208,8 +210,10 @@ static void AppendColorCombiner(std::string& out, TevStageConfig::Operation oper
out += variable_name + "[0] + " + variable_name + "[1] - vec3(0.5)";
break;
case Operation::Lerp:
// TODO(bunnei): Verify if HW actually does this per-component, otherwise we can just use builtin lerp
out += variable_name + "[0] * " + variable_name + "[2] + " + variable_name + "[1] * (vec3(1.0) - " + variable_name + "[2])";
// TODO(bunnei): Verify if HW actually does this per-component, otherwise we can just use
// builtin lerp
out += variable_name + "[0] * " + variable_name + "[2] + " + variable_name +
"[1] * (vec3(1.0) - " + variable_name + "[2])";
break;
case Operation::Subtract:
out += variable_name + "[0] - " + variable_name + "[1]";
@ -218,10 +222,12 @@ static void AppendColorCombiner(std::string& out, TevStageConfig::Operation oper
out += variable_name + "[0] * " + variable_name + "[1] + " + variable_name + "[2]";
break;
case Operation::AddThenMultiply:
out += "min(" + variable_name + "[0] + " + variable_name + "[1], vec3(1.0)) * " + variable_name + "[2]";
out += "min(" + variable_name + "[0] + " + variable_name + "[1], vec3(1.0)) * " +
variable_name + "[2]";
break;
case Operation::Dot3_RGB:
out += "vec3(dot(" + variable_name + "[0] - vec3(0.5), " + variable_name + "[1] - vec3(0.5)) * 4.0)";
out += "vec3(dot(" + variable_name + "[0] - vec3(0.5), " + variable_name +
"[1] - vec3(0.5)) * 4.0)";
break;
default:
out += "vec3(0.0)";
@ -233,7 +239,7 @@ static void AppendColorCombiner(std::string& out, TevStageConfig::Operation oper
/// Writes the combiner function for the alpha component for the specified TEV stage operation
static void AppendAlphaCombiner(std::string& out, TevStageConfig::Operation operation,
const std::string& variable_name) {
const std::string& variable_name) {
out += "clamp(";
using Operation = TevStageConfig::Operation;
switch (operation) {
@ -250,7 +256,8 @@ static void AppendAlphaCombiner(std::string& out, TevStageConfig::Operation oper
out += variable_name + "[0] + " + variable_name + "[1] - 0.5";
break;
case Operation::Lerp:
out += variable_name + "[0] * " + variable_name + "[2] + " + variable_name + "[1] * (1.0 - " + variable_name + "[2])";
out += variable_name + "[0] * " + variable_name + "[2] + " + variable_name +
"[1] * (1.0 - " + variable_name + "[2])";
break;
case Operation::Subtract:
out += variable_name + "[0] - " + variable_name + "[1]";
@ -259,7 +266,8 @@ static void AppendAlphaCombiner(std::string& out, TevStageConfig::Operation oper
out += variable_name + "[0] * " + variable_name + "[1] + " + variable_name + "[2]";
break;
case Operation::AddThenMultiply:
out += "min(" + variable_name + "[0] + " + variable_name + "[1], 1.0) * " + variable_name + "[2]";
out += "min(" + variable_name + "[0] + " + variable_name + "[1], 1.0) * " + variable_name +
"[2]";
break;
default:
out += "0.0";
@ -284,9 +292,10 @@ static void AppendAlphaTestCondition(std::string& out, Regs::CompareFunc func) {
case CompareFunc::LessThan:
case CompareFunc::LessThanOrEqual:
case CompareFunc::GreaterThan:
case CompareFunc::GreaterThanOrEqual:
{
static const char* op[] = { "!=", "==", ">=", ">", "<=", "<", };
case CompareFunc::GreaterThanOrEqual: {
static const char* op[] = {
"!=", "==", ">=", ">", "<=", "<",
};
unsigned index = (unsigned)func - (unsigned)CompareFunc::Equal;
out += "int(last_tex_env_out.a * 255.0f) " + std::string(op[index]) + " alphatest_ref";
break;
@ -301,7 +310,8 @@ static void AppendAlphaTestCondition(std::string& out, Regs::CompareFunc func) {
/// Writes the code to emulate the specified TEV stage
static void WriteTevStage(std::string& out, const PicaShaderConfig& config, unsigned index) {
const auto stage = static_cast<const Pica::Regs::TevStageConfig>(config.state.tev_stages[index]);
const auto stage =
static_cast<const Pica::Regs::TevStageConfig>(config.state.tev_stages[index]);
if (!IsPassThroughTevStage(stage)) {
std::string index_name = std::to_string(index);
@ -330,8 +340,12 @@ static void WriteTevStage(std::string& out, const PicaShaderConfig& config, unsi
out += ";\n";
out += "last_tex_env_out = vec4("
"clamp(color_output_" + index_name + " * " + std::to_string(stage.GetColorMultiplier()) + ".0, vec3(0.0), vec3(1.0)),"
"clamp(alpha_output_" + index_name + " * " + std::to_string(stage.GetAlphaMultiplier()) + ".0, 0.0, 1.0));\n";
"clamp(color_output_" +
index_name + " * " + std::to_string(stage.GetColorMultiplier()) +
".0, vec3(0.0), vec3(1.0)),"
"clamp(alpha_output_" +
index_name + " * " + std::to_string(stage.GetAlphaMultiplier()) +
".0, 0.0, 1.0));\n";
}
out += "combiner_buffer = next_combiner_buffer;\n";
@ -355,13 +369,17 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) {
// Compute fragment normals
if (lighting.bump_mode == Pica::Regs::LightingBumpMode::NormalMap) {
// Bump mapping is enabled using a normal map, read perturbation vector from the selected texture
// Bump mapping is enabled using a normal map, read perturbation vector from the selected
// texture
std::string bump_selector = std::to_string(lighting.bump_selector);
out += "vec3 surface_normal = 2.0 * texture(tex[" + bump_selector + "], texcoord[" + bump_selector + "]).rgb - 1.0;\n";
out += "vec3 surface_normal = 2.0 * texture(tex[" + bump_selector + "], texcoord[" +
bump_selector + "]).rgb - 1.0;\n";
// Recompute Z-component of perturbation if 'renorm' is enabled, this provides a higher precision result
// Recompute Z-component of perturbation if 'renorm' is enabled, this provides a higher
// precision result
if (lighting.bump_renorm) {
std::string val = "(1.0 - (surface_normal.x*surface_normal.x + surface_normal.y*surface_normal.y))";
std::string val =
"(1.0 - (surface_normal.x*surface_normal.x + surface_normal.y*surface_normal.y))";
out += "surface_normal.z = sqrt(max(" + val + ", 0.0));\n";
}
} else if (lighting.bump_mode == Pica::Regs::LightingBumpMode::TangentMap) {
@ -373,7 +391,8 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) {
out += "vec3 surface_normal = vec3(0.0, 0.0, 1.0);\n";
}
// Rotate the surface-local normal by the interpolated normal quaternion to convert it to eyespace
// Rotate the surface-local normal by the interpolated normal quaternion to convert it to
// eyespace
out += "vec3 normal = normalize(quaternion_rotate(normquat, surface_normal));\n";
// Gets the index into the specified lookup table for specular lighting
@ -406,12 +425,14 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) {
if (abs) {
// LUT index is in the range of (0.0, 1.0)
index = lighting.light[light_num].two_sided_diffuse ? "abs(" + index + ")" : "max(" + index + ", 0.f)";
index = lighting.light[light_num].two_sided_diffuse ? "abs(" + index + ")"
: "max(" + index + ", 0.f)";
return "(FLOAT_255 * clamp(" + index + ", 0.0, 1.0))";
} else {
// LUT index is in the range of (-1.0, 1.0)
index = "clamp(" + index + ", -1.0, 1.0)";
return "(FLOAT_255 * ((" + index + " < 0) ? " + index + " + 2.0 : " + index + ") / 2.0)";
return "(FLOAT_255 * ((" + index + " < 0) ? " + index + " + 2.0 : " + index +
") / 2.0)";
}
return std::string();
@ -434,52 +455,74 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) {
else
out += "light_vector = normalize(" + light_src + ".position + view);\n";
// Compute dot product of light_vector and normal, adjust if lighting is one-sided or two-sided
std::string dot_product = light_config.two_sided_diffuse ? "abs(dot(light_vector, normal))" : "max(dot(light_vector, normal), 0.0)";
// Compute dot product of light_vector and normal, adjust if lighting is one-sided or
// two-sided
std::string dot_product = light_config.two_sided_diffuse
? "abs(dot(light_vector, normal))"
: "max(dot(light_vector, normal), 0.0)";
// If enabled, compute distance attenuation value
std::string dist_atten = "1.0";
if (light_config.dist_atten_enable) {
std::string index = "(" + light_src + ".dist_atten_scale * length(-view - " + light_src + ".position) + " + light_src + ".dist_atten_bias)";
std::string index = "(" + light_src + ".dist_atten_scale * length(-view - " +
light_src + ".position) + " + light_src + ".dist_atten_bias)";
index = "((clamp(" + index + ", 0.0, FLOAT_255)))";
const unsigned lut_num = ((unsigned)Regs::LightingSampler::DistanceAttenuation + light_config.num);
const unsigned lut_num =
((unsigned)Regs::LightingSampler::DistanceAttenuation + light_config.num);
dist_atten = GetLutValue((Regs::LightingSampler)lut_num, index);
}
// If enabled, clamp specular component if lighting result is negative
std::string clamp_highlights = lighting.clamp_highlights ? "(dot(light_vector, normal) <= 0.0 ? 0.0 : 1.0)" : "1.0";
std::string clamp_highlights =
lighting.clamp_highlights ? "(dot(light_vector, normal) <= 0.0 ? 0.0 : 1.0)" : "1.0";
// Specular 0 component
std::string d0_lut_value = "1.0";
if (lighting.lut_d0.enable && Pica::Regs::IsLightingSamplerSupported(lighting.config, Pica::Regs::LightingSampler::Distribution0)) {
if (lighting.lut_d0.enable &&
Pica::Regs::IsLightingSamplerSupported(lighting.config,
Pica::Regs::LightingSampler::Distribution0)) {
// Lookup specular "distribution 0" LUT value
std::string index = GetLutIndex(light_config.num, lighting.lut_d0.type, lighting.lut_d0.abs_input);
d0_lut_value = "(" + std::to_string(lighting.lut_d0.scale) + " * " + GetLutValue(Regs::LightingSampler::Distribution0, index) + ")";
std::string index =
GetLutIndex(light_config.num, lighting.lut_d0.type, lighting.lut_d0.abs_input);
d0_lut_value = "(" + std::to_string(lighting.lut_d0.scale) + " * " +
GetLutValue(Regs::LightingSampler::Distribution0, index) + ")";
}
std::string specular_0 = "(" + d0_lut_value + " * " + light_src + ".specular_0)";
// If enabled, lookup ReflectRed value, otherwise, 1.0 is used
if (lighting.lut_rr.enable && Pica::Regs::IsLightingSamplerSupported(lighting.config, Pica::Regs::LightingSampler::ReflectRed)) {
std::string index = GetLutIndex(light_config.num, lighting.lut_rr.type, lighting.lut_rr.abs_input);
std::string value = "(" + std::to_string(lighting.lut_rr.scale) + " * " + GetLutValue(Regs::LightingSampler::ReflectRed, index) + ")";
if (lighting.lut_rr.enable &&
Pica::Regs::IsLightingSamplerSupported(lighting.config,
Pica::Regs::LightingSampler::ReflectRed)) {
std::string index =
GetLutIndex(light_config.num, lighting.lut_rr.type, lighting.lut_rr.abs_input);
std::string value = "(" + std::to_string(lighting.lut_rr.scale) + " * " +
GetLutValue(Regs::LightingSampler::ReflectRed, index) + ")";
out += "refl_value.r = " + value + ";\n";
} else {
out += "refl_value.r = 1.0;\n";
}
// If enabled, lookup ReflectGreen value, otherwise, ReflectRed value is used
if (lighting.lut_rg.enable && Pica::Regs::IsLightingSamplerSupported(lighting.config, Pica::Regs::LightingSampler::ReflectGreen)) {
std::string index = GetLutIndex(light_config.num, lighting.lut_rg.type, lighting.lut_rg.abs_input);
std::string value = "(" + std::to_string(lighting.lut_rg.scale) + " * " + GetLutValue(Regs::LightingSampler::ReflectGreen, index) + ")";
if (lighting.lut_rg.enable &&
Pica::Regs::IsLightingSamplerSupported(lighting.config,
Pica::Regs::LightingSampler::ReflectGreen)) {
std::string index =
GetLutIndex(light_config.num, lighting.lut_rg.type, lighting.lut_rg.abs_input);
std::string value = "(" + std::to_string(lighting.lut_rg.scale) + " * " +
GetLutValue(Regs::LightingSampler::ReflectGreen, index) + ")";
out += "refl_value.g = " + value + ";\n";
} else {
out += "refl_value.g = refl_value.r;\n";
}
// If enabled, lookup ReflectBlue value, otherwise, ReflectRed value is used
if (lighting.lut_rb.enable && Pica::Regs::IsLightingSamplerSupported(lighting.config, Pica::Regs::LightingSampler::ReflectBlue)) {
std::string index = GetLutIndex(light_config.num, lighting.lut_rb.type, lighting.lut_rb.abs_input);
std::string value = "(" + std::to_string(lighting.lut_rb.scale) + " * " + GetLutValue(Regs::LightingSampler::ReflectBlue, index) + ")";
if (lighting.lut_rb.enable &&
Pica::Regs::IsLightingSamplerSupported(lighting.config,
Pica::Regs::LightingSampler::ReflectBlue)) {
std::string index =
GetLutIndex(light_config.num, lighting.lut_rb.type, lighting.lut_rb.abs_input);
std::string value = "(" + std::to_string(lighting.lut_rb.scale) + " * " +
GetLutValue(Regs::LightingSampler::ReflectBlue, index) + ")";
out += "refl_value.b = " + value + ";\n";
} else {
out += "refl_value.b = refl_value.r;\n";
@ -487,18 +530,26 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) {
// Specular 1 component
std::string d1_lut_value = "1.0";
if (lighting.lut_d1.enable && Pica::Regs::IsLightingSamplerSupported(lighting.config, Pica::Regs::LightingSampler::Distribution1)) {
if (lighting.lut_d1.enable &&
Pica::Regs::IsLightingSamplerSupported(lighting.config,
Pica::Regs::LightingSampler::Distribution1)) {
// Lookup specular "distribution 1" LUT value
std::string index = GetLutIndex(light_config.num, lighting.lut_d1.type, lighting.lut_d1.abs_input);
d1_lut_value = "(" + std::to_string(lighting.lut_d1.scale) + " * " + GetLutValue(Regs::LightingSampler::Distribution1, index) + ")";
std::string index =
GetLutIndex(light_config.num, lighting.lut_d1.type, lighting.lut_d1.abs_input);
d1_lut_value = "(" + std::to_string(lighting.lut_d1.scale) + " * " +
GetLutValue(Regs::LightingSampler::Distribution1, index) + ")";
}
std::string specular_1 = "(" + d1_lut_value + " * refl_value * " + light_src + ".specular_1)";
std::string specular_1 =
"(" + d1_lut_value + " * refl_value * " + light_src + ".specular_1)";
// Fresnel
if (lighting.lut_fr.enable && Pica::Regs::IsLightingSamplerSupported(lighting.config, Pica::Regs::LightingSampler::Fresnel)) {
if (lighting.lut_fr.enable && Pica::Regs::IsLightingSamplerSupported(
lighting.config, Pica::Regs::LightingSampler::Fresnel)) {
// Lookup fresnel LUT value
std::string index = GetLutIndex(light_config.num, lighting.lut_fr.type, lighting.lut_fr.abs_input);
std::string value = "(" + std::to_string(lighting.lut_fr.scale) + " * " + GetLutValue(Regs::LightingSampler::Fresnel, index) + ")";
std::string index =
GetLutIndex(light_config.num, lighting.lut_fr.type, lighting.lut_fr.abs_input);
std::string value = "(" + std::to_string(lighting.lut_fr.scale) + " * " +
GetLutValue(Regs::LightingSampler::Fresnel, index) + ")";
// Enabled for difffuse lighting alpha component
if (lighting.fresnel_selector == Pica::Regs::LightingFresnelSelector::PrimaryAlpha ||
@ -512,10 +563,12 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) {
}
// Compute primary fragment color (diffuse lighting) function
out += "diffuse_sum.rgb += ((" + light_src + ".diffuse * " + dot_product + ") + " + light_src + ".ambient) * " + dist_atten + ";\n";
out += "diffuse_sum.rgb += ((" + light_src + ".diffuse * " + dot_product + ") + " +
light_src + ".ambient) * " + dist_atten + ";\n";
// Compute secondary fragment color (specular lighting) function
out += "specular_sum.rgb += (" + specular_0 + " + " + specular_1 + ") * " + clamp_highlights + " * " + dist_atten + ";\n";
out += "specular_sum.rgb += (" + specular_0 + " + " + specular_1 + ") * " +
clamp_highlights + " * " + dist_atten + ";\n";
}
// Sum final lighting result
@ -598,9 +651,9 @@ vec4 secondary_fragment_color = vec4(0.0);
out += "!";
// x2,y2 have +1 added to cover the entire pixel area
out += "(gl_FragCoord.x >= scissor_x1 * framebuffer_scale.x && "
"gl_FragCoord.y >= scissor_y1 * framebuffer_scale.y && "
"gl_FragCoord.x < (scissor_x2 + 1) * framebuffer_scale.x && "
"gl_FragCoord.y < (scissor_y2 + 1) * framebuffer_scale.y)) discard;\n";
"gl_FragCoord.y >= scissor_y1 * framebuffer_scale.y && "
"gl_FragCoord.x < (scissor_x2 + 1) * framebuffer_scale.x && "
"gl_FragCoord.y < (scissor_y2 + 1) * framebuffer_scale.y)) discard;\n";
}
out += "float z_over_w = 1.0 - gl_FragCoord.z * 2.0;\n";
@ -638,9 +691,11 @@ vec4 secondary_fragment_color = vec4(0.0);
out += "float fog_i = clamp(floor(fog_index), 0.0, 127.0);\n";
out += "float fog_f = fog_index - fog_i;\n";
out += "uint fog_lut_entry = texelFetch(fog_lut, int(fog_i), 0).r;\n";
out += "float fog_lut_entry_difference = float(int((fog_lut_entry & 0x1FFFU) << 19U) >> 19);\n"; // Extract signed difference
out += "float fog_lut_entry_difference = float(int((fog_lut_entry & 0x1FFFU) << 19U) >> "
"19);\n"; // Extract signed difference
out += "float fog_lut_entry_value = float((fog_lut_entry >> 13U) & 0x7FFU);\n";
out += "float fog_factor = (fog_lut_entry_value + fog_lut_entry_difference * fog_f) / 2047.0;\n";
out += "float fog_factor = (fog_lut_entry_value + fog_lut_entry_difference * fog_f) / "
"2047.0;\n";
out += "fog_factor = clamp(fog_factor, 0.0, 1.0);\n";
// Blend the fog
@ -658,14 +713,20 @@ vec4 secondary_fragment_color = vec4(0.0);
std::string GenerateVertexShader() {
std::string out = "#version 330 core\n";
out += "layout(location = " + std::to_string((int)ATTRIBUTE_POSITION) + ") in vec4 vert_position;\n";
out += "layout(location = " + std::to_string((int)ATTRIBUTE_COLOR) + ") in vec4 vert_color;\n";
out += "layout(location = " + std::to_string((int)ATTRIBUTE_TEXCOORD0) + ") in vec2 vert_texcoord0;\n";
out += "layout(location = " + std::to_string((int)ATTRIBUTE_TEXCOORD1) + ") in vec2 vert_texcoord1;\n";
out += "layout(location = " + std::to_string((int)ATTRIBUTE_TEXCOORD2) + ") in vec2 vert_texcoord2;\n";
out += "layout(location = " + std::to_string((int)ATTRIBUTE_TEXCOORD0_W) + ") in float vert_texcoord0_w;\n";
out += "layout(location = " + std::to_string((int)ATTRIBUTE_NORMQUAT) + ") in vec4 vert_normquat;\n";
out += "layout(location = " + std::to_string((int)ATTRIBUTE_VIEW) + ") in vec3 vert_view;\n";
out += "layout(location = " + std::to_string((int)ATTRIBUTE_POSITION) +
") in vec4 vert_position;\n";
out += "layout(location = " + std::to_string((int)ATTRIBUTE_COLOR) + ") in vec4 vert_color;\n";
out += "layout(location = " + std::to_string((int)ATTRIBUTE_TEXCOORD0) +
") in vec2 vert_texcoord0;\n";
out += "layout(location = " + std::to_string((int)ATTRIBUTE_TEXCOORD1) +
") in vec2 vert_texcoord1;\n";
out += "layout(location = " + std::to_string((int)ATTRIBUTE_TEXCOORD2) +
") in vec2 vert_texcoord2;\n";
out += "layout(location = " + std::to_string((int)ATTRIBUTE_TEXCOORD0_W) +
") in float vert_texcoord0_w;\n";
out += "layout(location = " + std::to_string((int)ATTRIBUTE_NORMQUAT) +
") in vec4 vert_normquat;\n";
out += "layout(location = " + std::to_string((int)ATTRIBUTE_VIEW) + ") in vec3 vert_view;\n";
out += R"(
out vec4 primary_color;

3
src/video_core/renderer_opengl/gl_shader_util.cpp
View file

@ -56,7 +56,8 @@ GLuint LoadProgram(const char* vertex_shader, const char* fragment_shader) {
if (result) {
LOG_DEBUG(Render_OpenGL, "%s", &fragment_shader_error[0]);
} else {
LOG_ERROR(Render_OpenGL, "Error compiling fragment shader:\n%s", &fragment_shader_error[0]);
LOG_ERROR(Render_OpenGL, "Error compiling fragment shader:\n%s",
&fragment_shader_error[0]);
}
}

49
src/video_core/renderer_opengl/gl_state.cpp
View file

@ -106,11 +106,11 @@ void OpenGLState::Apply() const {
// Color mask
if (color_mask.red_enabled != cur_state.color_mask.red_enabled ||
color_mask.green_enabled != cur_state.color_mask.green_enabled ||
color_mask.blue_enabled != cur_state.color_mask.blue_enabled ||
color_mask.alpha_enabled != cur_state.color_mask.alpha_enabled) {
glColorMask(color_mask.red_enabled, color_mask.green_enabled,
color_mask.blue_enabled, color_mask.alpha_enabled);
color_mask.green_enabled != cur_state.color_mask.green_enabled ||
color_mask.blue_enabled != cur_state.color_mask.blue_enabled ||
color_mask.alpha_enabled != cur_state.color_mask.alpha_enabled) {
glColorMask(color_mask.red_enabled, color_mask.green_enabled, color_mask.blue_enabled,
color_mask.alpha_enabled);
}
// Stencil test
@ -123,15 +123,16 @@ void OpenGLState::Apply() const {
}
if (stencil.test_func != cur_state.stencil.test_func ||
stencil.test_ref != cur_state.stencil.test_ref ||
stencil.test_mask != cur_state.stencil.test_mask) {
stencil.test_ref != cur_state.stencil.test_ref ||
stencil.test_mask != cur_state.stencil.test_mask) {
glStencilFunc(stencil.test_func, stencil.test_ref, stencil.test_mask);
}
if (stencil.action_depth_fail != cur_state.stencil.action_depth_fail ||
stencil.action_depth_pass != cur_state.stencil.action_depth_pass ||
stencil.action_stencil_fail != cur_state.stencil.action_stencil_fail) {
glStencilOp(stencil.action_stencil_fail, stencil.action_depth_fail, stencil.action_depth_pass);
stencil.action_depth_pass != cur_state.stencil.action_depth_pass ||
stencil.action_stencil_fail != cur_state.stencil.action_stencil_fail) {
glStencilOp(stencil.action_stencil_fail, stencil.action_depth_fail,
stencil.action_depth_pass);
}
// Stencil mask
@ -154,23 +155,22 @@ void OpenGLState::Apply() const {
}
if (blend.color.red != cur_state.blend.color.red ||
blend.color.green != cur_state.blend.color.green ||
blend.color.blue != cur_state.blend.color.blue ||
blend.color.alpha != cur_state.blend.color.alpha) {
glBlendColor(blend.color.red, blend.color.green,
blend.color.blue, blend.color.alpha);
blend.color.green != cur_state.blend.color.green ||
blend.color.blue != cur_state.blend.color.blue ||
blend.color.alpha != cur_state.blend.color.alpha) {
glBlendColor(blend.color.red, blend.color.green, blend.color.blue, blend.color.alpha);
}
if (blend.src_rgb_func != cur_state.blend.src_rgb_func ||
blend.dst_rgb_func != cur_state.blend.dst_rgb_func ||
blend.src_a_func != cur_state.blend.src_a_func ||
blend.dst_a_func != cur_state.blend.dst_a_func) {
glBlendFuncSeparate(blend.src_rgb_func, blend.dst_rgb_func,
blend.src_a_func, blend.dst_a_func);
blend.dst_rgb_func != cur_state.blend.dst_rgb_func ||
blend.src_a_func != cur_state.blend.src_a_func ||
blend.dst_a_func != cur_state.blend.dst_a_func) {
glBlendFuncSeparate(blend.src_rgb_func, blend.dst_rgb_func, blend.src_a_func,
blend.dst_a_func);
}
if (blend.rgb_equation != cur_state.blend.rgb_equation ||
blend.a_equation != cur_state.blend.a_equation) {
blend.a_equation != cur_state.blend.a_equation) {
glBlendEquationSeparate(blend.rgb_equation, blend.a_equation);
}
@ -237,8 +237,11 @@ void OpenGLState::Apply() const {
GLenum OpenGLState::CheckFBStatus(GLenum target) {
GLenum fb_status = glCheckFramebufferStatus(target);
if (fb_status != GL_FRAMEBUFFER_COMPLETE) {
const char* fb_description = (target == GL_READ_FRAMEBUFFER ? "READ" : (target == GL_DRAW_FRAMEBUFFER ? "DRAW" : "UNK"));
LOG_CRITICAL(Render_OpenGL, "OpenGL %s framebuffer check failed, status %X", fb_description, fb_status);
const char* fb_description =
(target == GL_READ_FRAMEBUFFER ? "READ"
: (target == GL_DRAW_FRAMEBUFFER ? "DRAW" : "UNK"));
LOG_CRITICAL(Render_OpenGL, "OpenGL %s framebuffer check failed, status %X", fb_description,
fb_status);
}
return fb_status;

40
src/video_core/renderer_opengl/gl_state.h
View file

@ -9,14 +9,14 @@
class OpenGLState {
public:
struct {
bool enabled; // GL_CULL_FACE
GLenum mode; // GL_CULL_FACE_MODE
bool enabled; // GL_CULL_FACE
GLenum mode; // GL_CULL_FACE_MODE
GLenum front_face; // GL_FRONT_FACE
} cull;
struct {
bool test_enabled; // GL_DEPTH_TEST
GLenum test_func; // GL_DEPTH_FUNC
bool test_enabled; // GL_DEPTH_TEST
GLenum test_func; // GL_DEPTH_FUNC
GLboolean write_mask; // GL_DEPTH_WRITEMASK
} depth;
@ -28,24 +28,24 @@ public:
} color_mask; // GL_COLOR_WRITEMASK
struct {
bool test_enabled; // GL_STENCIL_TEST
GLenum test_func; // GL_STENCIL_FUNC
GLint test_ref; // GL_STENCIL_REF
GLuint test_mask; // GL_STENCIL_VALUE_MASK
GLuint write_mask; // GL_STENCIL_WRITEMASK
bool test_enabled; // GL_STENCIL_TEST
GLenum test_func; // GL_STENCIL_FUNC
GLint test_ref; // GL_STENCIL_REF
GLuint test_mask; // GL_STENCIL_VALUE_MASK
GLuint write_mask; // GL_STENCIL_WRITEMASK
GLenum action_stencil_fail; // GL_STENCIL_FAIL
GLenum action_depth_fail; // GL_STENCIL_PASS_DEPTH_FAIL
GLenum action_depth_pass; // GL_STENCIL_PASS_DEPTH_PASS
GLenum action_depth_fail; // GL_STENCIL_PASS_DEPTH_FAIL
GLenum action_depth_pass; // GL_STENCIL_PASS_DEPTH_PASS
} stencil;
struct {
bool enabled; // GL_BLEND
bool enabled; // GL_BLEND
GLenum rgb_equation; // GL_BLEND_EQUATION_RGB
GLenum a_equation; // GL_BLEND_EQUATION_ALPHA
GLenum a_equation; // GL_BLEND_EQUATION_ALPHA
GLenum src_rgb_func; // GL_BLEND_SRC_RGB
GLenum dst_rgb_func; // GL_BLEND_DST_RGB
GLenum src_a_func; // GL_BLEND_SRC_ALPHA
GLenum dst_a_func; // GL_BLEND_DST_ALPHA
GLenum src_a_func; // GL_BLEND_SRC_ALPHA
GLenum dst_a_func; // GL_BLEND_DST_ALPHA
struct {
GLclampf red;
@ -60,7 +60,7 @@ public:
// 3 texture units - one for each that is used in PICA fragment shader emulation
struct {
GLuint texture_2d; // GL_TEXTURE_BINDING_2D
GLuint sampler; // GL_SAMPLER_BINDING
GLuint sampler; // GL_SAMPLER_BINDING
} texture_units[3];
struct {
@ -74,10 +74,10 @@ public:
struct {
GLuint read_framebuffer; // GL_READ_FRAMEBUFFER_BINDING
GLuint draw_framebuffer; // GL_DRAW_FRAMEBUFFER_BINDING
GLuint vertex_array; // GL_VERTEX_ARRAY_BINDING
GLuint vertex_buffer; // GL_ARRAY_BUFFER_BINDING
GLuint uniform_buffer; // GL_UNIFORM_BUFFER_BINDING
GLuint shader_program; // GL_CURRENT_PROGRAM
GLuint vertex_array; // GL_VERTEX_ARRAY_BINDING
GLuint vertex_buffer; // GL_ARRAY_BUFFER_BINDING
GLuint uniform_buffer; // GL_UNIFORM_BUFFER_BINDING
GLuint shader_program; // GL_CURRENT_PROGRAM
} draw;
OpenGLState();

72
src/video_core/renderer_opengl/pica_to_gl.h
View file

@ -25,8 +25,8 @@ namespace PicaToGL {
inline GLenum TextureFilterMode(Pica::Regs::TextureConfig::TextureFilter mode) {
static const GLenum filter_mode_table[] = {
GL_NEAREST, // TextureFilter::Nearest
GL_LINEAR // TextureFilter::Linear
GL_NEAREST, // TextureFilter::Nearest
GL_LINEAR // TextureFilter::Linear
};
// Range check table for input
@ -52,10 +52,10 @@ inline GLenum TextureFilterMode(Pica::Regs::TextureConfig::TextureFilter mode) {
inline GLenum WrapMode(Pica::Regs::TextureConfig::WrapMode mode) {
static const GLenum wrap_mode_table[] = {
GL_CLAMP_TO_EDGE, // WrapMode::ClampToEdge
GL_CLAMP_TO_BORDER,// WrapMode::ClampToBorder
GL_REPEAT, // WrapMode::Repeat
GL_MIRRORED_REPEAT // WrapMode::MirroredRepeat
GL_CLAMP_TO_EDGE, // WrapMode::ClampToEdge
GL_CLAMP_TO_BORDER, // WrapMode::ClampToBorder
GL_REPEAT, // WrapMode::Repeat
GL_MIRRORED_REPEAT // WrapMode::MirroredRepeat
};
// Range check table for input
@ -131,22 +131,22 @@ inline GLenum BlendFunc(Pica::Regs::BlendFactor factor) {
inline GLenum LogicOp(Pica::Regs::LogicOp op) {
static const GLenum logic_op_table[] = {
GL_CLEAR, // Clear
GL_AND, // And
GL_AND_REVERSE, // AndReverse
GL_COPY, // Copy
GL_SET, // Set
GL_COPY_INVERTED, // CopyInverted
GL_NOOP, // NoOp
GL_INVERT, // Invert
GL_NAND, // Nand
GL_OR, // Or
GL_NOR, // Nor
GL_XOR, // Xor
GL_EQUIV, // Equiv
GL_AND_INVERTED, // AndInverted
GL_OR_REVERSE, // OrReverse
GL_OR_INVERTED, // OrInverted
GL_CLEAR, // Clear
GL_AND, // And
GL_AND_REVERSE, // AndReverse
GL_COPY, // Copy
GL_SET, // Set
GL_COPY_INVERTED, // CopyInverted
GL_NOOP, // NoOp
GL_INVERT, // Invert
GL_NAND, // Nand
GL_OR, // Or
GL_NOR, // Nor
GL_XOR, // Xor
GL_EQUIV, // Equiv
GL_AND_INVERTED, // AndInverted
GL_OR_REVERSE, // OrReverse
GL_OR_INVERTED, // OrInverted
};
// Range check table for input
@ -185,14 +185,14 @@ inline GLenum CompareFunc(Pica::Regs::CompareFunc func) {
inline GLenum StencilOp(Pica::Regs::StencilAction action) {
static const GLenum stencil_op_table[] = {
GL_KEEP, // StencilAction::Keep
GL_ZERO, // StencilAction::Zero
GL_REPLACE, // StencilAction::Replace
GL_INCR, // StencilAction::Increment
GL_DECR, // StencilAction::Decrement
GL_INVERT, // StencilAction::Invert
GL_INCR_WRAP, // StencilAction::IncrementWrap
GL_DECR_WRAP // StencilAction::DecrementWrap
GL_KEEP, // StencilAction::Keep
GL_ZERO, // StencilAction::Zero
GL_REPLACE, // StencilAction::Replace
GL_INCR, // StencilAction::Increment
GL_DECR, // StencilAction::Decrement
GL_INVERT, // StencilAction::Invert
GL_INCR_WRAP, // StencilAction::IncrementWrap
GL_DECR_WRAP // StencilAction::DecrementWrap
};
// Range check table for input
@ -207,18 +207,12 @@ inline GLenum StencilOp(Pica::Regs::StencilAction action) {
}
inline GLvec4 ColorRGBA8(const u32 color) {
return { { (color >> 0 & 0xFF) / 255.0f,
(color >> 8 & 0xFF) / 255.0f,
(color >> 16 & 0xFF) / 255.0f,
(color >> 24 & 0xFF) / 255.0f
} };
return {{(color >> 0 & 0xFF) / 255.0f, (color >> 8 & 0xFF) / 255.0f,
(color >> 16 & 0xFF) / 255.0f, (color >> 24 & 0xFF) / 255.0f}};
}
inline std::array<GLfloat, 3> LightColor(const Pica::Regs::LightColor& color) {
return { { color.r / 255.0f,
color.g / 255.0f,
color.b / 255.0f
} };
return {{color.r / 255.0f, color.g / 255.0f, color.b / 255.0f}};
}
} // namespace

122
src/video_core/renderer_opengl/renderer_opengl.cpp
View file

@ -89,8 +89,12 @@ struct ScreenRectVertex {
static std::array<GLfloat, 3 * 2> MakeOrthographicMatrix(const float width, const float height) {
std::array<GLfloat, 3 * 2> matrix;
matrix[0] = 2.f / width; matrix[2] = 0.f; matrix[4] = -1.f;
matrix[1] = 0.f; matrix[3] = -2.f / height; matrix[5] = 1.f;
matrix[0] = 2.f / width;
matrix[2] = 0.f;
matrix[4] = -1.f;
matrix[1] = 0.f;
matrix[3] = -2.f / height;
matrix[5] = 1.f;
// Last matrix row is implicitly assumed to be [0, 0, 1].
return matrix;
@ -98,7 +102,7 @@ static std::array<GLfloat, 3 * 2> MakeOrthographicMatrix(const float width, cons
/// RendererOpenGL constructor
RendererOpenGL::RendererOpenGL() {
resolution_width = std::max(VideoCore::kScreenTopWidth, VideoCore::kScreenBottomWidth);
resolution_width = std::max(VideoCore::kScreenTopWidth, VideoCore::kScreenBottomWidth);
resolution_height = VideoCore::kScreenTopHeight + VideoCore::kScreenBottomHeight;
}
@ -116,13 +120,15 @@ void RendererOpenGL::SwapBuffers() {
const auto& framebuffer = GPU::g_regs.framebuffer_config[i];
// Main LCD (0): 0x1ED02204, Sub LCD (1): 0x1ED02A04
u32 lcd_color_addr = (i == 0) ? LCD_REG_INDEX(color_fill_top) : LCD_REG_INDEX(color_fill_bottom);
u32 lcd_color_addr =
(i == 0) ? LCD_REG_INDEX(color_fill_top) : LCD_REG_INDEX(color_fill_bottom);
lcd_color_addr = HW::VADDR_LCD + 4 * lcd_color_addr;
LCD::Regs::ColorFill color_fill = {0};
LCD::Read(color_fill.raw, lcd_color_addr);
if (color_fill.is_enabled) {
LoadColorToActiveGLTexture(color_fill.color_r, color_fill.color_g, color_fill.color_b, screen_infos[i].texture);
LoadColorToActiveGLTexture(color_fill.color_r, color_fill.color_g, color_fill.color_b,
screen_infos[i].texture);
// Resize the texture in case the framebuffer size has changed
screen_infos[i].texture.width = 1;
@ -172,15 +178,14 @@ void RendererOpenGL::SwapBuffers() {
* Loads framebuffer from emulated memory into the active OpenGL texture.
*/
void RendererOpenGL::LoadFBToScreenInfo(const GPU::Regs::FramebufferConfig& framebuffer,
ScreenInfo& screen_info) {
ScreenInfo& screen_info) {
const PAddr framebuffer_addr = framebuffer.active_fb == 0 ?
framebuffer.address_left1 : framebuffer.address_left2;
const PAddr framebuffer_addr =
framebuffer.active_fb == 0 ? framebuffer.address_left1 : framebuffer.address_left2;
LOG_TRACE(Render_OpenGL, "0x%08x bytes from 0x%08x(%dx%d), fmt %x",
framebuffer.stride * framebuffer.height,
framebuffer_addr, (int)framebuffer.width,
(int)framebuffer.height, (int)framebuffer.format);
framebuffer.stride * framebuffer.height, framebuffer_addr, (int)framebuffer.width,
(int)framebuffer.height, (int)framebuffer.format);
int bpp = GPU::Regs::BytesPerPixel(framebuffer.color_format);
size_t pixel_stride = framebuffer.stride / bpp;
@ -192,7 +197,8 @@ void RendererOpenGL::LoadFBToScreenInfo(const GPU::Regs::FramebufferConfig& fram
// only allows rows to have a memory alignement of 4.
ASSERT(pixel_stride % 4 == 0);
if (!Rasterizer()->AccelerateDisplay(framebuffer, framebuffer_addr, static_cast<u32>(pixel_stride), screen_info)) {
if (!Rasterizer()->AccelerateDisplay(framebuffer, framebuffer_addr,
static_cast<u32>(pixel_stride), screen_info)) {
// Reset the screen info's display texture to its own permanent texture
screen_info.display_texture = screen_info.texture.resource.handle;
screen_info.display_texcoords = MathUtil::Rectangle<float>(0.f, 0.f, 1.f, 1.f);
@ -208,12 +214,14 @@ void RendererOpenGL::LoadFBToScreenInfo(const GPU::Regs::FramebufferConfig& fram
glPixelStorei(GL_UNPACK_ROW_LENGTH, (GLint)pixel_stride);
// Update existing texture
// TODO: Test what happens on hardware when you change the framebuffer dimensions so that they
// TODO: Test what happens on hardware when you change the framebuffer dimensions so that
// they
// differ from the LCD resolution.
// TODO: Applications could theoretically crash Citra here by specifying too large
// framebuffer sizes. We should make sure that this cannot happen.
glTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, framebuffer.width, framebuffer.height,
screen_info.texture.gl_format, screen_info.texture.gl_type, framebuffer_data);
screen_info.texture.gl_format, screen_info.texture.gl_type,
framebuffer_data);
glPixelStorei(GL_UNPACK_ROW_LENGTH, 0);
@ -224,7 +232,8 @@ void RendererOpenGL::LoadFBToScreenInfo(const GPU::Regs::FramebufferConfig& fram
/**
* Fills active OpenGL texture with the given RGB color.
* Since the color is solid, the texture can be 1x1 but will stretch across whatever it's rendered on.
* Since the color is solid, the texture can be 1x1 but will stretch across whatever it's rendered
* on.
* This has the added benefit of being *really fast*.
*/
void RendererOpenGL::LoadColorToActiveGLTexture(u8 color_r, u8 color_g, u8 color_b,
@ -233,7 +242,7 @@ void RendererOpenGL::LoadColorToActiveGLTexture(u8 color_r, u8 color_g, u8 color
state.Apply();
glActiveTexture(GL_TEXTURE0);
u8 framebuffer_data[3] = { color_r, color_g, color_b };
u8 framebuffer_data[3] = {color_r, color_g, color_b};
// Update existing texture
glTexImage2D(GL_TEXTURE_2D, 0, GL_RGB, 1, 1, 0, GL_RGB, GL_UNSIGNED_BYTE, framebuffer_data);
@ -246,7 +255,8 @@ void RendererOpenGL::LoadColorToActiveGLTexture(u8 color_r, u8 color_g, u8 color
* Initializes the OpenGL state and creates persistent objects.
*/
void RendererOpenGL::InitOpenGLObjects() {
glClearColor(Settings::values.bg_red, Settings::values.bg_green, Settings::values.bg_blue, 0.0f);
glClearColor(Settings::values.bg_red, Settings::values.bg_green, Settings::values.bg_blue,
0.0f);
// Link shaders and get variable locations
shader.Create(vertex_shader, fragment_shader);
@ -270,8 +280,10 @@ void RendererOpenGL::InitOpenGLObjects() {
// Attach vertex data to VAO
glBufferData(GL_ARRAY_BUFFER, sizeof(ScreenRectVertex) * 4, nullptr, GL_STREAM_DRAW);
glVertexAttribPointer(attrib_position, 2, GL_FLOAT, GL_FALSE, sizeof(ScreenRectVertex), (GLvoid*)offsetof(ScreenRectVertex, position));
glVertexAttribPointer(attrib_tex_coord, 2, GL_FLOAT, GL_FALSE, sizeof(ScreenRectVertex), (GLvoid*)offsetof(ScreenRectVertex, tex_coord));
glVertexAttribPointer(attrib_position, 2, GL_FLOAT, GL_FALSE, sizeof(ScreenRectVertex),
(GLvoid*)offsetof(ScreenRectVertex, position));
glVertexAttribPointer(attrib_tex_coord, 2, GL_FLOAT, GL_FALSE, sizeof(ScreenRectVertex),
(GLvoid*)offsetof(ScreenRectVertex, tex_coord));
glEnableVertexAttribArray(attrib_position);
glEnableVertexAttribArray(attrib_tex_coord);
@ -352,23 +364,25 @@ void RendererOpenGL::ConfigureFramebufferTexture(TextureInfo& texture,
glActiveTexture(GL_TEXTURE0);
glTexImage2D(GL_TEXTURE_2D, 0, internal_format, texture.width, texture.height, 0,
texture.gl_format, texture.gl_type, nullptr);
texture.gl_format, texture.gl_type, nullptr);
state.texture_units[0].texture_2d = 0;
state.Apply();
}
/**
* Draws a single texture to the emulator window, rotating the texture to correct for the 3DS's LCD rotation.
* Draws a single texture to the emulator window, rotating the texture to correct for the 3DS's LCD
* rotation.
*/
void RendererOpenGL::DrawSingleScreenRotated(const ScreenInfo& screen_info, float x, float y, float w, float h) {
void RendererOpenGL::DrawSingleScreenRotated(const ScreenInfo& screen_info, float x, float y,
float w, float h) {
auto& texcoords = screen_info.display_texcoords;
std::array<ScreenRectVertex, 4> vertices = {{
ScreenRectVertex(x, y, texcoords.bottom, texcoords.left),
ScreenRectVertex(x+w, y, texcoords.bottom, texcoords.right),
ScreenRectVertex(x, y+h, texcoords.top, texcoords.left),
ScreenRectVertex(x+w, y+h, texcoords.top, texcoords.right),
ScreenRectVertex(x, y, texcoords.bottom, texcoords.left),
ScreenRectVertex(x + w, y, texcoords.bottom, texcoords.right),
ScreenRectVertex(x, y + h, texcoords.top, texcoords.left),
ScreenRectVertex(x + w, y + h, texcoords.top, texcoords.right),
}};
state.texture_units[0].texture_2d = screen_info.display_texture;
@ -391,18 +405,20 @@ void RendererOpenGL::DrawScreens() {
glClear(GL_COLOR_BUFFER_BIT);
// Set projection matrix
std::array<GLfloat, 3 * 2> ortho_matrix = MakeOrthographicMatrix((float)layout.width,
(float)layout.height);
std::array<GLfloat, 3 * 2> ortho_matrix =
MakeOrthographicMatrix((float)layout.width, (float)layout.height);
glUniformMatrix3x2fv(uniform_modelview_matrix, 1, GL_FALSE, ortho_matrix.data());
// Bind texture in Texture Unit 0
glActiveTexture(GL_TEXTURE0);
glUniform1i(uniform_color_texture, 0);
DrawSingleScreenRotated(screen_infos[0], (float)layout.top_screen.left, (float)layout.top_screen.top,
(float)layout.top_screen.GetWidth(), (float)layout.top_screen.GetHeight());
DrawSingleScreenRotated(screen_infos[1], (float)layout.bottom_screen.left,(float)layout.bottom_screen.top,
(float)layout.bottom_screen.GetWidth(), (float)layout.bottom_screen.GetHeight());
DrawSingleScreenRotated(screen_infos[0], (float)layout.top_screen.left,
(float)layout.top_screen.top, (float)layout.top_screen.GetWidth(),
(float)layout.top_screen.GetHeight());
DrawSingleScreenRotated(screen_infos[1], (float)layout.bottom_screen.left,
(float)layout.bottom_screen.top, (float)layout.bottom_screen.GetWidth(),
(float)layout.bottom_screen.GetHeight());
m_current_frame++;
}
@ -420,14 +436,16 @@ void RendererOpenGL::SetWindow(EmuWindow* window) {
}
static const char* GetSource(GLenum source) {
#define RET(s) case GL_DEBUG_SOURCE_##s: return #s
#define RET(s) \
case GL_DEBUG_SOURCE_##s: \
return #s
switch (source) {
RET(API);
RET(WINDOW_SYSTEM);
RET(SHADER_COMPILER);
RET(THIRD_PARTY);
RET(APPLICATION);
RET(OTHER);
RET(API);
RET(WINDOW_SYSTEM);
RET(SHADER_COMPILER);
RET(THIRD_PARTY);
RET(APPLICATION);
RET(OTHER);
default:
UNREACHABLE();
}
@ -435,23 +453,25 @@ static const char* GetSource(GLenum source) {
}
static const char* GetType(GLenum type) {
#define RET(t) case GL_DEBUG_TYPE_##t: return #t
#define RET(t) \
case GL_DEBUG_TYPE_##t: \
return #t
switch (type) {
RET(ERROR);
RET(DEPRECATED_BEHAVIOR);
RET(UNDEFINED_BEHAVIOR);
RET(PORTABILITY);
RET(PERFORMANCE);
RET(OTHER);
RET(MARKER);
RET(ERROR);
RET(DEPRECATED_BEHAVIOR);
RET(UNDEFINED_BEHAVIOR);
RET(PORTABILITY);
RET(PERFORMANCE);
RET(OTHER);
RET(MARKER);
default:
UNREACHABLE();
}
#undef RET
}
static void APIENTRY DebugHandler(GLenum source, GLenum type, GLuint id, GLenum severity, GLsizei length,
const GLchar* message, const void* user_param) {
static void APIENTRY DebugHandler(GLenum source, GLenum type, GLuint id, GLenum severity,
GLsizei length, const GLchar* message, const void* user_param) {
Log::Level level;
switch (severity) {
case GL_DEBUG_SEVERITY_HIGH:
@ -465,8 +485,8 @@ static void APIENTRY DebugHandler(GLenum source, GLenum type, GLuint id, GLenum
level = Log::Level::Debug;
break;
}
LOG_GENERIC(Log::Class::Render_OpenGL, level, "%s %s %d: %s",
GetSource(source), GetType(type), id, message);
LOG_GENERIC(Log::Class::Render_OpenGL, level, "%s %s %d: %s", GetSource(source), GetType(type),
id, message);
}
/// Initialize the renderer

15
src/video_core/renderer_opengl/renderer_opengl.h
View file

@ -38,7 +38,6 @@ struct ScreenInfo {
class RendererOpenGL : public RendererBase {
public:
RendererOpenGL();
~RendererOpenGL() override;
@ -67,15 +66,14 @@ private:
// Loads framebuffer from emulated memory into the display information structure
void LoadFBToScreenInfo(const GPU::Regs::FramebufferConfig& framebuffer,
ScreenInfo& screen_info);
ScreenInfo& screen_info);
// Fills active OpenGL texture with the given RGB color.
void LoadColorToActiveGLTexture(u8 color_r, u8 color_g, u8 color_b,
const TextureInfo& texture);
void LoadColorToActiveGLTexture(u8 color_r, u8 color_g, u8 color_b, const TextureInfo& texture);
EmuWindow* render_window; ///< Handle to render window
EmuWindow* render_window; ///< Handle to render window
int resolution_width; ///< Current resolution width
int resolution_height; ///< Current resolution height
int resolution_width; ///< Current resolution width
int resolution_height; ///< Current resolution height
OpenGLState state;
@ -83,7 +81,8 @@ private:
OGLVertexArray vertex_array;
OGLBuffer vertex_buffer;
OGLShader shader;
std::array<ScreenInfo, 2> screen_infos; ///< Display information for top and bottom screens respectively
std::array<ScreenInfo, 2>
screen_infos; ///< Display information for top and bottom screens respectively
// Shader uniform location indices
GLuint uniform_modelview_matrix;
GLuint uniform_color_texture;