piwalker/dolphin
1
0
Fork 0
mirror of https://github.com/dolphin-emu/dolphin.git synced 2025-07-24 14:49:42 -06:00
Code Issues Packages Projects Releases Wiki Activity

Reformat all the things. Have fun with merge conflicts.

Browse Source
This commit is contained in:
Pierre Bourdon
2016-06-24 10:43:46 +02:00
parent 2115e8a4a6
commit 3570c7f03a
1116 changed files with 187405 additions and 180344 deletions
Download Patch File Download Diff File Expand all files Collapse all files

71
Source/Core/VideoBackends/OGL/BoundingBox.cpp
View File

@ -15,57 +15,56 @@ static GLuint s_bbox_buffer_id;
namespace OGL
{
void BoundingBox::Init()
{
if (g_ActiveConfig.backend_info.bSupportsBBox)
{
int initial_values[4] = {0,0,0,0};
glGenBuffers(1, &s_bbox_buffer_id);
glBindBuffer(GL_SHADER_STORAGE_BUFFER, s_bbox_buffer_id);
glBufferData(GL_SHADER_STORAGE_BUFFER, 4 * sizeof(s32), initial_values, GL_DYNAMIC_DRAW);
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 3, s_bbox_buffer_id);
}
if (g_ActiveConfig.backend_info.bSupportsBBox)
{
int initial_values[4] = {0, 0, 0, 0};
glGenBuffers(1, &s_bbox_buffer_id);
glBindBuffer(GL_SHADER_STORAGE_BUFFER, s_bbox_buffer_id);
glBufferData(GL_SHADER_STORAGE_BUFFER, 4 * sizeof(s32), initial_values, GL_DYNAMIC_DRAW);
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 3, s_bbox_buffer_id);
}
}
void BoundingBox::Shutdown()
{
if (g_ActiveConfig.backend_info.bSupportsBBox)
glDeleteBuffers(1, &s_bbox_buffer_id);
if (g_ActiveConfig.backend_info.bSupportsBBox)
glDeleteBuffers(1, &s_bbox_buffer_id);
}
void BoundingBox::Set(int index, int value)
{
glBindBuffer(GL_SHADER_STORAGE_BUFFER, s_bbox_buffer_id);
glBufferSubData(GL_SHADER_STORAGE_BUFFER, index * sizeof(int), sizeof(int), &value);
glBindBuffer(GL_SHADER_STORAGE_BUFFER, 0);
glBindBuffer(GL_SHADER_STORAGE_BUFFER, s_bbox_buffer_id);
glBufferSubData(GL_SHADER_STORAGE_BUFFER, index * sizeof(int), sizeof(int), &value);
glBindBuffer(GL_SHADER_STORAGE_BUFFER, 0);
}
int BoundingBox::Get(int index)
{
int data = 0;
glBindBuffer(GL_SHADER_STORAGE_BUFFER, s_bbox_buffer_id);
int data = 0;
glBindBuffer(GL_SHADER_STORAGE_BUFFER, s_bbox_buffer_id);
if (!DriverDetails::HasBug(DriverDetails::BUG_SLOWGETBUFFERSUBDATA))
{
// Using glMapBufferRange to read back the contents of the SSBO is extremely slow
// on nVidia drivers. This is more noticeable at higher internal resolutions.
// Using glGetBufferSubData instead does not seem to exhibit this slowdown.
glGetBufferSubData(GL_SHADER_STORAGE_BUFFER, index * sizeof(int), sizeof(int), &data);
}
else
{
// Using glMapBufferRange is faster on AMD cards by a measurable margin.
void* ptr = glMapBufferRange(GL_SHADER_STORAGE_BUFFER, index * sizeof(int), sizeof(int), GL_MAP_READ_BIT);
if (ptr)
{
memcpy(&data, ptr, sizeof(int));
glUnmapBuffer(GL_SHADER_STORAGE_BUFFER);
}
}
if (!DriverDetails::HasBug(DriverDetails::BUG_SLOWGETBUFFERSUBDATA))
{
// Using glMapBufferRange to read back the contents of the SSBO is extremely slow
// on nVidia drivers. This is more noticeable at higher internal resolutions.
// Using glGetBufferSubData instead does not seem to exhibit this slowdown.
glGetBufferSubData(GL_SHADER_STORAGE_BUFFER, index * sizeof(int), sizeof(int), &data);
}
else
{
// Using glMapBufferRange is faster on AMD cards by a measurable margin.
void* ptr = glMapBufferRange(GL_SHADER_STORAGE_BUFFER, index * sizeof(int), sizeof(int),
GL_MAP_READ_BIT);
if (ptr)
{
memcpy(&data, ptr, sizeof(int));
glUnmapBuffer(GL_SHADER_STORAGE_BUFFER);
}
}
glBindBuffer(GL_SHADER_STORAGE_BUFFER, 0);
return data;
glBindBuffer(GL_SHADER_STORAGE_BUFFER, 0);
return data;
}
};

10
Source/Core/VideoBackends/OGL/BoundingBox.h
View File

@ -6,15 +6,13 @@
namespace OGL
{
class BoundingBox
{
public:
static void Init();
static void Shutdown();
static void Init();
static void Shutdown();
static void Set(int index, int value);
static int Get(int index);
static void Set(int index, int value);
static int Get(int index);
};
};

1008
Source/Core/VideoBackends/OGL/FramebufferManager.cpp
View File

File diff suppressed because it is too large Load Diff

118
Source/Core/VideoBackends/OGL/FramebufferManager.h
View File

@ -46,84 +46,92 @@
namespace OGL
{
struct XFBSource : public XFBSourceBase
{
XFBSource(GLuint tex, int layers) : texture(tex), m_layers(layers) {}
~XFBSource();
XFBSource(GLuint tex, int layers) : texture(tex), m_layers(layers) {}
~XFBSource();
void CopyEFB(float Gamma) override;
void DecodeToTexture(u32 xfbAddr, u32 fbWidth, u32 fbHeight) override;
void CopyEFB(float Gamma) override;
void DecodeToTexture(u32 xfbAddr, u32 fbWidth, u32 fbHeight) override;
const GLuint texture;
const int m_layers;
const GLuint texture;
const int m_layers;
};
class FramebufferManager : public FramebufferManagerBase
{
public:
FramebufferManager(int targetWidth, int targetHeight, int msaaSamples);
~FramebufferManager();
FramebufferManager(int targetWidth, int targetHeight, int msaaSamples);
~FramebufferManager();
// To get the EFB in texture form, these functions may have to transfer
// the EFB to a resolved texture first.
static GLuint GetEFBColorTexture(const EFBRectangle& sourceRc);
static GLuint GetEFBDepthTexture(const EFBRectangle& sourceRc);
// To get the EFB in texture form, these functions may have to transfer
// the EFB to a resolved texture first.
static GLuint GetEFBColorTexture(const EFBRectangle& sourceRc);
static GLuint GetEFBDepthTexture(const EFBRectangle& sourceRc);
static GLuint GetEFBFramebuffer(unsigned int layer = 0) { return (layer < m_EFBLayers) ? m_efbFramebuffer[layer] : m_efbFramebuffer.back(); }
static GLuint GetXFBFramebuffer() { return m_xfbFramebuffer; }
static GLuint GetEFBFramebuffer(unsigned int layer = 0)
{
return (layer < m_EFBLayers) ? m_efbFramebuffer[layer] : m_efbFramebuffer.back();
}
static GLuint GetXFBFramebuffer() { return m_xfbFramebuffer; }
// Resolved framebuffer is only used in MSAA mode.
static GLuint GetResolvedFramebuffer() { return m_resolvedFramebuffer[0]; }
static void SetFramebuffer(GLuint fb);
static void FramebufferTexture(GLenum target, GLenum attachment, GLenum textarget, GLuint texture,
GLint level);
// Resolved framebuffer is only used in MSAA mode.
static GLuint GetResolvedFramebuffer() { return m_resolvedFramebuffer[0]; }
// If in MSAA mode, this will perform a resolve of the specified rectangle, and return the resolve
// target as a texture ID.
// Thus, this call may be expensive. Don't repeat it unnecessarily.
// If not in MSAA mode, will just return the render target texture ID.
// After calling this, before you render anything else, you MUST bind the framebuffer you want to
// draw to.
static GLuint ResolveAndGetRenderTarget(const EFBRectangle& rect);
static void SetFramebuffer(GLuint fb);
static void FramebufferTexture(GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level);
// Same as above but for the depth Target.
// After calling this, before you render anything else, you MUST bind the framebuffer you want to
// draw to.
static GLuint ResolveAndGetDepthTarget(const EFBRectangle& rect);
// If in MSAA mode, this will perform a resolve of the specified rectangle, and return the resolve target as a texture ID.
// Thus, this call may be expensive. Don't repeat it unnecessarily.
// If not in MSAA mode, will just return the render target texture ID.
// After calling this, before you render anything else, you MUST bind the framebuffer you want to draw to.
static GLuint ResolveAndGetRenderTarget(const EFBRectangle &rect);
// Convert EFB content on pixel format change.
// convtype=0 -> rgb8->rgba6, convtype=2 -> rgba6->rgb8
static void ReinterpretPixelData(unsigned int convtype);
// Same as above but for the depth Target.
// After calling this, before you render anything else, you MUST bind the framebuffer you want to draw to.
static GLuint ResolveAndGetDepthTarget(const EFBRectangle &rect);
// Convert EFB content on pixel format change.
// convtype=0 -> rgb8->rgba6, convtype=2 -> rgba6->rgb8
static void ReinterpretPixelData(unsigned int convtype);
static void PokeEFB(EFBAccessType type, const EfbPokeData* points, size_t num_points);
static void PokeEFB(EFBAccessType type, const EfbPokeData* points, size_t num_points);
private:
std::unique_ptr<XFBSourceBase> CreateXFBSource(unsigned int target_width, unsigned int target_height, unsigned int layers) override;
void GetTargetSize(unsigned int *width, unsigned int *height) override;
std::unique_ptr<XFBSourceBase> CreateXFBSource(unsigned int target_width,
unsigned int target_height,
unsigned int layers) override;
void GetTargetSize(unsigned int* width, unsigned int* height) override;
void CopyToRealXFB(u32 xfbAddr, u32 fbStride, u32 fbHeight, const EFBRectangle& sourceRc,float Gamma) override;
void CopyToRealXFB(u32 xfbAddr, u32 fbStride, u32 fbHeight, const EFBRectangle& sourceRc,
float Gamma) override;
static int m_targetWidth;
static int m_targetHeight;
static int m_msaaSamples;
static int m_targetWidth;
static int m_targetHeight;
static int m_msaaSamples;
static GLenum m_textureType;
static std::vector<GLuint> m_efbFramebuffer;
static GLuint m_xfbFramebuffer;
static GLuint m_efbColor;
static GLuint m_efbDepth;
static GLuint m_efbColorSwap;// will be hot swapped with m_efbColor when reinterpreting EFB pixel formats
static GLenum m_textureType;
static std::vector<GLuint> m_efbFramebuffer;
static GLuint m_xfbFramebuffer;
static GLuint m_efbColor;
static GLuint m_efbDepth;
static GLuint
m_efbColorSwap; // will be hot swapped with m_efbColor when reinterpreting EFB pixel formats
// Only used in MSAA mode, TODO: try to avoid them
static std::vector<GLuint> m_resolvedFramebuffer;
static GLuint m_resolvedColorTexture;
static GLuint m_resolvedDepthTexture;
// Only used in MSAA mode, TODO: try to avoid them
static std::vector<GLuint> m_resolvedFramebuffer;
static GLuint m_resolvedColorTexture;
static GLuint m_resolvedDepthTexture;
// For pixel format draw
static SHADER m_pixel_format_shaders[2];
// For pixel format draw
static SHADER m_pixel_format_shaders[2];
// For EFB pokes
static GLuint m_EfbPokes_VBO;
static GLuint m_EfbPokes_VAO;
static SHADER m_EfbPokes;
// For EFB pokes
static GLuint m_EfbPokes_VBO;
static GLuint m_EfbPokes_VAO;
static SHADER m_EfbPokes;
};
} // namespace OGL

76
Source/Core/VideoBackends/OGL/NativeVertexFormat.cpp
View File

@ -3,10 +3,10 @@
// Refer to the license.txt file included.
#include "Common/Common.h"
#include "Common/GL/GLUtil.h"
#include "Common/MemoryUtil.h"
#include "Common/x64ABI.h"
#include "Common/x64Emitter.h"
#include "Common/GL/GLUtil.h"
#include "VideoBackends/OGL/ProgramShaderCache.h"
#include "VideoBackends/OGL/VertexManager.h"
@ -20,73 +20,73 @@
namespace OGL
{
NativeVertexFormat* VertexManager::CreateNativeVertexFormat(const PortableVertexDeclaration& vtx_decl)
NativeVertexFormat*
VertexManager::CreateNativeVertexFormat(const PortableVertexDeclaration& vtx_decl)
{
return new GLVertexFormat(vtx_decl);
return new GLVertexFormat(vtx_decl);
}
static inline GLuint VarToGL(VarType t)
{
static const GLuint lookup[5] = {
GL_UNSIGNED_BYTE, GL_BYTE, GL_UNSIGNED_SHORT, GL_SHORT, GL_FLOAT
};
return lookup[t];
static const GLuint lookup[5] = {GL_UNSIGNED_BYTE, GL_BYTE, GL_UNSIGNED_SHORT, GL_SHORT,
GL_FLOAT};
return lookup[t];
}
static void SetPointer(u32 attrib, u32 stride, const AttributeFormat &format)
static void SetPointer(u32 attrib, u32 stride, const AttributeFormat& format)
{
if (!format.enable)
return;
if (!format.enable)
return;
glEnableVertexAttribArray(attrib);
if (format.integer)
glVertexAttribIPointer(attrib, format.components, VarToGL(format.type), stride, (u8*)nullptr + format.offset);
else
glVertexAttribPointer(attrib, format.components, VarToGL(format.type), true, stride, (u8*)nullptr + format.offset);
glEnableVertexAttribArray(attrib);
if (format.integer)
glVertexAttribIPointer(attrib, format.components, VarToGL(format.type), stride,
(u8*)nullptr + format.offset);
else
glVertexAttribPointer(attrib, format.components, VarToGL(format.type), true, stride,
(u8*)nullptr + format.offset);
}
GLVertexFormat::GLVertexFormat(const PortableVertexDeclaration& _vtx_decl)
{
this->vtx_decl = _vtx_decl;
u32 vertex_stride = _vtx_decl.stride;
this->vtx_decl = _vtx_decl;
u32 vertex_stride = _vtx_decl.stride;
// We will not allow vertex components causing uneven strides.
if (vertex_stride & 3)
PanicAlert("Uneven vertex stride: %i", vertex_stride);
// We will not allow vertex components causing uneven strides.
if (vertex_stride & 3)
PanicAlert("Uneven vertex stride: %i", vertex_stride);
VertexManager* const vm = static_cast<VertexManager*>(g_vertex_manager.get());
VertexManager* const vm = static_cast<VertexManager*>(g_vertex_manager.get());
glGenVertexArrays(1, &VAO);
glBindVertexArray(VAO);
glGenVertexArrays(1, &VAO);
glBindVertexArray(VAO);
// the element buffer is bound directly to the vao, so we must it set for every vao
glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, vm->m_index_buffers);
glBindBuffer(GL_ARRAY_BUFFER, vm->m_vertex_buffers);
// the element buffer is bound directly to the vao, so we must it set for every vao
glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, vm->m_index_buffers);
glBindBuffer(GL_ARRAY_BUFFER, vm->m_vertex_buffers);
SetPointer(SHADER_POSITION_ATTRIB, vertex_stride, _vtx_decl.position);
SetPointer(SHADER_POSITION_ATTRIB, vertex_stride, _vtx_decl.position);
for (int i = 0; i < 3; i++)
SetPointer(SHADER_NORM0_ATTRIB+i, vertex_stride, _vtx_decl.normals[i]);
for (int i = 0; i < 3; i++)
SetPointer(SHADER_NORM0_ATTRIB + i, vertex_stride, _vtx_decl.normals[i]);
for (int i = 0; i < 2; i++)
SetPointer(SHADER_COLOR0_ATTRIB+i, vertex_stride, _vtx_decl.colors[i]);
for (int i = 0; i < 2; i++)
SetPointer(SHADER_COLOR0_ATTRIB + i, vertex_stride, _vtx_decl.colors[i]);
for (int i = 0; i < 8; i++)
SetPointer(SHADER_TEXTURE0_ATTRIB+i, vertex_stride, _vtx_decl.texcoords[i]);
for (int i = 0; i < 8; i++)
SetPointer(SHADER_TEXTURE0_ATTRIB + i, vertex_stride, _vtx_decl.texcoords[i]);
SetPointer(SHADER_POSMTX_ATTRIB, vertex_stride, _vtx_decl.posmtx);
SetPointer(SHADER_POSMTX_ATTRIB, vertex_stride, _vtx_decl.posmtx);
vm->m_last_vao = VAO;
vm->m_last_vao = VAO;
}
GLVertexFormat::~GLVertexFormat()
{
glDeleteVertexArrays(1, &VAO);
glDeleteVertexArrays(1, &VAO);
}
void GLVertexFormat::SetupVertexPointers()
{
}
}

264
Source/Core/VideoBackends/OGL/PerfQuery.cpp
View File

@ -16,243 +16,243 @@ namespace OGL
{
std::unique_ptr<PerfQueryBase> GetPerfQuery()
{
if (GLInterface->GetMode() == GLInterfaceMode::MODE_OPENGLES3 &&
GLExtensions::Supports("GL_NV_occlusion_query_samples"))
return std::make_unique<PerfQueryGLESNV>();
if (GLInterface->GetMode() == GLInterfaceMode::MODE_OPENGLES3 &&
GLExtensions::Supports("GL_NV_occlusion_query_samples"))
return std::make_unique<PerfQueryGLESNV>();
if (GLInterface->GetMode() == GLInterfaceMode::MODE_OPENGLES3)
return std::make_unique<PerfQueryGL>(GL_ANY_SAMPLES_PASSED);
if (GLInterface->GetMode() == GLInterfaceMode::MODE_OPENGLES3)
return std::make_unique<PerfQueryGL>(GL_ANY_SAMPLES_PASSED);
return std::make_unique<PerfQueryGL>(GL_SAMPLES_PASSED);
return std::make_unique<PerfQueryGL>(GL_SAMPLES_PASSED);
}
PerfQuery::PerfQuery()
: m_query_read_pos()
PerfQuery::PerfQuery() : m_query_read_pos()
{
ResetQuery();
ResetQuery();
}
void PerfQuery::EnableQuery(PerfQueryGroup type)
{
m_query->EnableQuery(type);
m_query->EnableQuery(type);
}
void PerfQuery::DisableQuery(PerfQueryGroup type)
{
m_query->DisableQuery(type);
m_query->DisableQuery(type);
}
bool PerfQuery::IsFlushed() const
{
return 0 == m_query_count;
return 0 == m_query_count;
}
// TODO: could selectively flush things, but I don't think that will do much
void PerfQuery::FlushResults()
{
m_query->FlushResults();
m_query->FlushResults();
}
void PerfQuery::ResetQuery()
{
m_query_count = 0;
std::fill_n(m_results, ArraySize(m_results), 0);
m_query_count = 0;
std::fill_n(m_results, ArraySize(m_results), 0);
}
u32 PerfQuery::GetQueryResult(PerfQueryType type)
{
u32 result = 0;
u32 result = 0;
if (type == PQ_ZCOMP_INPUT_ZCOMPLOC || type == PQ_ZCOMP_OUTPUT_ZCOMPLOC)
{
result = m_results[PQG_ZCOMP_ZCOMPLOC];
}
else if (type == PQ_ZCOMP_INPUT || type == PQ_ZCOMP_OUTPUT)
{
result = m_results[PQG_ZCOMP];
}
else if (type == PQ_BLEND_INPUT)
{
result = m_results[PQG_ZCOMP] + m_results[PQG_ZCOMP_ZCOMPLOC];
}
else if (type == PQ_EFB_COPY_CLOCKS)
{
result = m_results[PQG_EFB_COPY_CLOCKS];
}
if (type == PQ_ZCOMP_INPUT_ZCOMPLOC || type == PQ_ZCOMP_OUTPUT_ZCOMPLOC)
{
result = m_results[PQG_ZCOMP_ZCOMPLOC];
}
else if (type == PQ_ZCOMP_INPUT || type == PQ_ZCOMP_OUTPUT)
{
result = m_results[PQG_ZCOMP];
}
else if (type == PQ_BLEND_INPUT)
{
result = m_results[PQG_ZCOMP] + m_results[PQG_ZCOMP_ZCOMPLOC];
}
else if (type == PQ_EFB_COPY_CLOCKS)
{
result = m_results[PQG_EFB_COPY_CLOCKS];
}
return result / 4;
return result / 4;
}
// Implementations
PerfQueryGL::PerfQueryGL(GLenum query_type)
: m_query_type(query_type)
PerfQueryGL::PerfQueryGL(GLenum query_type) : m_query_type(query_type)
{
for (ActiveQuery& query : m_query_buffer)
glGenQueries(1, &query.query_id);
for (ActiveQuery& query : m_query_buffer)
glGenQueries(1, &query.query_id);
}
PerfQueryGL::~PerfQueryGL()
{
for (ActiveQuery& query : m_query_buffer)
glDeleteQueries(1, &query.query_id);
for (ActiveQuery& query : m_query_buffer)
glDeleteQueries(1, &query.query_id);
}
void PerfQueryGL::EnableQuery(PerfQueryGroup type)
{
// Is this sane?
if (m_query_count > m_query_buffer.size() / 2)
WeakFlush();
// Is this sane?
if (m_query_count > m_query_buffer.size() / 2)
WeakFlush();
if (m_query_buffer.size() == m_query_count)
{
FlushOne();
//ERROR_LOG(VIDEO, "Flushed query buffer early!");
}
if (m_query_buffer.size() == m_query_count)
{
FlushOne();
// ERROR_LOG(VIDEO, "Flushed query buffer early!");
}
// start query
if (type == PQG_ZCOMP_ZCOMPLOC || type == PQG_ZCOMP)
{
auto& entry = m_query_buffer[(m_query_read_pos + m_query_count) % m_query_buffer.size()];
// start query
if (type == PQG_ZCOMP_ZCOMPLOC || type == PQG_ZCOMP)
{
auto& entry = m_query_buffer[(m_query_read_pos + m_query_count) % m_query_buffer.size()];
glBeginQuery(m_query_type, entry.query_id);
entry.query_type = type;
glBeginQuery(m_query_type, entry.query_id);
entry.query_type = type;
++m_query_count;
}
++m_query_count;
}
}
void PerfQueryGL::DisableQuery(PerfQueryGroup type)
{
// stop query
if (type == PQG_ZCOMP_ZCOMPLOC || type == PQG_ZCOMP)
{
glEndQuery(m_query_type);
}
// stop query
if (type == PQG_ZCOMP_ZCOMPLOC || type == PQG_ZCOMP)
{
glEndQuery(m_query_type);
}
}
void PerfQueryGL::WeakFlush()
{
while (!IsFlushed())
{
auto& entry = m_query_buffer[m_query_read_pos];
while (!IsFlushed())
{
auto& entry = m_query_buffer[m_query_read_pos];
GLuint result = GL_FALSE;
glGetQueryObjectuiv(entry.query_id, GL_QUERY_RESULT_AVAILABLE, &result);
GLuint result = GL_FALSE;
glGetQueryObjectuiv(entry.query_id, GL_QUERY_RESULT_AVAILABLE, &result);
if (GL_TRUE == result)
{
FlushOne();
}
else
{
break;
}
}
if (GL_TRUE == result)
{
FlushOne();
}
else
{
break;
}
}
}
void PerfQueryGL::FlushOne()
{
auto& entry = m_query_buffer[m_query_read_pos];
auto& entry = m_query_buffer[m_query_read_pos];
GLuint result = 0;
glGetQueryObjectuiv(entry.query_id, GL_QUERY_RESULT, &result);
GLuint result = 0;
glGetQueryObjectuiv(entry.query_id, GL_QUERY_RESULT, &result);
// NOTE: Reported pixel metrics should be referenced to native resolution
m_results[entry.query_type] += (u64)result * EFB_WIDTH / g_renderer->GetTargetWidth() * EFB_HEIGHT / g_renderer->GetTargetHeight();
// NOTE: Reported pixel metrics should be referenced to native resolution
m_results[entry.query_type] += (u64)result * EFB_WIDTH / g_renderer->GetTargetWidth() *
EFB_HEIGHT / g_renderer->GetTargetHeight();
m_query_read_pos = (m_query_read_pos + 1) % m_query_buffer.size();
--m_query_count;
m_query_read_pos = (m_query_read_pos + 1) % m_query_buffer.size();
--m_query_count;
}
// TODO: could selectively flush things, but I don't think that will do much
void PerfQueryGL::FlushResults()
{
while (!IsFlushed())
FlushOne();
while (!IsFlushed())
FlushOne();
}
PerfQueryGLESNV::PerfQueryGLESNV()
{
for (ActiveQuery& query : m_query_buffer)
glGenOcclusionQueriesNV(1, &query.query_id);
for (ActiveQuery& query : m_query_buffer)
glGenOcclusionQueriesNV(1, &query.query_id);
}
PerfQueryGLESNV::~PerfQueryGLESNV()
{
for (ActiveQuery& query : m_query_buffer)
glDeleteOcclusionQueriesNV(1, &query.query_id);
for (ActiveQuery& query : m_query_buffer)
glDeleteOcclusionQueriesNV(1, &query.query_id);
}
void PerfQueryGLESNV::EnableQuery(PerfQueryGroup type)
{
// Is this sane?
if (m_query_count > m_query_buffer.size() / 2)
WeakFlush();
// Is this sane?
if (m_query_count > m_query_buffer.size() / 2)
WeakFlush();
if (m_query_buffer.size() == m_query_count)
{
FlushOne();
//ERROR_LOG(VIDEO, "Flushed query buffer early!");
}
if (m_query_buffer.size() == m_query_count)
{
FlushOne();
// ERROR_LOG(VIDEO, "Flushed query buffer early!");
}
// start query
if (type == PQG_ZCOMP_ZCOMPLOC || type == PQG_ZCOMP)
{
auto& entry = m_query_buffer[(m_query_read_pos + m_query_count) % m_query_buffer.size()];
// start query
if (type == PQG_ZCOMP_ZCOMPLOC || type == PQG_ZCOMP)
{
auto& entry = m_query_buffer[(m_query_read_pos + m_query_count) % m_query_buffer.size()];
glBeginOcclusionQueryNV(entry.query_id);
entry.query_type = type;
glBeginOcclusionQueryNV(entry.query_id);
entry.query_type = type;
++m_query_count;
}
++m_query_count;
}
}
void PerfQueryGLESNV::DisableQuery(PerfQueryGroup type)
{
// stop query
if (type == PQG_ZCOMP_ZCOMPLOC || type == PQG_ZCOMP)
{
glEndOcclusionQueryNV();
}
// stop query
if (type == PQG_ZCOMP_ZCOMPLOC || type == PQG_ZCOMP)
{
glEndOcclusionQueryNV();
}
}
void PerfQueryGLESNV::WeakFlush()
{
while (!IsFlushed())
{
auto& entry = m_query_buffer[m_query_read_pos];
while (!IsFlushed())
{
auto& entry = m_query_buffer[m_query_read_pos];
GLuint result = GL_FALSE;
glGetOcclusionQueryuivNV(entry.query_id, GL_PIXEL_COUNT_AVAILABLE_NV, &result);
GLuint result = GL_FALSE;
glGetOcclusionQueryuivNV(entry.query_id, GL_PIXEL_COUNT_AVAILABLE_NV, &result);
if (GL_TRUE == result)
{
FlushOne();
}
else
{
break;
}
}
if (GL_TRUE == result)
{
FlushOne();
}
else
{
break;
}
}
}
void PerfQueryGLESNV::FlushOne()
{
auto& entry = m_query_buffer[m_query_read_pos];
auto& entry = m_query_buffer[m_query_read_pos];
GLuint result = 0;
glGetOcclusionQueryuivNV(entry.query_id, GL_OCCLUSION_TEST_RESULT_HP, &result);
GLuint result = 0;
glGetOcclusionQueryuivNV(entry.query_id, GL_OCCLUSION_TEST_RESULT_HP, &result);
// NOTE: Reported pixel metrics should be referenced to native resolution
m_results[entry.query_type] += (u64)result * EFB_WIDTH / g_renderer->GetTargetWidth() * EFB_HEIGHT / g_renderer->GetTargetHeight();
// NOTE: Reported pixel metrics should be referenced to native resolution
m_results[entry.query_type] += (u64)result * EFB_WIDTH / g_renderer->GetTargetWidth() *
EFB_HEIGHT / g_renderer->GetTargetHeight();
m_query_read_pos = (m_query_read_pos + 1) % m_query_buffer.size();
--m_query_count;
m_query_read_pos = (m_query_read_pos + 1) % m_query_buffer.size();
--m_query_count;
}
// TODO: could selectively flush things, but I don't think that will do much
void PerfQueryGLESNV::FlushResults()
{
while (!IsFlushed())
FlushOne();
while (!IsFlushed())
FlushOne();
}
} // namespace
} // namespace

80
Source/Core/VideoBackends/OGL/PerfQuery.h
View File

@ -18,71 +18,67 @@ std::unique_ptr<PerfQueryBase> GetPerfQuery();
class PerfQuery : public PerfQueryBase
{
public:
PerfQuery();
~PerfQuery() {}
void EnableQuery(PerfQueryGroup type) override;
void DisableQuery(PerfQueryGroup type) override;
void ResetQuery() override;
u32 GetQueryResult(PerfQueryType type) override;
void FlushResults() override;
bool IsFlushed() const override;
PerfQuery();
~PerfQuery() {}
void EnableQuery(PerfQueryGroup type) override;
void DisableQuery(PerfQueryGroup type) override;
void ResetQuery() override;
u32 GetQueryResult(PerfQueryType type) override;
void FlushResults() override;
bool IsFlushed() const override;
protected:
struct ActiveQuery
{
GLuint query_id;
PerfQueryGroup query_type;
};
struct ActiveQuery
{
GLuint query_id;
PerfQueryGroup query_type;
};
// when testing in SMS: 64 was too small, 128 was ok
static const u32 PERF_QUERY_BUFFER_SIZE = 512;
// when testing in SMS: 64 was too small, 128 was ok
static const u32 PERF_QUERY_BUFFER_SIZE = 512;
// This contains gl query objects with unretrieved results.
std::array<ActiveQuery, PERF_QUERY_BUFFER_SIZE> m_query_buffer;
u32 m_query_read_pos;
// This contains gl query objects with unretrieved results.
std::array<ActiveQuery, PERF_QUERY_BUFFER_SIZE> m_query_buffer;
u32 m_query_read_pos;
private:
// Implementation
std::unique_ptr<PerfQuery> m_query;
// Implementation
std::unique_ptr<PerfQuery> m_query;
};
// Implementations
class PerfQueryGL : public PerfQuery
{
public:
PerfQueryGL(GLenum query_type);
~PerfQueryGL();
PerfQueryGL(GLenum query_type);
~PerfQueryGL();
void EnableQuery(PerfQueryGroup type) override;
void DisableQuery(PerfQueryGroup type) override;
void FlushResults() override;
void EnableQuery(PerfQueryGroup type) override;
void DisableQuery(PerfQueryGroup type) override;
void FlushResults() override;
private:
void WeakFlush();
// Only use when non-empty
void FlushOne();
void WeakFlush();
// Only use when non-empty
void FlushOne();
GLenum m_query_type;
GLenum m_query_type;
};
class PerfQueryGLESNV : public PerfQuery
{
public:
PerfQueryGLESNV();
~PerfQueryGLESNV();
PerfQueryGLESNV();
~PerfQueryGLESNV();
void EnableQuery(PerfQueryGroup type) override;
void DisableQuery(PerfQueryGroup type) override;
void FlushResults() override;
void EnableQuery(PerfQueryGroup type) override;
void DisableQuery(PerfQueryGroup type) override;
void FlushResults() override;
private:
void WeakFlush();
// Only use when non-empty
void FlushOne();
void WeakFlush();
// Only use when non-empty
void FlushOne();
};
} // namespace
} // namespace

406
Source/Core/VideoBackends/OGL/PostProcessing.cpp
View File

@ -20,262 +20,254 @@
namespace OGL
{
static const char s_vertex_shader[] = "out vec2 uv0;\n"
"uniform vec4 src_rect;\n"
"void main(void) {\n"
" vec2 rawpos = vec2(gl_VertexID&1, gl_VertexID&2);\n"
" gl_Position = vec4(rawpos*2.0-1.0, 0.0, 1.0);\n"
" uv0 = rawpos * src_rect.zw + src_rect.xy;\n"
"}\n";
static const char s_vertex_shader[] =
"out vec2 uv0;\n"
"uniform vec4 src_rect;\n"
"void main(void) {\n"
" vec2 rawpos = vec2(gl_VertexID&1, gl_VertexID&2);\n"
" gl_Position = vec4(rawpos*2.0-1.0, 0.0, 1.0);\n"
" uv0 = rawpos * src_rect.zw + src_rect.xy;\n"
"}\n";
OpenGLPostProcessing::OpenGLPostProcessing()
: m_initialized(false)
OpenGLPostProcessing::OpenGLPostProcessing() : m_initialized(false)
{
CreateHeader();
CreateHeader();
}
OpenGLPostProcessing::~OpenGLPostProcessing()
{
m_shader.Destroy();
m_shader.Destroy();
}
void OpenGLPostProcessing::BlitFromTexture(TargetRectangle src, TargetRectangle dst,
int src_texture, int src_width, int src_height, int layer)
int src_texture, int src_width, int src_height,
int layer)
{
ApplyShader();
ApplyShader();
glBindFramebuffer(GL_DRAW_FRAMEBUFFER, 0);
glBindFramebuffer(GL_DRAW_FRAMEBUFFER, 0);
glViewport(dst.left, dst.bottom, dst.GetWidth(), dst.GetHeight());
glViewport(dst.left, dst.bottom, dst.GetWidth(), dst.GetHeight());
OpenGL_BindAttributelessVAO();
OpenGL_BindAttributelessVAO();
m_shader.Bind();
m_shader.Bind();
glUniform4f(m_uniform_resolution, (float)src_width, (float)src_height, 1.0f / (float)src_width, 1.0f / (float)src_height);
glUniform4f(m_uniform_src_rect, src.left / (float) src_width, src.bottom / (float) src_height,
src.GetWidth() / (float) src_width, src.GetHeight() / (float) src_height);
glUniform1ui(m_uniform_time, (GLuint)m_timer.GetTimeElapsed());
glUniform1i(m_uniform_layer, layer);
glUniform4f(m_uniform_resolution, (float)src_width, (float)src_height, 1.0f / (float)src_width,
1.0f / (float)src_height);
glUniform4f(m_uniform_src_rect, src.left / (float)src_width, src.bottom / (float)src_height,
src.GetWidth() / (float)src_width, src.GetHeight() / (float)src_height);
glUniform1ui(m_uniform_time, (GLuint)m_timer.GetTimeElapsed());
glUniform1i(m_uniform_layer, layer);
if (m_config.IsDirty())
{
for (auto& it : m_config.GetOptions())
{
if (it.second.m_dirty)
{
switch (it.second.m_type)
{
case PostProcessingShaderConfiguration::ConfigurationOption::OptionType::OPTION_BOOL:
glUniform1i(m_uniform_bindings[it.first], it.second.m_bool_value);
break;
case PostProcessingShaderConfiguration::ConfigurationOption::OptionType::OPTION_INTEGER:
switch (it.second.m_integer_values.size())
{
case 1:
glUniform1i(m_uniform_bindings[it.first], it.second.m_integer_values[0]);
break;
case 2:
glUniform2i(m_uniform_bindings[it.first],
it.second.m_integer_values[0],
it.second.m_integer_values[1]);
break;
case 3:
glUniform3i(m_uniform_bindings[it.first],
it.second.m_integer_values[0],
it.second.m_integer_values[1],
it.second.m_integer_values[2]);
break;
case 4:
glUniform4i(m_uniform_bindings[it.first],
it.second.m_integer_values[0],
it.second.m_integer_values[1],
it.second.m_integer_values[2],
it.second.m_integer_values[3]);
break;
}
break;
case PostProcessingShaderConfiguration::ConfigurationOption::OptionType::OPTION_FLOAT:
switch (it.second.m_float_values.size())
{
case 1:
glUniform1f(m_uniform_bindings[it.first], it.second.m_float_values[0]);
break;
case 2:
glUniform2f(m_uniform_bindings[it.first],
it.second.m_float_values[0],
it.second.m_float_values[1]);
break;
case 3:
glUniform3f(m_uniform_bindings[it.first],
it.second.m_float_values[0],
it.second.m_float_values[1],
it.second.m_float_values[2]);
break;
case 4:
glUniform4f(m_uniform_bindings[it.first],
it.second.m_float_values[0],
it.second.m_float_values[1],
it.second.m_float_values[2],
it.second.m_float_values[3]);
break;
}
break;
}
it.second.m_dirty = false;
}
}
m_config.SetDirty(false);
}
if (m_config.IsDirty())
{
for (auto& it : m_config.GetOptions())
{
if (it.second.m_dirty)
{
switch (it.second.m_type)
{
case PostProcessingShaderConfiguration::ConfigurationOption::OptionType::OPTION_BOOL:
glUniform1i(m_uniform_bindings[it.first], it.second.m_bool_value);
break;
case PostProcessingShaderConfiguration::ConfigurationOption::OptionType::OPTION_INTEGER:
switch (it.second.m_integer_values.size())
{
case 1:
glUniform1i(m_uniform_bindings[it.first], it.second.m_integer_values[0]);
break;
case 2:
glUniform2i(m_uniform_bindings[it.first], it.second.m_integer_values[0],
it.second.m_integer_values[1]);
break;
case 3:
glUniform3i(m_uniform_bindings[it.first], it.second.m_integer_values[0],
it.second.m_integer_values[1], it.second.m_integer_values[2]);
break;
case 4:
glUniform4i(m_uniform_bindings[it.first], it.second.m_integer_values[0],
it.second.m_integer_values[1], it.second.m_integer_values[2],
it.second.m_integer_values[3]);
break;
}
break;
case PostProcessingShaderConfiguration::ConfigurationOption::OptionType::OPTION_FLOAT:
switch (it.second.m_float_values.size())
{
case 1:
glUniform1f(m_uniform_bindings[it.first], it.second.m_float_values[0]);
break;
case 2:
glUniform2f(m_uniform_bindings[it.first], it.second.m_float_values[0],
it.second.m_float_values[1]);
break;
case 3:
glUniform3f(m_uniform_bindings[it.first], it.second.m_float_values[0],
it.second.m_float_values[1], it.second.m_float_values[2]);
break;
case 4:
glUniform4f(m_uniform_bindings[it.first], it.second.m_float_values[0],
it.second.m_float_values[1], it.second.m_float_values[2],
it.second.m_float_values[3]);
break;
}
break;
}
it.second.m_dirty = false;
}
}
m_config.SetDirty(false);
}
glActiveTexture(GL_TEXTURE9);
glBindTexture(GL_TEXTURE_2D_ARRAY, src_texture);
g_sampler_cache->BindLinearSampler(9);
glDrawArrays(GL_TRIANGLE_STRIP, 0, 4);
glActiveTexture(GL_TEXTURE9);
glBindTexture(GL_TEXTURE_2D_ARRAY, src_texture);
g_sampler_cache->BindLinearSampler(9);
glDrawArrays(GL_TRIANGLE_STRIP, 0, 4);
}
void OpenGLPostProcessing::ApplyShader()
{
// shader didn't changed
if (m_initialized && m_config.GetShader() == g_ActiveConfig.sPostProcessingShader)
return;
// shader didn't changed
if (m_initialized && m_config.GetShader() == g_ActiveConfig.sPostProcessingShader)
return;
m_shader.Destroy();
m_uniform_bindings.clear();
m_shader.Destroy();
m_uniform_bindings.clear();
// load shader code
std::string code = m_config.LoadShader();
code = LoadShaderOptions(code);
// load shader code
std::string code = m_config.LoadShader();
code = LoadShaderOptions(code);
// and compile it
if (!ProgramShaderCache::CompileShader(m_shader, s_vertex_shader, code))
{
ERROR_LOG(VIDEO, "Failed to compile post-processing shader %s", m_config.GetShader().c_str());
g_ActiveConfig.sPostProcessingShader.clear();
code = m_config.LoadShader();
ProgramShaderCache::CompileShader(m_shader, s_vertex_shader, code);
}
// and compile it
if (!ProgramShaderCache::CompileShader(m_shader, s_vertex_shader, code))
{
ERROR_LOG(VIDEO, "Failed to compile post-processing shader %s", m_config.GetShader().c_str());
g_ActiveConfig.sPostProcessingShader.clear();
code = m_config.LoadShader();
ProgramShaderCache::CompileShader(m_shader, s_vertex_shader, code);
}
// read uniform locations
m_uniform_resolution = glGetUniformLocation(m_shader.glprogid, "resolution");
m_uniform_time = glGetUniformLocation(m_shader.glprogid, "time");
m_uniform_src_rect = glGetUniformLocation(m_shader.glprogid, "src_rect");
m_uniform_layer = glGetUniformLocation(m_shader.glprogid, "layer");
// read uniform locations
m_uniform_resolution = glGetUniformLocation(m_shader.glprogid, "resolution");
m_uniform_time = glGetUniformLocation(m_shader.glprogid, "time");
m_uniform_src_rect = glGetUniformLocation(m_shader.glprogid, "src_rect");
m_uniform_layer = glGetUniformLocation(m_shader.glprogid, "layer");
for (const auto& it : m_config.GetOptions())
{
std::string glsl_name = "option_" + it.first;
m_uniform_bindings[it.first] = glGetUniformLocation(m_shader.glprogid, glsl_name.c_str());
}
m_initialized = true;
for (const auto& it : m_config.GetOptions())
{
std::string glsl_name = "option_" + it.first;
m_uniform_bindings[it.first] = glGetUniformLocation(m_shader.glprogid, glsl_name.c_str());
}
m_initialized = true;
}
void OpenGLPostProcessing::CreateHeader()
{
m_glsl_header =
// Required variables
// Shouldn't be accessed directly by the PP shader
// Texture sampler
"SAMPLER_BINDING(8) uniform sampler2D samp8;\n"
"SAMPLER_BINDING(9) uniform sampler2DArray samp9;\n"
m_glsl_header =
// Required variables
// Shouldn't be accessed directly by the PP shader
// Texture sampler
"SAMPLER_BINDING(8) uniform sampler2D samp8;\n"
"SAMPLER_BINDING(9) uniform sampler2DArray samp9;\n"
// Output variable
"out float4 ocol0;\n"
// Input coordinates
"in float2 uv0;\n"
// Resolution
"uniform float4 resolution;\n"
// Time
"uniform uint time;\n"
// Layer
"uniform int layer;\n"
// Output variable
"out float4 ocol0;\n"
// Input coordinates
"in float2 uv0;\n"
// Resolution
"uniform float4 resolution;\n"
// Time
"uniform uint time;\n"
// Layer
"uniform int layer;\n"
// Interfacing functions
"float4 Sample()\n"
"{\n"
"\treturn texture(samp9, float3(uv0, layer));\n"
"}\n"
// Interfacing functions
"float4 Sample()\n"
"{\n"
"\treturn texture(samp9, float3(uv0, layer));\n"
"}\n"
"float4 SampleLocation(float2 location)\n"
"{\n"
"\treturn texture(samp9, float3(location, layer));\n"
"}\n"
"float4 SampleLocation(float2 location)\n"
"{\n"
"\treturn texture(samp9, float3(location, layer));\n"
"}\n"
"float4 SampleLayer(int layer)\n"
"{\n"
"\treturn texture(samp9, float3(uv0, layer));\n"
"}\n"
"float4 SampleLayer(int layer)\n"
"{\n"
"\treturn texture(samp9, float3(uv0, layer));\n"
"}\n"
"#define SampleOffset(offset) textureOffset(samp9, float3(uv0, layer), offset)\n"
"#define SampleOffset(offset) textureOffset(samp9, float3(uv0, layer), offset)\n"
"float4 SampleFontLocation(float2 location)\n"
"{\n"
"\treturn texture(samp8, location);\n"
"}\n"
"float4 SampleFontLocation(float2 location)\n"
"{\n"
"\treturn texture(samp8, location);\n"
"}\n"
"float2 GetResolution()\n"
"{\n"
"\treturn resolution.xy;\n"
"}\n"
"float2 GetResolution()\n"
"{\n"
"\treturn resolution.xy;\n"
"}\n"
"float2 GetInvResolution()\n"
"{\n"
"\treturn resolution.zw;\n"
"}\n"
"float2 GetInvResolution()\n"
"{\n"
"\treturn resolution.zw;\n"
"}\n"
"float2 GetCoordinates()\n"
"{\n"
"\treturn uv0;\n"
"}\n"
"float2 GetCoordinates()\n"
"{\n"
"\treturn uv0;\n"
"}\n"
"uint GetTime()\n"
"{\n"
"\treturn time;\n"
"}\n"
"uint GetTime()\n"
"{\n"
"\treturn time;\n"
"}\n"
"void SetOutput(float4 color)\n"
"{\n"
"\tocol0 = color;\n"
"}\n"
"void SetOutput(float4 color)\n"
"{\n"
"\tocol0 = color;\n"
"}\n"
"#define GetOption(x) (option_##x)\n"
"#define OptionEnabled(x) (option_##x != 0)\n";
"#define GetOption(x) (option_##x)\n"
"#define OptionEnabled(x) (option_##x != 0)\n";
}
std::string OpenGLPostProcessing::LoadShaderOptions(const std::string& code)
{
std::string glsl_options = "";
m_uniform_bindings.clear();
std::string glsl_options = "";
m_uniform_bindings.clear();
for (const auto& it : m_config.GetOptions())
{
if (it.second.m_type == PostProcessingShaderConfiguration::ConfigurationOption::OptionType::OPTION_BOOL)
{
glsl_options += StringFromFormat("uniform int option_%s;\n", it.first.c_str());
}
else if (it.second.m_type == PostProcessingShaderConfiguration::ConfigurationOption::OptionType::OPTION_INTEGER)
{
u32 count = static_cast<u32>(it.second.m_integer_values.size());
if (count == 1)
glsl_options += StringFromFormat("uniform int option_%s;\n", it.first.c_str());
else
glsl_options += StringFromFormat("uniform int%d option_%s;\n", count, it.first.c_str());
}
else if (it.second.m_type == PostProcessingShaderConfiguration::ConfigurationOption::OptionType::OPTION_FLOAT)
{
u32 count = static_cast<u32>(it.second.m_float_values.size());
if (count == 1)
glsl_options += StringFromFormat("uniform float option_%s;\n", it.first.c_str());
else
glsl_options += StringFromFormat("uniform float%d option_%s;\n", count, it.first.c_str());
}
for (const auto& it : m_config.GetOptions())
{
if (it.second.m_type ==
PostProcessingShaderConfiguration::ConfigurationOption::OptionType::OPTION_BOOL)
{
glsl_options += StringFromFormat("uniform int option_%s;\n", it.first.c_str());
}
else if (it.second.m_type ==
PostProcessingShaderConfiguration::ConfigurationOption::OptionType::OPTION_INTEGER)
{
u32 count = static_cast<u32>(it.second.m_integer_values.size());
if (count == 1)
glsl_options += StringFromFormat("uniform int option_%s;\n", it.first.c_str());
else
glsl_options += StringFromFormat("uniform int%d option_%s;\n", count, it.first.c_str());
}
else if (it.second.m_type ==
PostProcessingShaderConfiguration::ConfigurationOption::OptionType::OPTION_FLOAT)
{
u32 count = static_cast<u32>(it.second.m_float_values.size());
if (count == 1)
glsl_options += StringFromFormat("uniform float option_%s;\n", it.first.c_str());
else
glsl_options += StringFromFormat("uniform float%d option_%s;\n", count, it.first.c_str());
}
m_uniform_bindings[it.first] = 0;
}
m_uniform_bindings[it.first] = 0;
}
return m_glsl_header + glsl_options + code;
return m_glsl_header + glsl_options + code;
}
} // namespace OGL

31
Source/Core/VideoBackends/OGL/PostProcessing.h
View File

@ -16,30 +16,29 @@
namespace OGL
{
class OpenGLPostProcessing : public PostProcessingShaderImplementation
{
public:
OpenGLPostProcessing();
~OpenGLPostProcessing();
OpenGLPostProcessing();
~OpenGLPostProcessing();
void BlitFromTexture(TargetRectangle src, TargetRectangle dst,
int src_texture, int src_width, int src_height, int layer) override;
void ApplyShader() override;
void BlitFromTexture(TargetRectangle src, TargetRectangle dst, int src_texture, int src_width,
int src_height, int layer) override;
void ApplyShader() override;
private:
bool m_initialized;
SHADER m_shader;
GLuint m_uniform_resolution;
GLuint m_uniform_src_rect;
GLuint m_uniform_time;
GLuint m_uniform_layer;
std::string m_glsl_header;
bool m_initialized;
SHADER m_shader;
GLuint m_uniform_resolution;
GLuint m_uniform_src_rect;
GLuint m_uniform_time;
GLuint m_uniform_layer;
std::string m_glsl_header;
std::unordered_map<std::string, GLuint> m_uniform_bindings;
std::unordered_map<std::string, GLuint> m_uniform_bindings;
void CreateHeader();
std::string LoadShaderOptions(const std::string& code);
void CreateHeader();
std::string LoadShaderOptions(const std::string& code);
};
} // namespace

965
Source/Core/VideoBackends/OGL/ProgramShaderCache.cpp
View File

File diff suppressed because it is too large Load Diff

114
Source/Core/VideoBackends/OGL/ProgramShaderCache.h
View File

@ -6,8 +6,8 @@
#include <tuple>
#include "Common/LinearDiskCache.h"
#include "Common/GL/GLUtil.h"
#include "Common/LinearDiskCache.h"
#include "Core/ConfigManager.h"
@ -17,91 +17,83 @@
namespace OGL
{
class SHADERUID
{
public:
VertexShaderUid vuid;
PixelShaderUid puid;
GeometryShaderUid guid;
VertexShaderUid vuid;
PixelShaderUid puid;
GeometryShaderUid guid;
bool operator <(const SHADERUID& r) const
{
return std::tie(puid, vuid, guid) <
std::tie(r.puid, r.vuid, r.guid);
}
bool operator<(const SHADERUID& r) const
{
return std::tie(puid, vuid, guid) < std::tie(r.puid, r.vuid, r.guid);
}
bool operator ==(const SHADERUID& r) const
{
return std::tie(puid, vuid, guid) ==
std::tie(r.puid, r.vuid, r.guid);
}
bool operator==(const SHADERUID& r) const
{
return std::tie(puid, vuid, guid) == std::tie(r.puid, r.vuid, r.guid);
}
};
struct SHADER
{
SHADER() : glprogid(0) { }
void Destroy()
{
glDeleteProgram(glprogid);
glprogid = 0;
}
GLuint glprogid; // OpenGL program id
SHADER() : glprogid(0) {}
void Destroy()
{
glDeleteProgram(glprogid);
glprogid = 0;
}
GLuint glprogid; // OpenGL program id
std::string strvprog, strpprog, strgprog;
std::string strvprog, strpprog, strgprog;
void SetProgramVariables();
void SetProgramBindings();
void Bind();
void SetProgramVariables();
void SetProgramBindings();
void Bind();
};
class ProgramShaderCache
{
public:
struct PCacheEntry
{
SHADER shader;
bool in_cache;
struct PCacheEntry
{
SHADER shader;
bool in_cache;
void Destroy() { shader.Destroy(); }
};
void Destroy()
{
shader.Destroy();
}
};
static PCacheEntry GetShaderProgram();
static SHADER* SetShader(DSTALPHA_MODE dstAlphaMode, u32 primitive_type);
static void GetShaderId(SHADERUID* uid, DSTALPHA_MODE dstAlphaMode, u32 primitive_type);
static bool CompileShader(SHADER& shader, const std::string& vcode, const std::string& pcode,
const std::string& gcode = "");
static GLuint CompileSingleShader(GLuint type, const std::string& code);
static void UploadConstants();
static PCacheEntry GetShaderProgram();
static SHADER* SetShader(DSTALPHA_MODE dstAlphaMode, u32 primitive_type);
static void GetShaderId(SHADERUID *uid, DSTALPHA_MODE dstAlphaMode, u32 primitive_type);
static bool CompileShader(SHADER &shader, const std::string& vcode, const std::string& pcode, const std::string& gcode = "");
static GLuint CompileSingleShader(GLuint type, const std::string& code);
static void UploadConstants();
static void Init();
static void Shutdown();
static void CreateHeader();
static void Init();
static void Shutdown();
static void CreateHeader();
private:
class ProgramShaderCacheInserter : public LinearDiskCacheReader<SHADERUID, u8>
{
public:
void Read(const SHADERUID &key, const u8 *value, u32 value_size) override;
};
class ProgramShaderCacheInserter : public LinearDiskCacheReader<SHADERUID, u8>
{
public:
void Read(const SHADERUID& key, const u8* value, u32 value_size) override;
};
typedef std::map<SHADERUID, PCacheEntry> PCache;
static PCache pshaders;
static PCacheEntry* last_entry;
static SHADERUID last_uid;
typedef std::map<SHADERUID, PCacheEntry> PCache;
static PCache pshaders;
static PCacheEntry* last_entry;
static SHADERUID last_uid;
static UidChecker<PixelShaderUid, ShaderCode> pixel_uid_checker;
static UidChecker<VertexShaderUid, ShaderCode> vertex_uid_checker;
static UidChecker<GeometryShaderUid, ShaderCode> geometry_uid_checker;
static UidChecker<PixelShaderUid, ShaderCode> pixel_uid_checker;
static UidChecker<VertexShaderUid, ShaderCode> vertex_uid_checker;
static UidChecker<GeometryShaderUid, ShaderCode> geometry_uid_checker;
static u32 s_ubo_buffer_size;
static s32 s_ubo_align;
static u32 s_ubo_buffer_size;
static s32 s_ubo_align;
};
} // namespace OGL

445
Source/Core/VideoBackends/OGL/RasterFont.cpp
View File

@ -14,267 +14,268 @@
namespace OGL
{
static const int CHAR_WIDTH = 8;
static const int CHAR_HEIGHT = 13;
static const int CHAR_OFFSET = 32;
static const int CHAR_COUNT = 95;
static const u8 rasters[CHAR_COUNT][CHAR_HEIGHT] = {
{0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x18, 0x18, 0x00, 0x00, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18},
{0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x36, 0x36, 0x36, 0x36},
{0x00, 0x00, 0x00, 0x66, 0x66, 0xff, 0x66, 0x66, 0xff, 0x66, 0x66, 0x00, 0x00},
{0x00, 0x00, 0x18, 0x7e, 0xff, 0x1b, 0x1f, 0x7e, 0xf8, 0xd8, 0xff, 0x7e, 0x18},
{0x00, 0x00, 0x0e, 0x1b, 0xdb, 0x6e, 0x30, 0x18, 0x0c, 0x76, 0xdb, 0xd8, 0x70},
{0x00, 0x00, 0x7f, 0xc6, 0xcf, 0xd8, 0x70, 0x70, 0xd8, 0xcc, 0xcc, 0x6c, 0x38},
{0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x18, 0x1c, 0x0c, 0x0e},
{0x00, 0x00, 0x0c, 0x18, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x18, 0x0c},
{0x00, 0x00, 0x30, 0x18, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x18, 0x30},
{0x00, 0x00, 0x00, 0x00, 0x99, 0x5a, 0x3c, 0xff, 0x3c, 0x5a, 0x99, 0x00, 0x00},
{0x00, 0x00, 0x00, 0x18, 0x18, 0x18, 0xff, 0xff, 0x18, 0x18, 0x18, 0x00, 0x00},
{0x00, 0x00, 0x30, 0x18, 0x1c, 0x1c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xff, 0xff, 0x00, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x00, 0x38, 0x38, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x60, 0x60, 0x30, 0x30, 0x18, 0x18, 0x0c, 0x0c, 0x06, 0x06, 0x03, 0x03},
{0x00, 0x00, 0x3c, 0x66, 0xc3, 0xe3, 0xf3, 0xdb, 0xcf, 0xc7, 0xc3, 0x66, 0x3c},
{0x00, 0x00, 0x7e, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x78, 0x38, 0x18},
{0x00, 0x00, 0xff, 0xc0, 0xc0, 0x60, 0x30, 0x18, 0x0c, 0x06, 0x03, 0xe7, 0x7e},
{0x00, 0x00, 0x7e, 0xe7, 0x03, 0x03, 0x07, 0x7e, 0x07, 0x03, 0x03, 0xe7, 0x7e},
{0x00, 0x00, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0xff, 0xcc, 0x6c, 0x3c, 0x1c, 0x0c},
{0x00, 0x00, 0x7e, 0xe7, 0x03, 0x03, 0x07, 0xfe, 0xc0, 0xc0, 0xc0, 0xc0, 0xff},
{0x00, 0x00, 0x7e, 0xe7, 0xc3, 0xc3, 0xc7, 0xfe, 0xc0, 0xc0, 0xc0, 0xe7, 0x7e},
{0x00, 0x00, 0x30, 0x30, 0x30, 0x30, 0x18, 0x0c, 0x06, 0x03, 0x03, 0x03, 0xff},
{0x00, 0x00, 0x7e, 0xe7, 0xc3, 0xc3, 0xe7, 0x7e, 0xe7, 0xc3, 0xc3, 0xe7, 0x7e},
{0x00, 0x00, 0x7e, 0xe7, 0x03, 0x03, 0x03, 0x7f, 0xe7, 0xc3, 0xc3, 0xe7, 0x7e},
{0x00, 0x00, 0x00, 0x38, 0x38, 0x00, 0x00, 0x38, 0x38, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x30, 0x18, 0x1c, 0x1c, 0x00, 0x00, 0x1c, 0x1c, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x06, 0x0c, 0x18, 0x30, 0x60, 0xc0, 0x60, 0x30, 0x18, 0x0c, 0x06},
{0x00, 0x00, 0x00, 0x00, 0xff, 0xff, 0x00, 0xff, 0xff, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x60, 0x30, 0x18, 0x0c, 0x06, 0x03, 0x06, 0x0c, 0x18, 0x30, 0x60},
{0x00, 0x00, 0x18, 0x00, 0x00, 0x18, 0x18, 0x0c, 0x06, 0x03, 0xc3, 0xc3, 0x7e},
{0x00, 0x00, 0x3f, 0x60, 0xcf, 0xdb, 0xd3, 0xdd, 0xc3, 0x7e, 0x00, 0x00, 0x00},
{0x00, 0x00, 0xc3, 0xc3, 0xc3, 0xc3, 0xff, 0xc3, 0xc3, 0xc3, 0x66, 0x3c, 0x18},
{0x00, 0x00, 0xfe, 0xc7, 0xc3, 0xc3, 0xc7, 0xfe, 0xc7, 0xc3, 0xc3, 0xc7, 0xfe},
{0x00, 0x00, 0x7e, 0xe7, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0, 0xe7, 0x7e},
{0x00, 0x00, 0xfc, 0xce, 0xc7, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xc7, 0xce, 0xfc},
{0x00, 0x00, 0xff, 0xc0, 0xc0, 0xc0, 0xc0, 0xfc, 0xc0, 0xc0, 0xc0, 0xc0, 0xff},
{0x00, 0x00, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0, 0xfc, 0xc0, 0xc0, 0xc0, 0xff},
{0x00, 0x00, 0x7e, 0xe7, 0xc3, 0xc3, 0xcf, 0xc0, 0xc0, 0xc0, 0xc0, 0xe7, 0x7e},
{0x00, 0x00, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xff, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3},
{0x00, 0x00, 0x7e, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x7e},
{0x00, 0x00, 0x7c, 0xee, 0xc6, 0x06, 0x06, 0x06, 0x06, 0x06, 0x06, 0x06, 0x06},
{0x00, 0x00, 0xc3, 0xc6, 0xcc, 0xd8, 0xf0, 0xe0, 0xf0, 0xd8, 0xcc, 0xc6, 0xc3},
{0x00, 0x00, 0xff, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0},
{0x00, 0x00, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xdb, 0xff, 0xff, 0xe7, 0xc3},
{0x00, 0x00, 0xc7, 0xc7, 0xcf, 0xcf, 0xdf, 0xdb, 0xfb, 0xf3, 0xf3, 0xe3, 0xe3},
{0x00, 0x00, 0x7e, 0xe7, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xe7, 0x7e},
{0x00, 0x00, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0, 0xfe, 0xc7, 0xc3, 0xc3, 0xc7, 0xfe},
{0x00, 0x00, 0x3f, 0x6e, 0xdf, 0xdb, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0x66, 0x3c},
{0x00, 0x00, 0xc3, 0xc6, 0xcc, 0xd8, 0xf0, 0xfe, 0xc7, 0xc3, 0xc3, 0xc7, 0xfe},
{0x00, 0x00, 0x7e, 0xe7, 0x03, 0x03, 0x07, 0x7e, 0xe0, 0xc0, 0xc0, 0xe7, 0x7e},
{0x00, 0x00, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0xff},
{0x00, 0x00, 0x7e, 0xe7, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3},
{0x00, 0x00, 0x18, 0x3c, 0x3c, 0x66, 0x66, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3},
{0x00, 0x00, 0xc3, 0xe7, 0xff, 0xff, 0xdb, 0xdb, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3},
{0x00, 0x00, 0xc3, 0x66, 0x66, 0x3c, 0x3c, 0x18, 0x3c, 0x3c, 0x66, 0x66, 0xc3},
{0x00, 0x00, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x3c, 0x3c, 0x66, 0x66, 0xc3},
{0x00, 0x00, 0xff, 0xc0, 0xc0, 0x60, 0x30, 0x7e, 0x0c, 0x06, 0x03, 0x03, 0xff},
{0x00, 0x00, 0x3c, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x3c},
{0x00, 0x03, 0x03, 0x06, 0x06, 0x0c, 0x0c, 0x18, 0x18, 0x30, 0x30, 0x60, 0x60},
{0x00, 0x00, 0x3c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x3c},
{0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xc3, 0x66, 0x3c, 0x18},
{0xff, 0xff, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x18, 0x38, 0x30, 0x70},
{0x00, 0x00, 0x7f, 0xc3, 0xc3, 0x7f, 0x03, 0xc3, 0x7e, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0xfe, 0xc3, 0xc3, 0xc3, 0xc3, 0xfe, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0},
{0x00, 0x00, 0x7e, 0xc3, 0xc0, 0xc0, 0xc0, 0xc3, 0x7e, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x7f, 0xc3, 0xc3, 0xc3, 0xc3, 0x7f, 0x03, 0x03, 0x03, 0x03, 0x03},
{0x00, 0x00, 0x7f, 0xc0, 0xc0, 0xfe, 0xc3, 0xc3, 0x7e, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x30, 0x30, 0x30, 0x30, 0x30, 0xfc, 0x30, 0x30, 0x30, 0x33, 0x1e},
{0x7e, 0xc3, 0x03, 0x03, 0x7f, 0xc3, 0xc3, 0xc3, 0x7e, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xfe, 0xc0, 0xc0, 0xc0, 0xc0},
{0x00, 0x00, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x00, 0x00, 0x18, 0x00},
{0x38, 0x6c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x00, 0x00, 0x0c, 0x00},
{0x00, 0x00, 0xc6, 0xcc, 0xf8, 0xf0, 0xd8, 0xcc, 0xc6, 0xc0, 0xc0, 0xc0, 0xc0},
{0x00, 0x00, 0x7e, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x78},
{0x00, 0x00, 0xdb, 0xdb, 0xdb, 0xdb, 0xdb, 0xdb, 0xfe, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0xfc, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x7c, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0x7c, 0x00, 0x00, 0x00, 0x00},
{0xc0, 0xc0, 0xc0, 0xfe, 0xc3, 0xc3, 0xc3, 0xc3, 0xfe, 0x00, 0x00, 0x00, 0x00},
{0x03, 0x03, 0x03, 0x7f, 0xc3, 0xc3, 0xc3, 0xc3, 0x7f, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0, 0xe0, 0xfe, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0xfe, 0x03, 0x03, 0x7e, 0xc0, 0xc0, 0x7f, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x1c, 0x36, 0x30, 0x30, 0x30, 0x30, 0xfc, 0x30, 0x30, 0x30, 0x00},
{0x00, 0x00, 0x7e, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x18, 0x3c, 0x3c, 0x66, 0x66, 0xc3, 0xc3, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0xc3, 0xe7, 0xff, 0xdb, 0xc3, 0xc3, 0xc3, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0xc3, 0x66, 0x3c, 0x18, 0x3c, 0x66, 0xc3, 0x00, 0x00, 0x00, 0x00},
{0xc0, 0x60, 0x60, 0x30, 0x18, 0x3c, 0x66, 0x66, 0xc3, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0xff, 0x60, 0x30, 0x18, 0x0c, 0x06, 0xff, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x0f, 0x18, 0x18, 0x18, 0x38, 0xf0, 0x38, 0x18, 0x18, 0x18, 0x0f},
{0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18},
{0x00, 0x00, 0xf0, 0x18, 0x18, 0x18, 0x1c, 0x0f, 0x1c, 0x18, 0x18, 0x18, 0xf0},
{0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x06, 0x8f, 0xf1, 0x60, 0x00, 0x00, 0x00}
};
{0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x18, 0x18, 0x00, 0x00, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18},
{0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x36, 0x36, 0x36, 0x36},
{0x00, 0x00, 0x00, 0x66, 0x66, 0xff, 0x66, 0x66, 0xff, 0x66, 0x66, 0x00, 0x00},
{0x00, 0x00, 0x18, 0x7e, 0xff, 0x1b, 0x1f, 0x7e, 0xf8, 0xd8, 0xff, 0x7e, 0x18},
{0x00, 0x00, 0x0e, 0x1b, 0xdb, 0x6e, 0x30, 0x18, 0x0c, 0x76, 0xdb, 0xd8, 0x70},
{0x00, 0x00, 0x7f, 0xc6, 0xcf, 0xd8, 0x70, 0x70, 0xd8, 0xcc, 0xcc, 0x6c, 0x38},
{0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x18, 0x1c, 0x0c, 0x0e},
{0x00, 0x00, 0x0c, 0x18, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x18, 0x0c},
{0x00, 0x00, 0x30, 0x18, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x18, 0x30},
{0x00, 0x00, 0x00, 0x00, 0x99, 0x5a, 0x3c, 0xff, 0x3c, 0x5a, 0x99, 0x00, 0x00},
{0x00, 0x00, 0x00, 0x18, 0x18, 0x18, 0xff, 0xff, 0x18, 0x18, 0x18, 0x00, 0x00},
{0x00, 0x00, 0x30, 0x18, 0x1c, 0x1c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xff, 0xff, 0x00, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x00, 0x38, 0x38, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x60, 0x60, 0x30, 0x30, 0x18, 0x18, 0x0c, 0x0c, 0x06, 0x06, 0x03, 0x03},
{0x00, 0x00, 0x3c, 0x66, 0xc3, 0xe3, 0xf3, 0xdb, 0xcf, 0xc7, 0xc3, 0x66, 0x3c},
{0x00, 0x00, 0x7e, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x78, 0x38, 0x18},
{0x00, 0x00, 0xff, 0xc0, 0xc0, 0x60, 0x30, 0x18, 0x0c, 0x06, 0x03, 0xe7, 0x7e},
{0x00, 0x00, 0x7e, 0xe7, 0x03, 0x03, 0x07, 0x7e, 0x07, 0x03, 0x03, 0xe7, 0x7e},
{0x00, 0x00, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0xff, 0xcc, 0x6c, 0x3c, 0x1c, 0x0c},
{0x00, 0x00, 0x7e, 0xe7, 0x03, 0x03, 0x07, 0xfe, 0xc0, 0xc0, 0xc0, 0xc0, 0xff},
{0x00, 0x00, 0x7e, 0xe7, 0xc3, 0xc3, 0xc7, 0xfe, 0xc0, 0xc0, 0xc0, 0xe7, 0x7e},
{0x00, 0x00, 0x30, 0x30, 0x30, 0x30, 0x18, 0x0c, 0x06, 0x03, 0x03, 0x03, 0xff},
{0x00, 0x00, 0x7e, 0xe7, 0xc3, 0xc3, 0xe7, 0x7e, 0xe7, 0xc3, 0xc3, 0xe7, 0x7e},
{0x00, 0x00, 0x7e, 0xe7, 0x03, 0x03, 0x03, 0x7f, 0xe7, 0xc3, 0xc3, 0xe7, 0x7e},
{0x00, 0x00, 0x00, 0x38, 0x38, 0x00, 0x00, 0x38, 0x38, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x30, 0x18, 0x1c, 0x1c, 0x00, 0x00, 0x1c, 0x1c, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x06, 0x0c, 0x18, 0x30, 0x60, 0xc0, 0x60, 0x30, 0x18, 0x0c, 0x06},
{0x00, 0x00, 0x00, 0x00, 0xff, 0xff, 0x00, 0xff, 0xff, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x60, 0x30, 0x18, 0x0c, 0x06, 0x03, 0x06, 0x0c, 0x18, 0x30, 0x60},
{0x00, 0x00, 0x18, 0x00, 0x00, 0x18, 0x18, 0x0c, 0x06, 0x03, 0xc3, 0xc3, 0x7e},
{0x00, 0x00, 0x3f, 0x60, 0xcf, 0xdb, 0xd3, 0xdd, 0xc3, 0x7e, 0x00, 0x00, 0x00},
{0x00, 0x00, 0xc3, 0xc3, 0xc3, 0xc3, 0xff, 0xc3, 0xc3, 0xc3, 0x66, 0x3c, 0x18},
{0x00, 0x00, 0xfe, 0xc7, 0xc3, 0xc3, 0xc7, 0xfe, 0xc7, 0xc3, 0xc3, 0xc7, 0xfe},
{0x00, 0x00, 0x7e, 0xe7, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0, 0xe7, 0x7e},
{0x00, 0x00, 0xfc, 0xce, 0xc7, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xc7, 0xce, 0xfc},
{0x00, 0x00, 0xff, 0xc0, 0xc0, 0xc0, 0xc0, 0xfc, 0xc0, 0xc0, 0xc0, 0xc0, 0xff},
{0x00, 0x00, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0, 0xfc, 0xc0, 0xc0, 0xc0, 0xff},
{0x00, 0x00, 0x7e, 0xe7, 0xc3, 0xc3, 0xcf, 0xc0, 0xc0, 0xc0, 0xc0, 0xe7, 0x7e},
{0x00, 0x00, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xff, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3},
{0x00, 0x00, 0x7e, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x7e},
{0x00, 0x00, 0x7c, 0xee, 0xc6, 0x06, 0x06, 0x06, 0x06, 0x06, 0x06, 0x06, 0x06},
{0x00, 0x00, 0xc3, 0xc6, 0xcc, 0xd8, 0xf0, 0xe0, 0xf0, 0xd8, 0xcc, 0xc6, 0xc3},
{0x00, 0x00, 0xff, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0},
{0x00, 0x00, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xdb, 0xff, 0xff, 0xe7, 0xc3},
{0x00, 0x00, 0xc7, 0xc7, 0xcf, 0xcf, 0xdf, 0xdb, 0xfb, 0xf3, 0xf3, 0xe3, 0xe3},
{0x00, 0x00, 0x7e, 0xe7, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xe7, 0x7e},
{0x00, 0x00, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0, 0xfe, 0xc7, 0xc3, 0xc3, 0xc7, 0xfe},
{0x00, 0x00, 0x3f, 0x6e, 0xdf, 0xdb, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0x66, 0x3c},
{0x00, 0x00, 0xc3, 0xc6, 0xcc, 0xd8, 0xf0, 0xfe, 0xc7, 0xc3, 0xc3, 0xc7, 0xfe},
{0x00, 0x00, 0x7e, 0xe7, 0x03, 0x03, 0x07, 0x7e, 0xe0, 0xc0, 0xc0, 0xe7, 0x7e},
{0x00, 0x00, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0xff},
{0x00, 0x00, 0x7e, 0xe7, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3},
{0x00, 0x00, 0x18, 0x3c, 0x3c, 0x66, 0x66, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3},
{0x00, 0x00, 0xc3, 0xe7, 0xff, 0xff, 0xdb, 0xdb, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3},
{0x00, 0x00, 0xc3, 0x66, 0x66, 0x3c, 0x3c, 0x18, 0x3c, 0x3c, 0x66, 0x66, 0xc3},
{0x00, 0x00, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x3c, 0x3c, 0x66, 0x66, 0xc3},
{0x00, 0x00, 0xff, 0xc0, 0xc0, 0x60, 0x30, 0x7e, 0x0c, 0x06, 0x03, 0x03, 0xff},
{0x00, 0x00, 0x3c, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x3c},
{0x00, 0x03, 0x03, 0x06, 0x06, 0x0c, 0x0c, 0x18, 0x18, 0x30, 0x30, 0x60, 0x60},
{0x00, 0x00, 0x3c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x3c},
{0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xc3, 0x66, 0x3c, 0x18},
{0xff, 0xff, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x18, 0x38, 0x30, 0x70},
{0x00, 0x00, 0x7f, 0xc3, 0xc3, 0x7f, 0x03, 0xc3, 0x7e, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0xfe, 0xc3, 0xc3, 0xc3, 0xc3, 0xfe, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0},
{0x00, 0x00, 0x7e, 0xc3, 0xc0, 0xc0, 0xc0, 0xc3, 0x7e, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x7f, 0xc3, 0xc3, 0xc3, 0xc3, 0x7f, 0x03, 0x03, 0x03, 0x03, 0x03},
{0x00, 0x00, 0x7f, 0xc0, 0xc0, 0xfe, 0xc3, 0xc3, 0x7e, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x30, 0x30, 0x30, 0x30, 0x30, 0xfc, 0x30, 0x30, 0x30, 0x33, 0x1e},
{0x7e, 0xc3, 0x03, 0x03, 0x7f, 0xc3, 0xc3, 0xc3, 0x7e, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xc3, 0xfe, 0xc0, 0xc0, 0xc0, 0xc0},
{0x00, 0x00, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x00, 0x00, 0x18, 0x00},
{0x38, 0x6c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x0c, 0x00, 0x00, 0x0c, 0x00},
{0x00, 0x00, 0xc6, 0xcc, 0xf8, 0xf0, 0xd8, 0xcc, 0xc6, 0xc0, 0xc0, 0xc0, 0xc0},
{0x00, 0x00, 0x7e, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x78},
{0x00, 0x00, 0xdb, 0xdb, 0xdb, 0xdb, 0xdb, 0xdb, 0xfe, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0xfc, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x7c, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0x7c, 0x00, 0x00, 0x00, 0x00},
{0xc0, 0xc0, 0xc0, 0xfe, 0xc3, 0xc3, 0xc3, 0xc3, 0xfe, 0x00, 0x00, 0x00, 0x00},
{0x03, 0x03, 0x03, 0x7f, 0xc3, 0xc3, 0xc3, 0xc3, 0x7f, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0xc0, 0xc0, 0xc0, 0xc0, 0xc0, 0xe0, 0xfe, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0xfe, 0x03, 0x03, 0x7e, 0xc0, 0xc0, 0x7f, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x1c, 0x36, 0x30, 0x30, 0x30, 0x30, 0xfc, 0x30, 0x30, 0x30, 0x00},
{0x00, 0x00, 0x7e, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0xc6, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x18, 0x3c, 0x3c, 0x66, 0x66, 0xc3, 0xc3, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0xc3, 0xe7, 0xff, 0xdb, 0xc3, 0xc3, 0xc3, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0xc3, 0x66, 0x3c, 0x18, 0x3c, 0x66, 0xc3, 0x00, 0x00, 0x00, 0x00},
{0xc0, 0x60, 0x60, 0x30, 0x18, 0x3c, 0x66, 0x66, 0xc3, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0xff, 0x60, 0x30, 0x18, 0x0c, 0x06, 0xff, 0x00, 0x00, 0x00, 0x00},
{0x00, 0x00, 0x0f, 0x18, 0x18, 0x18, 0x38, 0xf0, 0x38, 0x18, 0x18, 0x18, 0x0f},
{0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18, 0x18},
{0x00, 0x00, 0xf0, 0x18, 0x18, 0x18, 0x1c, 0x0f, 0x1c, 0x18, 0x18, 0x18, 0xf0},
{0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x06, 0x8f, 0xf1, 0x60, 0x00, 0x00, 0x00}};
static const char *s_vertexShaderSrc =
"uniform vec2 charSize;\n"
"uniform vec2 offset;"
"in vec2 rawpos;\n"
"in vec2 tex0;\n"
"out vec2 uv0;\n"
"void main(void) {\n"
" gl_Position = vec4(rawpos + offset,0,1);\n"
" uv0 = tex0 * charSize;\n"
"}\n";
static const char* s_vertexShaderSrc = "uniform vec2 charSize;\n"
"uniform vec2 offset;"
"in vec2 rawpos;\n"
"in vec2 tex0;\n"
"out vec2 uv0;\n"
"void main(void) {\n"
" gl_Position = vec4(rawpos + offset,0,1);\n"
" uv0 = tex0 * charSize;\n"
"}\n";
static const char *s_fragmentShaderSrc =
"SAMPLER_BINDING(8) uniform sampler2D samp8;\n"
"uniform vec4 color;\n"
"in vec2 uv0;\n"
"out vec4 ocol0;\n"
"void main(void) {\n"
" ocol0 = texture(samp8,uv0) * color;\n"
"}\n";
static const char* s_fragmentShaderSrc = "SAMPLER_BINDING(8) uniform sampler2D samp8;\n"
"uniform vec4 color;\n"
"in vec2 uv0;\n"
"out vec4 ocol0;\n"
"void main(void) {\n"
" ocol0 = texture(samp8,uv0) * color;\n"
"}\n";
static SHADER s_shader;
RasterFont::RasterFont()
{
// generate the texture
glGenTextures(1, &texture);
glActiveTexture(GL_TEXTURE8);
glBindTexture(GL_TEXTURE_2D, texture);
std::vector<u32> texture_data(CHAR_WIDTH * CHAR_COUNT * CHAR_HEIGHT);
for (int y = 0; y < CHAR_HEIGHT; y++)
{
for (int c = 0; c < CHAR_COUNT; c++)
{
for (int x = 0; x < CHAR_WIDTH; x++)
{
bool pixel = (0 != (rasters[c][y] & (1 << (CHAR_WIDTH - x - 1))));
texture_data[CHAR_WIDTH * CHAR_COUNT * y + CHAR_WIDTH * c + x] = pixel ? -1 : 0;
}
}
}
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, 0);
glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, CHAR_WIDTH * CHAR_COUNT, CHAR_HEIGHT, 0, GL_RGBA, GL_UNSIGNED_BYTE, texture_data.data());
// generate the texture
glGenTextures(1, &texture);
glActiveTexture(GL_TEXTURE8);
glBindTexture(GL_TEXTURE_2D, texture);
std::vector<u32> texture_data(CHAR_WIDTH * CHAR_COUNT * CHAR_HEIGHT);
for (int y = 0; y < CHAR_HEIGHT; y++)
{
for (int c = 0; c < CHAR_COUNT; c++)
{
for (int x = 0; x < CHAR_WIDTH; x++)
{
bool pixel = (0 != (rasters[c][y] & (1 << (CHAR_WIDTH - x - 1))));
texture_data[CHAR_WIDTH * CHAR_COUNT * y + CHAR_WIDTH * c + x] = pixel ? -1 : 0;
}
}
}
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, 0);
glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, CHAR_WIDTH * CHAR_COUNT, CHAR_HEIGHT, 0, GL_RGBA,
GL_UNSIGNED_BYTE, texture_data.data());
// generate shader
ProgramShaderCache::CompileShader(s_shader, s_vertexShaderSrc, s_fragmentShaderSrc);
s_shader.Bind();
// generate shader
ProgramShaderCache::CompileShader(s_shader, s_vertexShaderSrc, s_fragmentShaderSrc);
s_shader.Bind();
// bound uniforms
glUniform2f(glGetUniformLocation(s_shader.glprogid,"charSize"), 1.0f / GLfloat(CHAR_COUNT), 1.0f);
uniform_color_id = glGetUniformLocation(s_shader.glprogid,"color");
glUniform4f(uniform_color_id, 1.0f, 1.0f, 1.0f, 1.0f);
uniform_offset_id = glGetUniformLocation(s_shader.glprogid, "offset");
glUniform2f(uniform_offset_id, 0.0f, 0.0f);
// bound uniforms
glUniform2f(glGetUniformLocation(s_shader.glprogid, "charSize"), 1.0f / GLfloat(CHAR_COUNT),
1.0f);
uniform_color_id = glGetUniformLocation(s_shader.glprogid, "color");
glUniform4f(uniform_color_id, 1.0f, 1.0f, 1.0f, 1.0f);
uniform_offset_id = glGetUniformLocation(s_shader.glprogid, "offset");
glUniform2f(uniform_offset_id, 0.0f, 0.0f);
// generate VBO & VAO
glGenBuffers(1, &VBO);
glGenVertexArrays(1, &VAO);
glBindBuffer(GL_ARRAY_BUFFER, VBO);
glBindVertexArray(VAO);
glEnableVertexAttribArray(SHADER_POSITION_ATTRIB);
glVertexAttribPointer(SHADER_POSITION_ATTRIB, 2, GL_FLOAT, 0, sizeof(GLfloat)*4, nullptr);
glEnableVertexAttribArray(SHADER_TEXTURE0_ATTRIB);
glVertexAttribPointer(SHADER_TEXTURE0_ATTRIB, 2, GL_FLOAT, 0, sizeof(GLfloat)*4, (GLfloat*)nullptr+2);
// generate VBO & VAO
glGenBuffers(1, &VBO);
glGenVertexArrays(1, &VAO);
glBindBuffer(GL_ARRAY_BUFFER, VBO);
glBindVertexArray(VAO);
glEnableVertexAttribArray(SHADER_POSITION_ATTRIB);
glVertexAttribPointer(SHADER_POSITION_ATTRIB, 2, GL_FLOAT, 0, sizeof(GLfloat) * 4, nullptr);
glEnableVertexAttribArray(SHADER_TEXTURE0_ATTRIB);
glVertexAttribPointer(SHADER_TEXTURE0_ATTRIB, 2, GL_FLOAT, 0, sizeof(GLfloat) * 4,
(GLfloat*)nullptr + 2);
}
RasterFont::~RasterFont()
{
glDeleteTextures(1, &texture);
glDeleteBuffers(1, &VBO);
glDeleteVertexArrays(1, &VAO);
s_shader.Destroy();
glDeleteTextures(1, &texture);
glDeleteBuffers(1, &VBO);
glDeleteVertexArrays(1, &VAO);
s_shader.Destroy();
}
void RasterFont::printMultilineText(const std::string& text, double start_x, double start_y, double z, int bbWidth, int bbHeight, u32 color)
void RasterFont::printMultilineText(const std::string& text, double start_x, double start_y,
double z, int bbWidth, int bbHeight, u32 color)
{
std::vector<GLfloat> vertices(text.length() * 6 * 4);
std::vector<GLfloat> vertices(text.length() * 6 * 4);
int usage = 0;
GLfloat delta_x = GLfloat(2 * CHAR_WIDTH) / GLfloat(bbWidth);
GLfloat delta_y = GLfloat(2 * CHAR_HEIGHT) / GLfloat(bbHeight);
GLfloat border_x = 2.0f / GLfloat(bbWidth);
GLfloat border_y = 4.0f / GLfloat(bbHeight);
int usage = 0;
GLfloat delta_x = GLfloat(2 * CHAR_WIDTH) / GLfloat(bbWidth);
GLfloat delta_y = GLfloat(2 * CHAR_HEIGHT) / GLfloat(bbHeight);
GLfloat border_x = 2.0f / GLfloat(bbWidth);
GLfloat border_y = 4.0f / GLfloat(bbHeight);
GLfloat x = GLfloat(start_x);
GLfloat y = GLfloat(start_y);
GLfloat x = GLfloat(start_x);
GLfloat y = GLfloat(start_y);
for (const char& c : text)
{
if (c == '\n')
{
x = GLfloat(start_x);
y -= delta_y + border_y;
continue;
}
for (const char& c : text)
{
if (c == '\n')
{
x = GLfloat(start_x);
y -= delta_y + border_y;
continue;
}
// do not print spaces, they can be skipped easily
if (c == ' ')
{
x += delta_x + border_x;
continue;
}
// do not print spaces, they can be skipped easily
if (c == ' ')
{
x += delta_x + border_x;
continue;
}
if (c < CHAR_OFFSET || c >= CHAR_COUNT + CHAR_OFFSET)
continue;
if (c < CHAR_OFFSET || c >= CHAR_COUNT + CHAR_OFFSET)
continue;
vertices[usage++] = x;
vertices[usage++] = y;
vertices[usage++] = GLfloat(c - CHAR_OFFSET);
vertices[usage++] = 0.0f;
vertices[usage++] = x;
vertices[usage++] = y;
vertices[usage++] = GLfloat(c - CHAR_OFFSET);
vertices[usage++] = 0.0f;
vertices[usage++] = x + delta_x;
vertices[usage++] = y;
vertices[usage++] = GLfloat(c - CHAR_OFFSET + 1);
vertices[usage++] = 0.0f;
vertices[usage++] = x + delta_x;
vertices[usage++] = y;
vertices[usage++] = GLfloat(c - CHAR_OFFSET + 1);
vertices[usage++] = 0.0f;
vertices[usage++] = x + delta_x;
vertices[usage++] = y + delta_y;
vertices[usage++] = GLfloat(c - CHAR_OFFSET + 1);
vertices[usage++] = 1.0f;
vertices[usage++] = x + delta_x;
vertices[usage++] = y + delta_y;
vertices[usage++] = GLfloat(c - CHAR_OFFSET + 1);
vertices[usage++] = 1.0f;
vertices[usage++] = x;
vertices[usage++] = y;
vertices[usage++] = GLfloat(c - CHAR_OFFSET);
vertices[usage++] = 0.0f;
vertices[usage++] = x;
vertices[usage++] = y;
vertices[usage++] = GLfloat(c - CHAR_OFFSET);
vertices[usage++] = 0.0f;
vertices[usage++] = x + delta_x;
vertices[usage++] = y + delta_y;
vertices[usage++] = GLfloat(c - CHAR_OFFSET + 1);
vertices[usage++] = 1.0f;
vertices[usage++] = x + delta_x;
vertices[usage++] = y + delta_y;
vertices[usage++] = GLfloat(c - CHAR_OFFSET + 1);
vertices[usage++] = 1.0f;
vertices[usage++] = x;
vertices[usage++] = y + delta_y;
vertices[usage++] = GLfloat(c - CHAR_OFFSET);
vertices[usage++] = 1.0f;
vertices[usage++] = x;
vertices[usage++] = y + delta_y;
vertices[usage++] = GLfloat(c - CHAR_OFFSET);
vertices[usage++] = 1.0f;
x += delta_x + border_x;
}
x += delta_x + border_x;
}
if (!usage)
{
return;
}
if (!usage)
{
return;
}
glBindVertexArray(VAO);
glBindBuffer(GL_ARRAY_BUFFER, VBO);
glBufferData(GL_ARRAY_BUFFER, usage*sizeof(GLfloat), vertices.data(), GL_STREAM_DRAW);
glBindVertexArray(VAO);
glBindBuffer(GL_ARRAY_BUFFER, VBO);
glBufferData(GL_ARRAY_BUFFER, usage * sizeof(GLfloat), vertices.data(), GL_STREAM_DRAW);
s_shader.Bind();
s_shader.Bind();
// shadows
glUniform2f(uniform_offset_id, 2.0f / GLfloat(bbWidth), -2.0f / GLfloat(bbHeight));
glUniform4f(uniform_color_id, 0.0f, 0.0f, 0.0f, GLfloat((color>>24)&0xff)/255.f);
glDrawArrays(GL_TRIANGLES, 0, usage/4);
// shadows
glUniform2f(uniform_offset_id, 2.0f / GLfloat(bbWidth), -2.0f / GLfloat(bbHeight));
glUniform4f(uniform_color_id, 0.0f, 0.0f, 0.0f, GLfloat((color >> 24) & 0xff) / 255.f);
glDrawArrays(GL_TRIANGLES, 0, usage / 4);
glUniform2f(uniform_offset_id, 0.0f, 0.0f);
glUniform4f(uniform_color_id, GLfloat((color>>16)&0xff)/255.f,GLfloat((color>>8)&0xff)/255.f,GLfloat((color>>0)&0xff)/255.f,GLfloat((color>>24)&0xff)/255.f);
glDrawArrays(GL_TRIANGLES, 0, usage/4);
glUniform2f(uniform_offset_id, 0.0f, 0.0f);
glUniform4f(uniform_color_id, GLfloat((color >> 16) & 0xff) / 255.f,
GLfloat((color >> 8) & 0xff) / 255.f, GLfloat((color >> 0) & 0xff) / 255.f,
GLfloat((color >> 24) & 0xff) / 255.f);
glDrawArrays(GL_TRIANGLES, 0, usage / 4);
}
}

23
Source/Core/VideoBackends/OGL/RasterFont.h
View File

@ -10,22 +10,21 @@
namespace OGL
{
class RasterFont
{
public:
RasterFont();
~RasterFont();
static int debug;
RasterFont();
~RasterFont();
static int debug;
void printMultilineText(const std::string& text, double x, double y, double z, int bbWidth,
int bbHeight, u32 color);
void printMultilineText(const std::string& text, double x, double y, double z, int bbWidth, int bbHeight, u32 color);
private:
u32 VBO;
u32 VAO;
u32 texture;
u32 uniform_color_id;
u32 uniform_offset_id;
u32 VBO;
u32 VAO;
u32 texture;
u32 uniform_color_id;
u32 uniform_offset_id;
};
}

2661
Source/Core/VideoBackends/OGL/Render.cpp
View File

File diff suppressed because it is too large Load Diff

134
Source/Core/VideoBackends/OGL/Render.h
View File

@ -9,106 +9,108 @@
namespace OGL
{
void ClearEFBCache();
enum GLSL_VERSION
{
GLSL_130,
GLSL_140,
GLSL_150,
GLSL_330,
GLSL_400, // and above
GLSLES_300, // GLES 3.0
GLSLES_310, // GLES 3.1
GLSLES_320, // GLES 3.2
GLSL_130,
GLSL_140,
GLSL_150,
GLSL_330,
GLSL_400, // and above
GLSLES_300, // GLES 3.0
GLSLES_310, // GLES 3.1
GLSLES_320, // GLES 3.2
};
enum class ES_TEXBUF_TYPE
{
TEXBUF_NONE,
TEXBUF_CORE,
TEXBUF_OES,
TEXBUF_EXT
TEXBUF_NONE,
TEXBUF_CORE,
TEXBUF_OES,
TEXBUF_EXT
};
// ogl-only config, so not in VideoConfig.h
struct VideoConfig
{
bool bSupportsGLSLCache;
bool bSupportsGLPinnedMemory;
bool bSupportsGLSync;
bool bSupportsGLBaseVertex;
bool bSupportsGLBufferStorage;
bool bSupportsMSAA;
GLSL_VERSION eSupportedGLSLVersion;
bool bSupportViewportFloat;
bool bSupportsAEP;
bool bSupportsDebug;
bool bSupportsCopySubImage;
u8 SupportedESPointSize;
ES_TEXBUF_TYPE SupportedESTextureBuffer;
bool bSupports2DTextureStorage;
bool bSupports3DTextureStorage;
bool bSupportsEarlyFragmentTests;
bool bSupportsConservativeDepth;
bool bSupportsAniso;
bool bSupportsGLSLCache;
bool bSupportsGLPinnedMemory;
bool bSupportsGLSync;
bool bSupportsGLBaseVertex;
bool bSupportsGLBufferStorage;
bool bSupportsMSAA;
GLSL_VERSION eSupportedGLSLVersion;
bool bSupportViewportFloat;
bool bSupportsAEP;
bool bSupportsDebug;
bool bSupportsCopySubImage;
u8 SupportedESPointSize;
ES_TEXBUF_TYPE SupportedESTextureBuffer;
bool bSupports2DTextureStorage;
bool bSupports3DTextureStorage;
bool bSupportsEarlyFragmentTests;
bool bSupportsConservativeDepth;
bool bSupportsAniso;
const char* gl_vendor;
const char* gl_renderer;
const char* gl_version;
const char* glsl_version;
const char* gl_vendor;
const char* gl_renderer;
const char* gl_version;
const char* glsl_version;
s32 max_samples;
s32 max_samples;
};
extern VideoConfig g_ogl_config;
class Renderer : public ::Renderer
{
public:
Renderer();
~Renderer();
Renderer();
~Renderer();
static void Init();
static void Shutdown();
static void Init();
static void Shutdown();
void SetColorMask() override;
void SetBlendMode(bool forceUpdate) override;
void SetScissorRect(const EFBRectangle& rc) override;
void SetGenerationMode() override;
void SetDepthMode() override;
void SetLogicOpMode() override;
void SetDitherMode() override;
void SetSamplerState(int stage, int texindex, bool custom_tex) override;
void SetInterlacingMode() override;
void SetViewport() override;
void SetColorMask() override;
void SetBlendMode(bool forceUpdate) override;
void SetScissorRect(const EFBRectangle& rc) override;
void SetGenerationMode() override;
void SetDepthMode() override;
void SetLogicOpMode() override;
void SetDitherMode() override;
void SetSamplerState(int stage, int texindex, bool custom_tex) override;
void SetInterlacingMode() override;
void SetViewport() override;
void RenderText(const std::string& text, int left, int top, u32 color) override;
void RenderText(const std::string& text, int left, int top, u32 color) override;
u32 AccessEFB(EFBAccessType type, u32 x, u32 y, u32 poke_data) override;
void PokeEFB(EFBAccessType type, const EfbPokeData* points, size_t num_points) override;
u32 AccessEFB(EFBAccessType type, u32 x, u32 y, u32 poke_data) override;
void PokeEFB(EFBAccessType type, const EfbPokeData* points, size_t num_points) override;
u16 BBoxRead(int index) override;
void BBoxWrite(int index, u16 value) override;
u16 BBoxRead(int index) override;
void BBoxWrite(int index, u16 value) override;
void ResetAPIState() override;
void RestoreAPIState() override;
void ResetAPIState() override;
void RestoreAPIState() override;
TargetRectangle ConvertEFBRectangle(const EFBRectangle& rc) override;
TargetRectangle ConvertEFBRectangle(const EFBRectangle& rc) override;
void SwapImpl(u32 xfbAddr, u32 fbWidth, u32 fbStride, u32 fbHeight, const EFBRectangle& rc, float Gamma) override;
void SwapImpl(u32 xfbAddr, u32 fbWidth, u32 fbStride, u32 fbHeight, const EFBRectangle& rc,
float Gamma) override;
void ClearScreen(const EFBRectangle& rc, bool colorEnable, bool alphaEnable, bool zEnable, u32 color, u32 z) override;
void ClearScreen(const EFBRectangle& rc, bool colorEnable, bool alphaEnable, bool zEnable,
u32 color, u32 z) override;
void ReinterpretPixelData(unsigned int convtype) override;
void ReinterpretPixelData(unsigned int convtype) override;
bool SaveScreenshot(const std::string &filename, const TargetRectangle &rc) override;
bool SaveScreenshot(const std::string& filename, const TargetRectangle& rc) override;
int GetMaxTextureSize() override;
int GetMaxTextureSize() override;
private:
void UpdateEFBCache(EFBAccessType type, u32 cacheRectIdx, const EFBRectangle& efbPixelRc, const TargetRectangle& targetPixelRc, const void* data);
void UpdateEFBCache(EFBAccessType type, u32 cacheRectIdx, const EFBRectangle& efbPixelRc,
const TargetRectangle& targetPixelRc, const void* data);
void BlitScreen(TargetRectangle src, TargetRectangle dst, GLuint src_texture, int src_width, int src_height);
void BlitScreen(TargetRectangle src, TargetRectangle dst, GLuint src_texture, int src_width,
int src_height);
};
}

197
Source/Core/VideoBackends/OGL/SamplerCache.cpp
View File

@ -13,154 +13,143 @@
namespace OGL
{
std::unique_ptr<SamplerCache> g_sampler_cache;
SamplerCache::SamplerCache()
: m_last_max_anisotropy()
SamplerCache::SamplerCache() : m_last_max_anisotropy()
{
glGenSamplers(2, m_sampler_id);
glSamplerParameteri(m_sampler_id[0], GL_TEXTURE_MIN_FILTER, GL_NEAREST);
glSamplerParameteri(m_sampler_id[0], GL_TEXTURE_MAG_FILTER, GL_NEAREST);
glSamplerParameteri(m_sampler_id[0], GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
glSamplerParameteri(m_sampler_id[0], GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
glSamplerParameteri(m_sampler_id[1], GL_TEXTURE_MIN_FILTER, GL_LINEAR);
glSamplerParameteri(m_sampler_id[1], GL_TEXTURE_MAG_FILTER, GL_LINEAR);
glSamplerParameteri(m_sampler_id[1], GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
glSamplerParameteri(m_sampler_id[1], GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
glGenSamplers(2, m_sampler_id);
glSamplerParameteri(m_sampler_id[0], GL_TEXTURE_MIN_FILTER, GL_NEAREST);
glSamplerParameteri(m_sampler_id[0], GL_TEXTURE_MAG_FILTER, GL_NEAREST);
glSamplerParameteri(m_sampler_id[0], GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
glSamplerParameteri(m_sampler_id[0], GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
glSamplerParameteri(m_sampler_id[1], GL_TEXTURE_MIN_FILTER, GL_LINEAR);
glSamplerParameteri(m_sampler_id[1], GL_TEXTURE_MAG_FILTER, GL_LINEAR);
glSamplerParameteri(m_sampler_id[1], GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
glSamplerParameteri(m_sampler_id[1], GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
}
SamplerCache::~SamplerCache()
{
Clear();
glDeleteSamplers(2, m_sampler_id);
Clear();
glDeleteSamplers(2, m_sampler_id);
}
void SamplerCache::BindNearestSampler(int stage)
{
glBindSampler(stage, m_sampler_id[0]);
glBindSampler(stage, m_sampler_id[0]);
}
void SamplerCache::BindLinearSampler(int stage)
{
glBindSampler(stage, m_sampler_id[1]);
glBindSampler(stage, m_sampler_id[1]);
}
void SamplerCache::SetSamplerState(int stage, const TexMode0& tm0, const TexMode1& tm1, bool custom_tex)
void SamplerCache::SetSamplerState(int stage, const TexMode0& tm0, const TexMode1& tm1,
bool custom_tex)
{
// TODO: can this go somewhere else?
if (m_last_max_anisotropy != g_ActiveConfig.iMaxAnisotropy)
{
m_last_max_anisotropy = g_ActiveConfig.iMaxAnisotropy;
Clear();
}
// TODO: can this go somewhere else?
if (m_last_max_anisotropy != g_ActiveConfig.iMaxAnisotropy)
{
m_last_max_anisotropy = g_ActiveConfig.iMaxAnisotropy;
Clear();
}
Params params(tm0, tm1);
Params params(tm0, tm1);
// take equivalent forced linear when bForceFiltering
if (g_ActiveConfig.bForceFiltering)
{
params.tm0.min_filter = SamplerCommon::AreBpTexMode0MipmapsEnabled(tm0) ? 6 : 4;
params.tm0.mag_filter = 1;
}
// take equivalent forced linear when bForceFiltering
if (g_ActiveConfig.bForceFiltering)
{
params.tm0.min_filter = SamplerCommon::AreBpTexMode0MipmapsEnabled(tm0) ? 6 : 4;
params.tm0.mag_filter = 1;
}
// custom textures may have higher resolution, so disable the max_lod
if (custom_tex)
{
params.tm1.max_lod = 255;
}
// custom textures may have higher resolution, so disable the max_lod
if (custom_tex)
{
params.tm1.max_lod = 255;
}
// TODO: Should keep a circular buffer for each stage of recently used samplers.
// TODO: Should keep a circular buffer for each stage of recently used samplers.
auto& active_sampler = m_active_samplers[stage];
if (active_sampler.first != params || !active_sampler.second.sampler_id)
{
// Active sampler does not match parameters (or is invalid), bind the proper one.
active_sampler.first = params;
active_sampler.second = GetEntry(params);
glBindSampler(stage, active_sampler.second.sampler_id);
}
auto& active_sampler = m_active_samplers[stage];
if (active_sampler.first != params || !active_sampler.second.sampler_id)
{
// Active sampler does not match parameters (or is invalid), bind the proper one.
active_sampler.first = params;
active_sampler.second = GetEntry(params);
glBindSampler(stage, active_sampler.second.sampler_id);
}
}
SamplerCache::Value& SamplerCache::GetEntry(const Params& params)
{
auto& val = m_cache[params];
if (!val.sampler_id)
{
// Sampler not found in cache, create it.
glGenSamplers(1, &val.sampler_id);
SetParameters(val.sampler_id, params);
auto& val = m_cache[params];
if (!val.sampler_id)
{
// Sampler not found in cache, create it.
glGenSamplers(1, &val.sampler_id);
SetParameters(val.sampler_id, params);
// TODO: Maybe kill old samplers if the cache gets huge. It doesn't seem to get huge though.
//ERROR_LOG(VIDEO, "Sampler cache size is now %ld.", m_cache.size());
}
// TODO: Maybe kill old samplers if the cache gets huge. It doesn't seem to get huge though.
// ERROR_LOG(VIDEO, "Sampler cache size is now %ld.", m_cache.size());
}
return val;
return val;
}
void SamplerCache::SetParameters(GLuint sampler_id, const Params& params)
{
static const GLint min_filters[8] =
{
GL_NEAREST,
GL_NEAREST_MIPMAP_NEAREST,
GL_NEAREST_MIPMAP_LINEAR,
GL_NEAREST,
GL_LINEAR,
GL_LINEAR_MIPMAP_NEAREST,
GL_LINEAR_MIPMAP_LINEAR,
GL_LINEAR,
};
static const GLint min_filters[8] = {
GL_NEAREST, GL_NEAREST_MIPMAP_NEAREST, GL_NEAREST_MIPMAP_LINEAR, GL_NEAREST,
GL_LINEAR, GL_LINEAR_MIPMAP_NEAREST, GL_LINEAR_MIPMAP_LINEAR, GL_LINEAR,
};
static const GLint wrap_settings[4] =
{
GL_CLAMP_TO_EDGE,
GL_REPEAT,
GL_MIRRORED_REPEAT,
GL_REPEAT,
};
static const GLint wrap_settings[4] = {
GL_CLAMP_TO_EDGE, GL_REPEAT, GL_MIRRORED_REPEAT, GL_REPEAT,
};
auto& tm0 = params.tm0;
auto& tm1 = params.tm1;
auto& tm0 = params.tm0;
auto& tm1 = params.tm1;
glSamplerParameteri(sampler_id, GL_TEXTURE_WRAP_S, wrap_settings[tm0.wrap_s]);
glSamplerParameteri(sampler_id, GL_TEXTURE_WRAP_T, wrap_settings[tm0.wrap_t]);
glSamplerParameteri(sampler_id, GL_TEXTURE_WRAP_S, wrap_settings[tm0.wrap_s]);
glSamplerParameteri(sampler_id, GL_TEXTURE_WRAP_T, wrap_settings[tm0.wrap_t]);
glSamplerParameterf(sampler_id, GL_TEXTURE_MIN_LOD, tm1.min_lod / 16.f);
glSamplerParameterf(sampler_id, GL_TEXTURE_MAX_LOD, tm1.max_lod / 16.f);
glSamplerParameterf(sampler_id, GL_TEXTURE_MIN_LOD, tm1.min_lod / 16.f);
glSamplerParameterf(sampler_id, GL_TEXTURE_MAX_LOD, tm1.max_lod / 16.f);
if (GLInterface->GetMode() == GLInterfaceMode::MODE_OPENGL)
glSamplerParameterf(sampler_id, GL_TEXTURE_LOD_BIAS, (s32)tm0.lod_bias / 32.f);
if (GLInterface->GetMode() == GLInterfaceMode::MODE_OPENGL)
glSamplerParameterf(sampler_id, GL_TEXTURE_LOD_BIAS, (s32)tm0.lod_bias / 32.f);
GLint min_filter = min_filters[tm0.min_filter];
GLint mag_filter = tm0.mag_filter ? GL_LINEAR : GL_NEAREST;
GLint min_filter = min_filters[tm0.min_filter];
GLint mag_filter = tm0.mag_filter ? GL_LINEAR : GL_NEAREST;
if (g_ActiveConfig.iMaxAnisotropy > 0 && g_ogl_config.bSupportsAniso &&
!SamplerCommon::IsBpTexMode0PointFiltering(tm0))
{
// https://www.opengl.org/registry/specs/EXT/texture_filter_anisotropic.txt
// For predictable results on all hardware/drivers, only use one of:
// GL_LINEAR + GL_LINEAR (No Mipmaps [Bilinear])
// GL_LINEAR + GL_LINEAR_MIPMAP_LINEAR (w/ Mipmaps [Trilinear])
// Letting the game set other combinations will have varying arbitrary results;
// possibly being interpreted as equal to bilinear/trilinear, implicitly
// disabling anisotropy, or changing the anisotropic algorithm employed.
min_filter = SamplerCommon::AreBpTexMode0MipmapsEnabled(tm0) ? GL_LINEAR_MIPMAP_LINEAR : GL_LINEAR;
mag_filter = GL_LINEAR;
glSamplerParameterf(sampler_id, GL_TEXTURE_MAX_ANISOTROPY_EXT, (float)(1 << g_ActiveConfig.iMaxAnisotropy));
}
if (g_ActiveConfig.iMaxAnisotropy > 0 && g_ogl_config.bSupportsAniso &&
!SamplerCommon::IsBpTexMode0PointFiltering(tm0))
{
// https://www.opengl.org/registry/specs/EXT/texture_filter_anisotropic.txt
// For predictable results on all hardware/drivers, only use one of:
// GL_LINEAR + GL_LINEAR (No Mipmaps [Bilinear])
// GL_LINEAR + GL_LINEAR_MIPMAP_LINEAR (w/ Mipmaps [Trilinear])
// Letting the game set other combinations will have varying arbitrary results;
// possibly being interpreted as equal to bilinear/trilinear, implicitly
// disabling anisotropy, or changing the anisotropic algorithm employed.
min_filter =
SamplerCommon::AreBpTexMode0MipmapsEnabled(tm0) ? GL_LINEAR_MIPMAP_LINEAR : GL_LINEAR;
mag_filter = GL_LINEAR;
glSamplerParameterf(sampler_id, GL_TEXTURE_MAX_ANISOTROPY_EXT,
(float)(1 << g_ActiveConfig.iMaxAnisotropy));
}
glSamplerParameteri(sampler_id, GL_TEXTURE_MIN_FILTER, min_filter);
glSamplerParameteri(sampler_id, GL_TEXTURE_MAG_FILTER, mag_filter);
glSamplerParameteri(sampler_id, GL_TEXTURE_MIN_FILTER, min_filter);
glSamplerParameteri(sampler_id, GL_TEXTURE_MAG_FILTER, mag_filter);
}
void SamplerCache::Clear()
{
for (auto& p : m_cache)
{
glDeleteSamplers(1, &p.second.sampler_id);
}
m_cache.clear();
for (auto& p : m_cache)
{
glDeleteSamplers(1, &p.second.sampler_id);
}
m_cache.clear();
}
}

90
Source/Core/VideoBackends/OGL/SamplerCache.h
View File

@ -8,79 +8,61 @@
#include <memory>
#include "Common/CommonTypes.h"
#include "Common/NonCopyable.h"
#include "Common/GL/GLUtil.h"
#include "Common/NonCopyable.h"
#include "VideoBackends/OGL/Render.h"
namespace OGL
{
class SamplerCache : NonCopyable
{
public:
SamplerCache();
~SamplerCache();
SamplerCache();
~SamplerCache();
void SetSamplerState(int stage, const TexMode0& tm0, const TexMode1& tm1, bool custom_tex);
void Clear();
void BindNearestSampler(int stage);
void BindLinearSampler(int stage);
void SetSamplerState(int stage, const TexMode0& tm0, const TexMode1& tm1, bool custom_tex);
void Clear();
void BindNearestSampler(int stage);
void BindLinearSampler(int stage);
private:
struct Params
{
union
{
struct
{
TexMode0 tm0;
TexMode1 tm1;
};
struct Params
{
union {
struct
{
TexMode0 tm0;
TexMode1 tm1;
};
u64 hex;
};
u64 hex;
};
Params()
: hex()
{}
Params() : hex() {}
Params(const TexMode0& _tm0, const TexMode1& _tm1) : tm0(_tm0), tm1(_tm1)
{
static_assert(sizeof(Params) == 8, "Assuming I can treat this as a 64bit int.");
}
Params(const TexMode0& _tm0, const TexMode1& _tm1)
: tm0(_tm0)
, tm1(_tm1)
{
static_assert(sizeof(Params) == 8, "Assuming I can treat this as a 64bit int.");
}
bool operator<(const Params& other) const { return hex < other.hex; }
bool operator!=(const Params& other) const { return hex != other.hex; }
};
bool operator<(const Params& other) const
{
return hex < other.hex;
}
struct Value
{
Value() : sampler_id() {}
GLuint sampler_id;
};
bool operator!=(const Params& other) const
{
return hex != other.hex;
}
};
void SetParameters(GLuint sampler_id, const Params& params);
Value& GetEntry(const Params& params);
struct Value
{
Value()
: sampler_id()
{}
std::map<Params, Value> m_cache;
std::pair<Params, Value> m_active_samplers[8];
GLuint sampler_id;
};
void SetParameters(GLuint sampler_id, const Params& params);
Value& GetEntry(const Params& params);
std::map<Params, Value> m_cache;
std::pair<Params, Value> m_active_samplers[8];
int m_last_max_anisotropy;
u32 m_sampler_id[2];
int m_last_max_anisotropy;
u32 m_sampler_id[2];
};
extern std::unique_ptr<SamplerCache> g_sampler_cache;
}

460
Source/Core/VideoBackends/OGL/StreamBuffer.cpp
View File

@ -2,8 +2,8 @@
// Licensed under GPLv2+
// Refer to the license.txt file included.
#include "Common/MemoryUtil.h"
#include "Common/GL/GLUtil.h"
#include "Common/MemoryUtil.h"
#include "VideoBackends/OGL/Render.h"
#include "VideoBackends/OGL/StreamBuffer.h"
@ -13,27 +13,26 @@
namespace OGL
{
// moved out of constructor, so m_buffer is allowed to be const
static u32 GenBuffer()
{
u32 id;
glGenBuffers(1, &id);
return id;
u32 id;
glGenBuffers(1, &id);
return id;
}
StreamBuffer::StreamBuffer(u32 type, u32 size)
: m_buffer(GenBuffer()), m_buffertype(type), m_size(ROUND_UP_POW2(size)), m_bit_per_slot(IntLog2(ROUND_UP_POW2(size) / SYNC_POINTS))
: m_buffer(GenBuffer()), m_buffertype(type), m_size(ROUND_UP_POW2(size)),
m_bit_per_slot(IntLog2(ROUND_UP_POW2(size) / SYNC_POINTS))
{
m_iterator = 0;
m_used_iterator = 0;
m_free_iterator = 0;
m_iterator = 0;
m_used_iterator = 0;
m_free_iterator = 0;
}
StreamBuffer::~StreamBuffer()
{
glDeleteBuffers(1, &m_buffer);
glDeleteBuffers(1, &m_buffer);
}
/* Shared synchronization code for ring buffers
@ -44,76 +43,79 @@ StreamBuffer::~StreamBuffer()
* To reduce overhead, the complete buffer is splitted up into SYNC_POINTS chunks.
* For each of this chunks, there is a fence which checks if this chunk is still in use.
*
* As our API allows to alloc more memory then it has to use, we have to catch how much is already written.
* As our API allows to alloc more memory then it has to use, we have to catch how much is already
* written.
*
* m_iterator - writing position
* m_free_iterator - last position checked if free
* m_used_iterator - last position known to be written
*
* So on alloc, we have to wait for all slots between m_free_iterator and m_iterator (and set m_free_iterator to m_iterator afterwards).
* So on alloc, we have to wait for all slots between m_free_iterator and m_iterator (and set
* m_free_iterator to m_iterator afterwards).
*
* We also assume that this buffer is accessed by the GPU between the Unmap and Map function,
* so we may create the fences on the start of mapping.
* Some here, new fences for the chunks between m_used_iterator and m_iterator (also update m_used_iterator).
* Some here, new fences for the chunks between m_used_iterator and m_iterator (also update
* m_used_iterator).
*
* As ring buffers have an ugly behavior on rollover, have fun to read this code ;)
*/
void StreamBuffer::CreateFences()
{
for (int i = 0; i < SYNC_POINTS; i++)
{
m_fences[i] = glFenceSync(GL_SYNC_GPU_COMMANDS_COMPLETE, 0);
}
for (int i = 0; i < SYNC_POINTS; i++)
{
m_fences[i] = glFenceSync(GL_SYNC_GPU_COMMANDS_COMPLETE, 0);
}
}
void StreamBuffer::DeleteFences()
{
for (int i = Slot(m_free_iterator) + 1; i < SYNC_POINTS; i++)
{
glDeleteSync(m_fences[i]);
}
for (int i = 0; i < Slot(m_iterator); i++)
{
glDeleteSync(m_fences[i]);
}
for (int i = Slot(m_free_iterator) + 1; i < SYNC_POINTS; i++)
{
glDeleteSync(m_fences[i]);
}
for (int i = 0; i < Slot(m_iterator); i++)
{
glDeleteSync(m_fences[i]);
}
}
void StreamBuffer::AllocMemory(u32 size)
{
// insert waiting slots for used memory
for (int i = Slot(m_used_iterator); i < Slot(m_iterator); i++)
{
m_fences[i] = glFenceSync(GL_SYNC_GPU_COMMANDS_COMPLETE, 0);
}
m_used_iterator = m_iterator;
// insert waiting slots for used memory
for (int i = Slot(m_used_iterator); i < Slot(m_iterator); i++)
{
m_fences[i] = glFenceSync(GL_SYNC_GPU_COMMANDS_COMPLETE, 0);
}
m_used_iterator = m_iterator;
// wait for new slots to end of buffer
for (int i = Slot(m_free_iterator) + 1; i <= Slot(m_iterator + size) && i < SYNC_POINTS; i++)
{
glClientWaitSync(m_fences[i], GL_SYNC_FLUSH_COMMANDS_BIT, GL_TIMEOUT_IGNORED);
glDeleteSync(m_fences[i]);
}
m_free_iterator = m_iterator + size;
// wait for new slots to end of buffer
for (int i = Slot(m_free_iterator) + 1; i <= Slot(m_iterator + size) && i < SYNC_POINTS; i++)
{
glClientWaitSync(m_fences[i], GL_SYNC_FLUSH_COMMANDS_BIT, GL_TIMEOUT_IGNORED);
glDeleteSync(m_fences[i]);
}
m_free_iterator = m_iterator + size;
// if buffer is full
if (m_iterator + size >= m_size)
{
// insert waiting slots in unused space at the end of the buffer
for (int i = Slot(m_used_iterator); i < SYNC_POINTS; i++)
{
m_fences[i] = glFenceSync(GL_SYNC_GPU_COMMANDS_COMPLETE, 0);
}
// if buffer is full
if (m_iterator + size >= m_size)
{
// insert waiting slots in unused space at the end of the buffer
for (int i = Slot(m_used_iterator); i < SYNC_POINTS; i++)
{
m_fences[i] = glFenceSync(GL_SYNC_GPU_COMMANDS_COMPLETE, 0);
}
// move to the start
m_used_iterator = m_iterator = 0; // offset 0 is always aligned
// move to the start
m_used_iterator = m_iterator = 0; // offset 0 is always aligned
// wait for space at the start
for (int i = 0; i <= Slot(m_iterator + size); i++)
{
glClientWaitSync(m_fences[i], GL_SYNC_FLUSH_COMMANDS_BIT, GL_TIMEOUT_IGNORED);
glDeleteSync(m_fences[i]);
}
m_free_iterator = m_iterator + size;
}
// wait for space at the start
for (int i = 0; i <= Slot(m_iterator + size); i++)
{
glClientWaitSync(m_fences[i], GL_SYNC_FLUSH_COMMANDS_BIT, GL_TIMEOUT_IGNORED);
glDeleteSync(m_fences[i]);
}
m_free_iterator = m_iterator + size;
}
}
/* The usual way to stream data to the GPU.
@ -126,34 +128,32 @@ void StreamBuffer::AllocMemory(u32 size)
class MapAndOrphan : public StreamBuffer
{
public:
MapAndOrphan(u32 type, u32 size) : StreamBuffer(type, size)
{
glBindBuffer(m_buffertype, m_buffer);
glBufferData(m_buffertype, m_size, nullptr, GL_STREAM_DRAW);
}
MapAndOrphan(u32 type, u32 size) : StreamBuffer(type, size)
{
glBindBuffer(m_buffertype, m_buffer);
glBufferData(m_buffertype, m_size, nullptr, GL_STREAM_DRAW);
}
~MapAndOrphan()
{
}
~MapAndOrphan() {}
std::pair<u8*, u32> Map(u32 size) override
{
if (m_iterator + size >= m_size)
{
glBufferData(m_buffertype, m_size, nullptr, GL_STREAM_DRAW);
m_iterator = 0;
}
u8* pointer = (u8*)glMapBufferRange(m_buffertype, m_iterator, size,
GL_MAP_WRITE_BIT | GL_MAP_FLUSH_EXPLICIT_BIT |
GL_MAP_UNSYNCHRONIZED_BIT);
return std::make_pair(pointer, m_iterator);
}
std::pair<u8*, u32> Map(u32 size) override
{
if (m_iterator + size >= m_size)
{
glBufferData(m_buffertype, m_size, nullptr, GL_STREAM_DRAW);
m_iterator = 0;
}
u8* pointer = (u8*)glMapBufferRange(m_buffertype, m_iterator, size,
GL_MAP_WRITE_BIT | GL_MAP_FLUSH_EXPLICIT_BIT | GL_MAP_UNSYNCHRONIZED_BIT);
return std::make_pair(pointer, m_iterator);
}
void Unmap(u32 used_size) override
{
glFlushMappedBufferRange(m_buffertype, 0, used_size);
glUnmapBuffer(m_buffertype);
m_iterator += used_size;
}
void Unmap(u32 used_size) override
{
glFlushMappedBufferRange(m_buffertype, 0, used_size);
glUnmapBuffer(m_buffertype);
m_iterator += used_size;
}
};
/* A modified streaming way without reallocation
@ -166,32 +166,29 @@ public:
class MapAndSync : public StreamBuffer
{
public:
MapAndSync(u32 type, u32 size) : StreamBuffer(type, size)
{
CreateFences();
glBindBuffer(m_buffertype, m_buffer);
glBufferData(m_buffertype, m_size, nullptr, GL_STREAM_DRAW);
}
MapAndSync(u32 type, u32 size) : StreamBuffer(type, size)
{
CreateFences();
glBindBuffer(m_buffertype, m_buffer);
glBufferData(m_buffertype, m_size, nullptr, GL_STREAM_DRAW);
}
~MapAndSync()
{
DeleteFences();
}
~MapAndSync() { DeleteFences(); }
std::pair<u8*, u32> Map(u32 size) override
{
AllocMemory(size);
u8* pointer = (u8*)glMapBufferRange(m_buffertype, m_iterator, size,
GL_MAP_WRITE_BIT | GL_MAP_FLUSH_EXPLICIT_BIT |
GL_MAP_UNSYNCHRONIZED_BIT);
return std::make_pair(pointer, m_iterator);
}
std::pair<u8*, u32> Map(u32 size) override
{
AllocMemory(size);
u8* pointer = (u8*)glMapBufferRange(m_buffertype, m_iterator, size,
GL_MAP_WRITE_BIT | GL_MAP_FLUSH_EXPLICIT_BIT | GL_MAP_UNSYNCHRONIZED_BIT);
return std::make_pair(pointer, m_iterator);
}
void Unmap(u32 used_size) override
{
glFlushMappedBufferRange(m_buffertype, 0, used_size);
glUnmapBuffer(m_buffertype);
m_iterator += used_size;
}
void Unmap(u32 used_size) override
{
glFlushMappedBufferRange(m_buffertype, 0, used_size);
glUnmapBuffer(m_buffertype);
m_iterator += used_size;
}
};
/* Streaming fifo without mapping overhead.
@ -210,42 +207,45 @@ public:
class BufferStorage : public StreamBuffer
{
public:
BufferStorage(u32 type, u32 size, bool _coherent = false) : StreamBuffer(type, size), coherent(_coherent)
{
CreateFences();
glBindBuffer(m_buffertype, m_buffer);
BufferStorage(u32 type, u32 size, bool _coherent = false)
: StreamBuffer(type, size), coherent(_coherent)
{
CreateFences();
glBindBuffer(m_buffertype, m_buffer);
// PERSISTANT_BIT to make sure that the buffer can be used while mapped
// COHERENT_BIT is set so we don't have to use a MemoryBarrier on write
// CLIENT_STORAGE_BIT is set since we access the buffer more frequently on the client side then server side
glBufferStorage(m_buffertype, m_size, nullptr,
GL_MAP_WRITE_BIT | GL_MAP_PERSISTENT_BIT | (coherent ? GL_MAP_COHERENT_BIT : 0));
m_pointer = (u8*)glMapBufferRange(m_buffertype, 0, m_size,
GL_MAP_WRITE_BIT | GL_MAP_PERSISTENT_BIT | (coherent ? GL_MAP_COHERENT_BIT : GL_MAP_FLUSH_EXPLICIT_BIT));
}
// PERSISTANT_BIT to make sure that the buffer can be used while mapped
// COHERENT_BIT is set so we don't have to use a MemoryBarrier on write
// CLIENT_STORAGE_BIT is set since we access the buffer more frequently on the client side then
// server side
glBufferStorage(m_buffertype, m_size, nullptr, GL_MAP_WRITE_BIT | GL_MAP_PERSISTENT_BIT |
(coherent ? GL_MAP_COHERENT_BIT : 0));
m_pointer = (u8*)glMapBufferRange(
m_buffertype, 0, m_size, GL_MAP_WRITE_BIT | GL_MAP_PERSISTENT_BIT |
(coherent ? GL_MAP_COHERENT_BIT : GL_MAP_FLUSH_EXPLICIT_BIT));
}
~BufferStorage()
{
DeleteFences();
glUnmapBuffer(m_buffertype);
glBindBuffer(m_buffertype, 0);
}
~BufferStorage()
{
DeleteFences();
glUnmapBuffer(m_buffertype);
glBindBuffer(m_buffertype, 0);
}
std::pair<u8*, u32> Map(u32 size) override
{
AllocMemory(size);
return std::make_pair(m_pointer + m_iterator, m_iterator);
}
std::pair<u8*, u32> Map(u32 size) override
{
AllocMemory(size);
return std::make_pair(m_pointer + m_iterator, m_iterator);
}
void Unmap(u32 used_size) override
{
if (!coherent)
glFlushMappedBufferRange(m_buffertype, m_iterator, used_size);
m_iterator += used_size;
}
void Unmap(u32 used_size) override
{
if (!coherent)
glFlushMappedBufferRange(m_buffertype, m_iterator, used_size);
m_iterator += used_size;
}
u8* m_pointer;
const bool coherent;
u8* m_pointer;
const bool coherent;
};
/* --- AMD only ---
@ -258,38 +258,36 @@ public:
class PinnedMemory : public StreamBuffer
{
public:
PinnedMemory(u32 type, u32 size) : StreamBuffer(type, size)
{
CreateFences();
m_pointer = (u8*)AllocateAlignedMemory(ROUND_UP(m_size,ALIGN_PINNED_MEMORY), ALIGN_PINNED_MEMORY );
glBindBuffer(GL_EXTERNAL_VIRTUAL_MEMORY_BUFFER_AMD, m_buffer);
glBufferData(GL_EXTERNAL_VIRTUAL_MEMORY_BUFFER_AMD, ROUND_UP(m_size,ALIGN_PINNED_MEMORY), m_pointer, GL_STREAM_COPY);
glBindBuffer(GL_EXTERNAL_VIRTUAL_MEMORY_BUFFER_AMD, 0);
glBindBuffer(m_buffertype, m_buffer);
}
PinnedMemory(u32 type, u32 size) : StreamBuffer(type, size)
{
CreateFences();
m_pointer =
(u8*)AllocateAlignedMemory(ROUND_UP(m_size, ALIGN_PINNED_MEMORY), ALIGN_PINNED_MEMORY);
glBindBuffer(GL_EXTERNAL_VIRTUAL_MEMORY_BUFFER_AMD, m_buffer);
glBufferData(GL_EXTERNAL_VIRTUAL_MEMORY_BUFFER_AMD, ROUND_UP(m_size, ALIGN_PINNED_MEMORY),
m_pointer, GL_STREAM_COPY);
glBindBuffer(GL_EXTERNAL_VIRTUAL_MEMORY_BUFFER_AMD, 0);
glBindBuffer(m_buffertype, m_buffer);
}
~PinnedMemory()
{
DeleteFences();
glBindBuffer(m_buffertype, 0);
glFinish(); // ogl pipeline must be flushed, else this buffer can be in use
FreeAlignedMemory(m_pointer);
m_pointer = nullptr;
}
~PinnedMemory()
{
DeleteFences();
glBindBuffer(m_buffertype, 0);
glFinish(); // ogl pipeline must be flushed, else this buffer can be in use
FreeAlignedMemory(m_pointer);
m_pointer = nullptr;
}
std::pair<u8*, u32> Map(u32 size) override
{
AllocMemory(size);
return std::make_pair(m_pointer + m_iterator, m_iterator);
}
std::pair<u8*, u32> Map(u32 size) override
{
AllocMemory(size);
return std::make_pair(m_pointer + m_iterator, m_iterator);
}
void Unmap(u32 used_size) override
{
m_iterator += used_size;
}
u8* m_pointer;
static const u32 ALIGN_PINNED_MEMORY = 4096;
void Unmap(u32 used_size) override { m_iterator += used_size; }
u8* m_pointer;
static const u32 ALIGN_PINNED_MEMORY = 4096;
};
/* Fifo based on the glBufferSubData call.
@ -300,29 +298,17 @@ public:
class BufferSubData : public StreamBuffer
{
public:
BufferSubData(u32 type, u32 size) : StreamBuffer(type, size)
{
glBindBuffer(m_buffertype, m_buffer);
glBufferData(m_buffertype, size, nullptr, GL_STATIC_DRAW);
m_pointer = new u8[m_size];
}
BufferSubData(u32 type, u32 size) : StreamBuffer(type, size)
{
glBindBuffer(m_buffertype, m_buffer);
glBufferData(m_buffertype, size, nullptr, GL_STATIC_DRAW);
m_pointer = new u8[m_size];
}
~BufferSubData()
{
delete [] m_pointer;
}
std::pair<u8*, u32> Map(u32 size) override
{
return std::make_pair(m_pointer, 0);
}
void Unmap(u32 used_size) override
{
glBufferSubData(m_buffertype, 0, used_size, m_pointer);
}
u8* m_pointer;
~BufferSubData() { delete[] m_pointer; }
std::pair<u8*, u32> Map(u32 size) override { return std::make_pair(m_pointer, 0); }
void Unmap(u32 used_size) override { glBufferSubData(m_buffertype, 0, used_size, m_pointer); }
u8* m_pointer;
};
/* Fifo based on the glBufferData call.
@ -333,69 +319,63 @@ public:
class BufferData : public StreamBuffer
{
public:
BufferData(u32 type, u32 size) : StreamBuffer(type, size)
{
glBindBuffer(m_buffertype, m_buffer);
m_pointer = new u8[m_size];
}
BufferData(u32 type, u32 size) : StreamBuffer(type, size)
{
glBindBuffer(m_buffertype, m_buffer);
m_pointer = new u8[m_size];
}
~BufferData()
{
delete [] m_pointer;
}
~BufferData() { delete[] m_pointer; }
std::pair<u8*, u32> Map(u32 size) override { return std::make_pair(m_pointer, 0); }
void Unmap(u32 used_size) override
{
glBufferData(m_buffertype, used_size, m_pointer, GL_STREAM_DRAW);
}
std::pair<u8*, u32> Map(u32 size) override
{
return std::make_pair(m_pointer, 0);
}
void Unmap(u32 used_size) override
{
glBufferData(m_buffertype, used_size, m_pointer, GL_STREAM_DRAW);
}
u8* m_pointer;
u8* m_pointer;
};
// Chooses the best streaming method based on the supported extensions and known issues
std::unique_ptr<StreamBuffer> StreamBuffer::Create(u32 type, u32 size)
{
// without basevertex support, only streaming methods whith uploads everything to zero works fine:
if (!g_ogl_config.bSupportsGLBaseVertex)
{
if (!DriverDetails::HasBug(DriverDetails::BUG_BROKENBUFFERSTREAM))
return std::make_unique<BufferSubData>(type, size);
// without basevertex support, only streaming methods whith uploads everything to zero works fine:
if (!g_ogl_config.bSupportsGLBaseVertex)
{
if (!DriverDetails::HasBug(DriverDetails::BUG_BROKENBUFFERSTREAM))
return std::make_unique<BufferSubData>(type, size);
// BufferData is by far the worst way, only use it if needed
return std::make_unique<BufferData>(type, size);
}
// BufferData is by far the worst way, only use it if needed
return std::make_unique<BufferData>(type, size);
}
// Prefer the syncing buffers over the orphaning one
if (g_ogl_config.bSupportsGLSync)
{
// pinned memory is much faster than buffer storage on AMD cards
if (g_ogl_config.bSupportsGLPinnedMemory &&
!(DriverDetails::HasBug(DriverDetails::BUG_BROKENPINNEDMEMORY) && type == GL_ELEMENT_ARRAY_BUFFER))
return std::make_unique<PinnedMemory>(type, size);
// Prefer the syncing buffers over the orphaning one
if (g_ogl_config.bSupportsGLSync)
{
// pinned memory is much faster than buffer storage on AMD cards
if (g_ogl_config.bSupportsGLPinnedMemory &&
!(DriverDetails::HasBug(DriverDetails::BUG_BROKENPINNEDMEMORY) &&
type == GL_ELEMENT_ARRAY_BUFFER))
return std::make_unique<PinnedMemory>(type, size);
// buffer storage works well in most situations
bool coherent = DriverDetails::HasBug(DriverDetails::BUG_BROKENEXPLICITFLUSH);
if (g_ogl_config.bSupportsGLBufferStorage &&
!(DriverDetails::HasBug(DriverDetails::BUG_BROKENBUFFERSTORAGE) && type == GL_ARRAY_BUFFER) &&
!(DriverDetails::HasBug(DriverDetails::BUG_INTELBROKENBUFFERSTORAGE) && type == GL_ELEMENT_ARRAY_BUFFER))
return std::make_unique<BufferStorage>(type, size, coherent);
// buffer storage works well in most situations
bool coherent = DriverDetails::HasBug(DriverDetails::BUG_BROKENEXPLICITFLUSH);
if (g_ogl_config.bSupportsGLBufferStorage &&
!(DriverDetails::HasBug(DriverDetails::BUG_BROKENBUFFERSTORAGE) &&
type == GL_ARRAY_BUFFER) &&
!(DriverDetails::HasBug(DriverDetails::BUG_INTELBROKENBUFFERSTORAGE) &&
type == GL_ELEMENT_ARRAY_BUFFER))
return std::make_unique<BufferStorage>(type, size, coherent);
// don't fall back to MapAnd* for Nvidia drivers
if (DriverDetails::HasBug(DriverDetails::BUG_BROKENUNSYNCMAPPING))
return std::make_unique<BufferSubData>(type, size);
// don't fall back to MapAnd* for Nvidia drivers
if (DriverDetails::HasBug(DriverDetails::BUG_BROKENUNSYNCMAPPING))
return std::make_unique<BufferSubData>(type, size);
// mapping fallback
if (g_ogl_config.bSupportsGLSync)
return std::make_unique<MapAndSync>(type, size);
}
// mapping fallback
if (g_ogl_config.bSupportsGLSync)
return std::make_unique<MapAndSync>(type, size);
}
// default fallback, should work everywhere, but isn't the best way to do this job
return std::make_unique<MapAndOrphan>(type, size);
// default fallback, should work everywhere, but isn't the best way to do this job
return std::make_unique<MapAndOrphan>(type, size);
}
}

73
Source/Core/VideoBackends/OGL/StreamBuffer.h
View File

@ -16,56 +16,53 @@
namespace OGL
{
class StreamBuffer
{
public:
static std::unique_ptr<StreamBuffer> Create(u32 type, u32 size);
virtual ~StreamBuffer();
static std::unique_ptr<StreamBuffer> Create(u32 type, u32 size);
virtual ~StreamBuffer();
/* This mapping function will return a pair of:
* - the pointer to the mapped buffer
* - the offset into the real GPU buffer (always multiple of stride)
* On mapping, the maximum of size for allocation has to be set.
* The size really pushed into this fifo only has to be known on Unmapping.
* Mapping invalidates the current buffer content,
* so it isn't allowed to access the old content any more.
*/
virtual std::pair<u8*, u32> Map(u32 size) = 0;
virtual void Unmap(u32 used_size) = 0;
/* This mapping function will return a pair of:
* - the pointer to the mapped buffer
* - the offset into the real GPU buffer (always multiple of stride)
* On mapping, the maximum of size for allocation has to be set.
* The size really pushed into this fifo only has to be known on Unmapping.
* Mapping invalidates the current buffer content,
* so it isn't allowed to access the old content any more.
*/
virtual std::pair<u8*, u32> Map(u32 size) = 0;
virtual void Unmap(u32 used_size) = 0;
std::pair<u8*, u32> Map(u32 size, u32 stride)
{
u32 padding = m_iterator % stride;
if (padding)
{
m_iterator += stride - padding;
}
return Map(size);
}
std::pair<u8*, u32> Map(u32 size, u32 stride)
{
u32 padding = m_iterator % stride;
if (padding)
{
m_iterator += stride - padding;
}
return Map(size);
}
const u32 m_buffer;
const u32 m_buffer;
protected:
StreamBuffer(u32 type, u32 size);
void CreateFences();
void DeleteFences();
void AllocMemory(u32 size);
StreamBuffer(u32 type, u32 size);
void CreateFences();
void DeleteFences();
void AllocMemory(u32 size);
const u32 m_buffertype;
const u32 m_size;
const u32 m_buffertype;
const u32 m_size;
u32 m_iterator;
u32 m_used_iterator;
u32 m_free_iterator;
u32 m_iterator;
u32 m_used_iterator;
u32 m_free_iterator;
private:
static constexpr int SYNC_POINTS = 16;
int Slot(u32 x) const { return x >> m_bit_per_slot; }
const int m_bit_per_slot;
static constexpr int SYNC_POINTS = 16;
int Slot(u32 x) const { return x >> m_bit_per_slot; }
const int m_bit_per_slot;
std::array<GLsync, SYNC_POINTS> m_fences{};
std::array<GLsync, SYNC_POINTS> m_fences{};
};
}

678
Source/Core/VideoBackends/OGL/TextureCache.cpp
View File

@ -36,7 +36,6 @@
namespace OGL
{
static SHADER s_ColorCopyProgram;
static SHADER s_ColorMatrixProgram;
static SHADER s_DepthMatrixProgram;
@ -58,270 +57,250 @@ static GLuint s_palette_buffer_offset_uniform[3];
static GLuint s_palette_multiplier_uniform[3];
static GLuint s_palette_copy_position_uniform[3];
bool SaveTexture(const std::string& filename, u32 textarget, u32 tex, int virtual_width, int virtual_height, unsigned int level)
bool SaveTexture(const std::string& filename, u32 textarget, u32 tex, int virtual_width,
int virtual_height, unsigned int level)
{
if (GLInterface->GetMode() != GLInterfaceMode::MODE_OPENGL)
return false;
int width = std::max(virtual_width >> level, 1);
int height = std::max(virtual_height >> level, 1);
std::vector<u8> data(width * height * 4);
glActiveTexture(GL_TEXTURE9);
glBindTexture(textarget, tex);
glGetTexImage(textarget, level, GL_RGBA, GL_UNSIGNED_BYTE, data.data());
TextureCache::SetStage();
if (GLInterface->GetMode() != GLInterfaceMode::MODE_OPENGL)
return false;
int width = std::max(virtual_width >> level, 1);
int height = std::max(virtual_height >> level, 1);
std::vector<u8> data(width * height * 4);
glActiveTexture(GL_TEXTURE9);
glBindTexture(textarget, tex);
glGetTexImage(textarget, level, GL_RGBA, GL_UNSIGNED_BYTE, data.data());
TextureCache::SetStage();
return TextureToPng(data.data(), width * 4, filename, width, height, true);
return TextureToPng(data.data(), width * 4, filename, width, height, true);
}
TextureCache::TCacheEntry::~TCacheEntry()
{
if (texture)
{
for (auto& gtex : s_Textures)
if (gtex == texture)
gtex = 0;
glDeleteTextures(1, &texture);
texture = 0;
}
if (texture)
{
for (auto& gtex : s_Textures)
if (gtex == texture)
gtex = 0;
glDeleteTextures(1, &texture);
texture = 0;
}
if (framebuffer)
{
glDeleteFramebuffers(1, &framebuffer);
framebuffer = 0;
}
if (framebuffer)
{
glDeleteFramebuffers(1, &framebuffer);
framebuffer = 0;
}
}
TextureCache::TCacheEntry::TCacheEntry(const TCacheEntryConfig& _config)
: TCacheEntryBase(_config)
TextureCache::TCacheEntry::TCacheEntry(const TCacheEntryConfig& _config) : TCacheEntryBase(_config)
{
glGenTextures(1, &texture);
glGenTextures(1, &texture);
framebuffer = 0;
framebuffer = 0;
}
void TextureCache::TCacheEntry::Bind(unsigned int stage)
{
if (s_Textures[stage] != texture)
{
if (s_ActiveTexture != stage)
{
glActiveTexture(GL_TEXTURE0 + stage);
s_ActiveTexture = stage;
}
if (s_Textures[stage] != texture)
{
if (s_ActiveTexture != stage)
{
glActiveTexture(GL_TEXTURE0 + stage);
s_ActiveTexture = stage;
}
glBindTexture(GL_TEXTURE_2D_ARRAY, texture);
s_Textures[stage] = texture;
}
glBindTexture(GL_TEXTURE_2D_ARRAY, texture);
s_Textures[stage] = texture;
}
}
bool TextureCache::TCacheEntry::Save(const std::string& filename, unsigned int level)
{
return SaveTexture(filename, GL_TEXTURE_2D_ARRAY, texture, config.width, config.height, level);
return SaveTexture(filename, GL_TEXTURE_2D_ARRAY, texture, config.width, config.height, level);
}
TextureCache::TCacheEntryBase* TextureCache::CreateTexture(const TCacheEntryConfig& config)
{
TCacheEntry* entry = new TCacheEntry(config);
TCacheEntry* entry = new TCacheEntry(config);
glActiveTexture(GL_TEXTURE9);
glBindTexture(GL_TEXTURE_2D_ARRAY, entry->texture);
glActiveTexture(GL_TEXTURE9);
glBindTexture(GL_TEXTURE_2D_ARRAY, entry->texture);
glTexParameteri(GL_TEXTURE_2D_ARRAY, GL_TEXTURE_MAX_LEVEL, config.levels - 1);
glTexParameteri(GL_TEXTURE_2D_ARRAY, GL_TEXTURE_MAX_LEVEL, config.levels - 1);
if (config.rendertarget)
{
for (u32 level = 0; level <= config.levels; level++)
{
glTexImage3D(GL_TEXTURE_2D_ARRAY, level, GL_RGBA, config.width, config.height, config.layers, 0, GL_RGBA, GL_UNSIGNED_BYTE, nullptr);
}
glGenFramebuffers(1, &entry->framebuffer);
FramebufferManager::SetFramebuffer(entry->framebuffer);
FramebufferManager::FramebufferTexture(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D_ARRAY, entry->texture, 0);
}
if (config.rendertarget)
{
for (u32 level = 0; level <= config.levels; level++)
{
glTexImage3D(GL_TEXTURE_2D_ARRAY, level, GL_RGBA, config.width, config.height, config.layers,
0, GL_RGBA, GL_UNSIGNED_BYTE, nullptr);
}
glGenFramebuffers(1, &entry->framebuffer);
FramebufferManager::SetFramebuffer(entry->framebuffer);
FramebufferManager::FramebufferTexture(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0,
GL_TEXTURE_2D_ARRAY, entry->texture, 0);
}
TextureCache::SetStage();
return entry;
TextureCache::SetStage();
return entry;
}
void TextureCache::TCacheEntry::CopyRectangleFromTexture(
const TCacheEntryBase* source,
const MathUtil::Rectangle<int> &srcrect,
const MathUtil::Rectangle<int> &dstrect)
void TextureCache::TCacheEntry::CopyRectangleFromTexture(const TCacheEntryBase* source,
const MathUtil::Rectangle<int>& srcrect,
const MathUtil::Rectangle<int>& dstrect)
{
TCacheEntry* srcentry = (TCacheEntry*)source;
if (srcrect.GetWidth() == dstrect.GetWidth()
&& srcrect.GetHeight() == dstrect.GetHeight()
&& g_ogl_config.bSupportsCopySubImage)
{
glCopyImageSubData(
srcentry->texture,
GL_TEXTURE_2D_ARRAY,
0,
srcrect.left,
srcrect.top,
0,
texture,
GL_TEXTURE_2D_ARRAY,
0,
dstrect.left,
dstrect.top,
0,
dstrect.GetWidth(),
dstrect.GetHeight(),
srcentry->config.layers);
return;
}
else if (!framebuffer)
{
glGenFramebuffers(1, &framebuffer);
FramebufferManager::SetFramebuffer(framebuffer);
FramebufferManager::FramebufferTexture(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D_ARRAY, texture, 0);
}
g_renderer->ResetAPIState();
FramebufferManager::SetFramebuffer(framebuffer);
glActiveTexture(GL_TEXTURE9);
glBindTexture(GL_TEXTURE_2D_ARRAY, srcentry->texture);
g_sampler_cache->BindLinearSampler(9);
glViewport(dstrect.left, dstrect.top, dstrect.GetWidth(), dstrect.GetHeight());
s_ColorCopyProgram.Bind();
glUniform4f(s_ColorCopyPositionUniform,
float(srcrect.left),
float(srcrect.top),
float(srcrect.GetWidth()),
float(srcrect.GetHeight()));
glDrawArrays(GL_TRIANGLE_STRIP, 0, 4);
FramebufferManager::SetFramebuffer(0);
g_renderer->RestoreAPIState();
TCacheEntry* srcentry = (TCacheEntry*)source;
if (srcrect.GetWidth() == dstrect.GetWidth() && srcrect.GetHeight() == dstrect.GetHeight() &&
g_ogl_config.bSupportsCopySubImage)
{
glCopyImageSubData(srcentry->texture, GL_TEXTURE_2D_ARRAY, 0, srcrect.left, srcrect.top, 0,
texture, GL_TEXTURE_2D_ARRAY, 0, dstrect.left, dstrect.top, 0,
dstrect.GetWidth(), dstrect.GetHeight(), srcentry->config.layers);
return;
}
else if (!framebuffer)
{
glGenFramebuffers(1, &framebuffer);
FramebufferManager::SetFramebuffer(framebuffer);
FramebufferManager::FramebufferTexture(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0,
GL_TEXTURE_2D_ARRAY, texture, 0);
}
g_renderer->ResetAPIState();
FramebufferManager::SetFramebuffer(framebuffer);
glActiveTexture(GL_TEXTURE9);
glBindTexture(GL_TEXTURE_2D_ARRAY, srcentry->texture);
g_sampler_cache->BindLinearSampler(9);
glViewport(dstrect.left, dstrect.top, dstrect.GetWidth(), dstrect.GetHeight());
s_ColorCopyProgram.Bind();
glUniform4f(s_ColorCopyPositionUniform, float(srcrect.left), float(srcrect.top),
float(srcrect.GetWidth()), float(srcrect.GetHeight()));
glDrawArrays(GL_TRIANGLE_STRIP, 0, 4);
FramebufferManager::SetFramebuffer(0);
g_renderer->RestoreAPIState();
}
void TextureCache::TCacheEntry::Load(unsigned int width, unsigned int height,
unsigned int expanded_width, unsigned int level)
unsigned int expanded_width, unsigned int level)
{
if (level >= config.levels)
PanicAlert("Texture only has %d levels, can't update level %d", config.levels, level);
if (width != std::max(1u, config.width >> level) || height != std::max(1u, config.height >> level))
PanicAlert("size of level %d must be %dx%d, but %dx%d requested",
level, std::max(1u, config.width >> level), std::max(1u, config.height >> level), width, height);
if (level >= config.levels)
PanicAlert("Texture only has %d levels, can't update level %d", config.levels, level);
if (width != std::max(1u, config.width >> level) ||
height != std::max(1u, config.height >> level))
PanicAlert("size of level %d must be %dx%d, but %dx%d requested", level,
std::max(1u, config.width >> level), std::max(1u, config.height >> level), width,
height);
glActiveTexture(GL_TEXTURE9);
glBindTexture(GL_TEXTURE_2D_ARRAY, texture);
glActiveTexture(GL_TEXTURE9);
glBindTexture(GL_TEXTURE_2D_ARRAY, texture);
if (expanded_width != width)
glPixelStorei(GL_UNPACK_ROW_LENGTH, expanded_width);
if (expanded_width != width)
glPixelStorei(GL_UNPACK_ROW_LENGTH, expanded_width);
glTexImage3D(GL_TEXTURE_2D_ARRAY, level, GL_RGBA, width, height, 1, 0, GL_RGBA, GL_UNSIGNED_BYTE, temp);
glTexImage3D(GL_TEXTURE_2D_ARRAY, level, GL_RGBA, width, height, 1, 0, GL_RGBA, GL_UNSIGNED_BYTE,
temp);
if (expanded_width != width)
glPixelStorei(GL_UNPACK_ROW_LENGTH, 0);
if (expanded_width != width)
glPixelStorei(GL_UNPACK_ROW_LENGTH, 0);
TextureCache::SetStage();
TextureCache::SetStage();
}
void TextureCache::TCacheEntry::FromRenderTarget(u8* dstPointer, PEControl::PixelFormat srcFormat, const EFBRectangle& srcRect,
bool scaleByHalf, unsigned int cbufid, const float *colmat)
void TextureCache::TCacheEntry::FromRenderTarget(u8* dstPointer, PEControl::PixelFormat srcFormat,
const EFBRectangle& srcRect, bool scaleByHalf,
unsigned int cbufid, const float* colmat)
{
g_renderer->ResetAPIState(); // reset any game specific settings
g_renderer->ResetAPIState(); // reset any game specific settings
// Make sure to resolve anything we need to read from.
const GLuint read_texture = (srcFormat == PEControl::Z24) ?
FramebufferManager::ResolveAndGetDepthTarget(srcRect) :
FramebufferManager::ResolveAndGetRenderTarget(srcRect);
// Make sure to resolve anything we need to read from.
const GLuint read_texture = (srcFormat == PEControl::Z24) ?
FramebufferManager::ResolveAndGetDepthTarget(srcRect) :
FramebufferManager::ResolveAndGetRenderTarget(srcRect);
FramebufferManager::SetFramebuffer(framebuffer);
FramebufferManager::SetFramebuffer(framebuffer);
OpenGL_BindAttributelessVAO();
OpenGL_BindAttributelessVAO();
glActiveTexture(GL_TEXTURE9);
glBindTexture(GL_TEXTURE_2D_ARRAY, read_texture);
if (scaleByHalf)
g_sampler_cache->BindLinearSampler(9);
else
g_sampler_cache->BindNearestSampler(9);
glActiveTexture(GL_TEXTURE9);
glBindTexture(GL_TEXTURE_2D_ARRAY, read_texture);
if (scaleByHalf)
g_sampler_cache->BindLinearSampler(9);
else
g_sampler_cache->BindNearestSampler(9);
glViewport(0, 0, config.width, config.height);
glViewport(0, 0, config.width, config.height);
GLuint uniform_location;
if (srcFormat == PEControl::Z24)
{
s_DepthMatrixProgram.Bind();
if (s_DepthCbufid != cbufid)
glUniform4fv(s_DepthMatrixUniform, 5, colmat);
s_DepthCbufid = cbufid;
uniform_location = s_DepthCopyPositionUniform;
}
else
{
s_ColorMatrixProgram.Bind();
if (s_ColorCbufid != cbufid)
glUniform4fv(s_ColorMatrixUniform, 7, colmat);
s_ColorCbufid = cbufid;
uniform_location = s_ColorMatrixPositionUniform;
}
GLuint uniform_location;
if (srcFormat == PEControl::Z24)
{
s_DepthMatrixProgram.Bind();
if (s_DepthCbufid != cbufid)
glUniform4fv(s_DepthMatrixUniform, 5, colmat);
s_DepthCbufid = cbufid;
uniform_location = s_DepthCopyPositionUniform;
}
else
{
s_ColorMatrixProgram.Bind();
if (s_ColorCbufid != cbufid)
glUniform4fv(s_ColorMatrixUniform, 7, colmat);
s_ColorCbufid = cbufid;
uniform_location = s_ColorMatrixPositionUniform;
}
TargetRectangle R = g_renderer->ConvertEFBRectangle(srcRect);
glUniform4f(uniform_location, static_cast<float>(R.left), static_cast<float>(R.top),
static_cast<float>(R.right), static_cast<float>(R.bottom));
TargetRectangle R = g_renderer->ConvertEFBRectangle(srcRect);
glUniform4f(uniform_location, static_cast<float>(R.left), static_cast<float>(R.top),
static_cast<float>(R.right), static_cast<float>(R.bottom));
glDrawArrays(GL_TRIANGLE_STRIP, 0, 4);
glDrawArrays(GL_TRIANGLE_STRIP, 0, 4);
FramebufferManager::SetFramebuffer(0);
g_renderer->RestoreAPIState();
FramebufferManager::SetFramebuffer(0);
g_renderer->RestoreAPIState();
}
void TextureCache::CopyEFB(u8* dst, u32 format, u32 native_width, u32 bytes_per_row, u32 num_blocks_y, u32 memory_stride,
PEControl::PixelFormat srcFormat, const EFBRectangle& srcRect,
bool isIntensity, bool scaleByHalf)
void TextureCache::CopyEFB(u8* dst, u32 format, u32 native_width, u32 bytes_per_row,
u32 num_blocks_y, u32 memory_stride, PEControl::PixelFormat srcFormat,
const EFBRectangle& srcRect, bool isIntensity, bool scaleByHalf)
{
TextureConverter::EncodeToRamFromTexture(
dst,
format,
native_width,
bytes_per_row,
num_blocks_y,
memory_stride,
srcFormat,
isIntensity,
scaleByHalf,
srcRect);
TextureConverter::EncodeToRamFromTexture(dst, format, native_width, bytes_per_row, num_blocks_y,
memory_stride, srcFormat, isIntensity, scaleByHalf,
srcRect);
}
TextureCache::TextureCache()
{
CompileShaders();
CompileShaders();
s_ActiveTexture = -1;
for (auto& gtex : s_Textures)
gtex = -1;
s_ActiveTexture = -1;
for (auto& gtex : s_Textures)
gtex = -1;
if (g_ActiveConfig.backend_info.bSupportsPaletteConversion)
{
s32 buffer_size = 1024 * 1024;
s32 max_buffer_size = 0;
if (g_ActiveConfig.backend_info.bSupportsPaletteConversion)
{
s32 buffer_size = 1024 * 1024;
s32 max_buffer_size = 0;
// The minimum MAX_TEXTURE_BUFFER_SIZE that the spec mandates
// is 65KB, we are asking for a 1MB buffer here.
// Make sure to check the maximum size and if it is below 1MB
// then use the maximum the hardware supports instead.
glGetIntegerv(GL_MAX_TEXTURE_BUFFER_SIZE, &max_buffer_size);
buffer_size = std::min(buffer_size, max_buffer_size);
// The minimum MAX_TEXTURE_BUFFER_SIZE that the spec mandates
// is 65KB, we are asking for a 1MB buffer here.
// Make sure to check the maximum size and if it is below 1MB
// then use the maximum the hardware supports instead.
glGetIntegerv(GL_MAX_TEXTURE_BUFFER_SIZE, &max_buffer_size);
buffer_size = std::min(buffer_size, max_buffer_size);
s_palette_stream_buffer = StreamBuffer::Create(GL_TEXTURE_BUFFER, buffer_size);
glGenTextures(1, &s_palette_resolv_texture);
glBindTexture(GL_TEXTURE_BUFFER, s_palette_resolv_texture);
glTexBuffer(GL_TEXTURE_BUFFER, GL_R16UI, s_palette_stream_buffer->m_buffer);
}
s_palette_stream_buffer = StreamBuffer::Create(GL_TEXTURE_BUFFER, buffer_size);
glGenTextures(1, &s_palette_resolv_texture);
glBindTexture(GL_TEXTURE_BUFFER, s_palette_resolv_texture);
glTexBuffer(GL_TEXTURE_BUFFER, GL_R16UI, s_palette_stream_buffer->m_buffer);
}
}
TextureCache::~TextureCache()
{
DeleteShaders();
DeleteShaders();
if (g_ActiveConfig.backend_info.bSupportsPaletteConversion)
{
s_palette_stream_buffer.reset();
glDeleteTextures(1, &s_palette_resolv_texture);
}
if (g_ActiveConfig.backend_info.bSupportsPaletteConversion)
{
s_palette_stream_buffer.reset();
glDeleteTextures(1, &s_palette_resolv_texture);
}
}
void TextureCache::DisableStage(unsigned int stage)
@ -330,109 +309,117 @@ void TextureCache::DisableStage(unsigned int stage)
void TextureCache::SetStage()
{
// -1 is the initial value as we don't know which texture should be bound
if (s_ActiveTexture != (u32)-1)
glActiveTexture(GL_TEXTURE0 + s_ActiveTexture);
// -1 is the initial value as we don't know which texture should be bound
if (s_ActiveTexture != (u32)-1)
glActiveTexture(GL_TEXTURE0 + s_ActiveTexture);
}
void TextureCache::CompileShaders()
{
constexpr const char* color_copy_program =
"SAMPLER_BINDING(9) uniform sampler2DArray samp9;\n"
"in vec3 f_uv0;\n"
"out vec4 ocol0;\n"
"\n"
"void main(){\n"
" vec4 texcol = texture(samp9, f_uv0);\n"
" ocol0 = texcol;\n"
"}\n";
constexpr const char* color_copy_program = "SAMPLER_BINDING(9) uniform sampler2DArray samp9;\n"
"in vec3 f_uv0;\n"
"out vec4 ocol0;\n"
"\n"
"void main(){\n"
" vec4 texcol = texture(samp9, f_uv0);\n"
" ocol0 = texcol;\n"
"}\n";
constexpr const char* color_matrix_program =
"SAMPLER_BINDING(9) uniform sampler2DArray samp9;\n"
"uniform vec4 colmat[7];\n"
"in vec3 f_uv0;\n"
"out vec4 ocol0;\n"
"\n"
"void main(){\n"
" vec4 texcol = texture(samp9, f_uv0);\n"
" texcol = round(texcol * colmat[5]) * colmat[6];\n"
" ocol0 = texcol * mat4(colmat[0], colmat[1], colmat[2], colmat[3]) + colmat[4];\n"
"}\n";
constexpr const char* color_matrix_program =
"SAMPLER_BINDING(9) uniform sampler2DArray samp9;\n"
"uniform vec4 colmat[7];\n"
"in vec3 f_uv0;\n"
"out vec4 ocol0;\n"
"\n"
"void main(){\n"
" vec4 texcol = texture(samp9, f_uv0);\n"
" texcol = round(texcol * colmat[5]) * colmat[6];\n"
" ocol0 = texcol * mat4(colmat[0], colmat[1], colmat[2], colmat[3]) + colmat[4];\n"
"}\n";
constexpr const char* depth_matrix_program =
"SAMPLER_BINDING(9) uniform sampler2DArray samp9;\n"
"uniform vec4 colmat[5];\n"
"in vec3 f_uv0;\n"
"out vec4 ocol0;\n"
"\n"
"void main(){\n"
" vec4 texcol = texture(samp9, vec3(f_uv0.xy, %s));\n"
" int depth = int(texcol.x * 16777216.0);\n"
constexpr const char* depth_matrix_program =
"SAMPLER_BINDING(9) uniform sampler2DArray samp9;\n"
"uniform vec4 colmat[5];\n"
"in vec3 f_uv0;\n"
"out vec4 ocol0;\n"
"\n"
"void main(){\n"
" vec4 texcol = texture(samp9, vec3(f_uv0.xy, %s));\n"
" int depth = int(texcol.x * 16777216.0);\n"
// Convert to Z24 format
" ivec4 workspace;\n"
" workspace.r = (depth >> 16) & 255;\n"
" workspace.g = (depth >> 8) & 255;\n"
" workspace.b = depth & 255;\n"
// Convert to Z24 format
" ivec4 workspace;\n"
" workspace.r = (depth >> 16) & 255;\n"
" workspace.g = (depth >> 8) & 255;\n"
" workspace.b = depth & 255;\n"
// Convert to Z4 format
" workspace.a = (depth >> 16) & 0xF0;\n"
// Convert to Z4 format
" workspace.a = (depth >> 16) & 0xF0;\n"
// Normalize components to [0.0..1.0]
" texcol = vec4(workspace) / 255.0;\n"
// Normalize components to [0.0..1.0]
" texcol = vec4(workspace) / 255.0;\n"
" ocol0 = texcol * mat4(colmat[0], colmat[1], colmat[2], colmat[3]) + colmat[4];\n"
"}\n";
" ocol0 = texcol * mat4(colmat[0], colmat[1], colmat[2], colmat[3]) + colmat[4];\n"
"}\n";
constexpr const char* vertex_program =
"out vec3 %s_uv0;\n"
"SAMPLER_BINDING(9) uniform sampler2DArray samp9;\n"
"uniform vec4 copy_position;\n" // left, top, right, bottom
"void main()\n"
"{\n"
" vec2 rawpos = vec2(gl_VertexID&1, gl_VertexID&2);\n"
" %s_uv0 = vec3(mix(copy_position.xy, copy_position.zw, rawpos) / vec2(textureSize(samp9, 0).xy), 0.0);\n"
" gl_Position = vec4(rawpos*2.0-1.0, 0.0, 1.0);\n"
"}\n";
constexpr const char* vertex_program =
"out vec3 %s_uv0;\n"
"SAMPLER_BINDING(9) uniform sampler2DArray samp9;\n"
"uniform vec4 copy_position;\n" // left, top, right, bottom
"void main()\n"
"{\n"
" vec2 rawpos = vec2(gl_VertexID&1, gl_VertexID&2);\n"
" %s_uv0 = vec3(mix(copy_position.xy, copy_position.zw, rawpos) / vec2(textureSize(samp9, "
"0).xy), 0.0);\n"
" gl_Position = vec4(rawpos*2.0-1.0, 0.0, 1.0);\n"
"}\n";
const std::string geo_program = g_ActiveConfig.iStereoMode > 0 ?
"layout(triangles) in;\n"
"layout(triangle_strip, max_vertices = 6) out;\n"
"in vec3 v_uv0[3];\n"
"out vec3 f_uv0;\n"
"SAMPLER_BINDING(9) uniform sampler2DArray samp9;\n"
"void main()\n"
"{\n"
" int layers = textureSize(samp9, 0).z;\n"
" for (int layer = 0; layer < layers; ++layer) {\n"
" for (int i = 0; i < 3; ++i) {\n"
" f_uv0 = vec3(v_uv0[i].xy, layer);\n"
" gl_Position = gl_in[i].gl_Position;\n"
" gl_Layer = layer;\n"
" EmitVertex();\n"
" }\n"
" EndPrimitive();\n"
" }\n"
"}\n" : "";
const std::string geo_program = g_ActiveConfig.iStereoMode > 0 ?
"layout(triangles) in;\n"
"layout(triangle_strip, max_vertices = 6) out;\n"
"in vec3 v_uv0[3];\n"
"out vec3 f_uv0;\n"
"SAMPLER_BINDING(9) uniform sampler2DArray samp9;\n"
"void main()\n"
"{\n"
" int layers = textureSize(samp9, 0).z;\n"
" for (int layer = 0; layer < layers; ++layer) {\n"
" for (int i = 0; i < 3; ++i) {\n"
" f_uv0 = vec3(v_uv0[i].xy, layer);\n"
" gl_Position = gl_in[i].gl_Position;\n"
" gl_Layer = layer;\n"
" EmitVertex();\n"
" }\n"
" EndPrimitive();\n"
" }\n"
"}\n" :
"";
const char* prefix = geo_program.empty() ? "f" : "v";
const char* depth_layer = g_ActiveConfig.bStereoEFBMonoDepth ? "0.0" : "f_uv0.z";
const char* prefix = geo_program.empty() ? "f" : "v";
const char* depth_layer = g_ActiveConfig.bStereoEFBMonoDepth ? "0.0" : "f_uv0.z";
ProgramShaderCache::CompileShader(s_ColorCopyProgram, StringFromFormat(vertex_program, prefix, prefix).c_str(), color_copy_program, geo_program);
ProgramShaderCache::CompileShader(s_ColorMatrixProgram, StringFromFormat(vertex_program, prefix, prefix).c_str(), color_matrix_program, geo_program);
ProgramShaderCache::CompileShader(s_DepthMatrixProgram, StringFromFormat(vertex_program, prefix, prefix).c_str(), StringFromFormat(depth_matrix_program, depth_layer).c_str(), geo_program);
ProgramShaderCache::CompileShader(s_ColorCopyProgram,
StringFromFormat(vertex_program, prefix, prefix).c_str(),
color_copy_program, geo_program);
ProgramShaderCache::CompileShader(s_ColorMatrixProgram,
StringFromFormat(vertex_program, prefix, prefix).c_str(),
color_matrix_program, geo_program);
ProgramShaderCache::CompileShader(
s_DepthMatrixProgram, StringFromFormat(vertex_program, prefix, prefix).c_str(),
StringFromFormat(depth_matrix_program, depth_layer).c_str(), geo_program);
s_ColorMatrixUniform = glGetUniformLocation(s_ColorMatrixProgram.glprogid, "colmat");
s_DepthMatrixUniform = glGetUniformLocation(s_DepthMatrixProgram.glprogid, "colmat");
s_ColorCbufid = -1;
s_DepthCbufid = -1;
s_ColorMatrixUniform = glGetUniformLocation(s_ColorMatrixProgram.glprogid, "colmat");
s_DepthMatrixUniform = glGetUniformLocation(s_DepthMatrixProgram.glprogid, "colmat");
s_ColorCbufid = -1;
s_DepthCbufid = -1;
s_ColorCopyPositionUniform = glGetUniformLocation(s_ColorCopyProgram.glprogid, "copy_position");
s_ColorMatrixPositionUniform = glGetUniformLocation(s_ColorMatrixProgram.glprogid, "copy_position");
s_DepthCopyPositionUniform = glGetUniformLocation(s_DepthMatrixProgram.glprogid, "copy_position");
s_ColorCopyPositionUniform = glGetUniformLocation(s_ColorCopyProgram.glprogid, "copy_position");
s_ColorMatrixPositionUniform =
glGetUniformLocation(s_ColorMatrixProgram.glprogid, "copy_position");
s_DepthCopyPositionUniform = glGetUniformLocation(s_DepthMatrixProgram.glprogid, "copy_position");
std::string palette_shader =
R"GLSL(
std::string palette_shader =
R"GLSL(
uniform int texture_buffer_offset;
uniform float multiplier;
SAMPLER_BINDING(9) uniform sampler2DArray samp9;
@ -511,83 +498,88 @@ void TextureCache::CompileShaders()
}
)GLSL";
if (g_ActiveConfig.backend_info.bSupportsPaletteConversion)
{
ProgramShaderCache::CompileShader(
s_palette_pixel_shader[GX_TL_IA8],
StringFromFormat(vertex_program, prefix, prefix),
"#define DECODE DecodePixel_IA8" + palette_shader,
geo_program);
s_palette_buffer_offset_uniform[GX_TL_IA8] = glGetUniformLocation(s_palette_pixel_shader[GX_TL_IA8].glprogid, "texture_buffer_offset");
s_palette_multiplier_uniform[GX_TL_IA8] = glGetUniformLocation(s_palette_pixel_shader[GX_TL_IA8].glprogid, "multiplier");
s_palette_copy_position_uniform[GX_TL_IA8] = glGetUniformLocation(s_palette_pixel_shader[GX_TL_IA8].glprogid, "copy_position");
if (g_ActiveConfig.backend_info.bSupportsPaletteConversion)
{
ProgramShaderCache::CompileShader(
s_palette_pixel_shader[GX_TL_IA8], StringFromFormat(vertex_program, prefix, prefix),
"#define DECODE DecodePixel_IA8" + palette_shader, geo_program);
s_palette_buffer_offset_uniform[GX_TL_IA8] =
glGetUniformLocation(s_palette_pixel_shader[GX_TL_IA8].glprogid, "texture_buffer_offset");
s_palette_multiplier_uniform[GX_TL_IA8] =
glGetUniformLocation(s_palette_pixel_shader[GX_TL_IA8].glprogid, "multiplier");
s_palette_copy_position_uniform[GX_TL_IA8] =
glGetUniformLocation(s_palette_pixel_shader[GX_TL_IA8].glprogid, "copy_position");
ProgramShaderCache::CompileShader(
s_palette_pixel_shader[GX_TL_RGB565],
StringFromFormat(vertex_program, prefix, prefix),
"#define DECODE DecodePixel_RGB565" + palette_shader,
geo_program);
s_palette_buffer_offset_uniform[GX_TL_RGB565] = glGetUniformLocation(s_palette_pixel_shader[GX_TL_RGB565].glprogid, "texture_buffer_offset");
s_palette_multiplier_uniform[GX_TL_RGB565] = glGetUniformLocation(s_palette_pixel_shader[GX_TL_RGB565].glprogid, "multiplier");
s_palette_copy_position_uniform[GX_TL_RGB565] = glGetUniformLocation(s_palette_pixel_shader[GX_TL_RGB565].glprogid, "copy_position");
ProgramShaderCache::CompileShader(
s_palette_pixel_shader[GX_TL_RGB565], StringFromFormat(vertex_program, prefix, prefix),
"#define DECODE DecodePixel_RGB565" + palette_shader, geo_program);
s_palette_buffer_offset_uniform[GX_TL_RGB565] = glGetUniformLocation(
s_palette_pixel_shader[GX_TL_RGB565].glprogid, "texture_buffer_offset");
s_palette_multiplier_uniform[GX_TL_RGB565] =
glGetUniformLocation(s_palette_pixel_shader[GX_TL_RGB565].glprogid, "multiplier");
s_palette_copy_position_uniform[GX_TL_RGB565] =
glGetUniformLocation(s_palette_pixel_shader[GX_TL_RGB565].glprogid, "copy_position");
ProgramShaderCache::CompileShader(
s_palette_pixel_shader[GX_TL_RGB5A3],
StringFromFormat(vertex_program, prefix, prefix),
"#define DECODE DecodePixel_RGB5A3" + palette_shader,
geo_program);
s_palette_buffer_offset_uniform[GX_TL_RGB5A3] = glGetUniformLocation(s_palette_pixel_shader[GX_TL_RGB5A3].glprogid, "texture_buffer_offset");
s_palette_multiplier_uniform[GX_TL_RGB5A3] = glGetUniformLocation(s_palette_pixel_shader[GX_TL_RGB5A3].glprogid, "multiplier");
s_palette_copy_position_uniform[GX_TL_RGB5A3] = glGetUniformLocation(s_palette_pixel_shader[GX_TL_RGB5A3].glprogid, "copy_position");
}
ProgramShaderCache::CompileShader(
s_palette_pixel_shader[GX_TL_RGB5A3], StringFromFormat(vertex_program, prefix, prefix),
"#define DECODE DecodePixel_RGB5A3" + palette_shader, geo_program);
s_palette_buffer_offset_uniform[GX_TL_RGB5A3] = glGetUniformLocation(
s_palette_pixel_shader[GX_TL_RGB5A3].glprogid, "texture_buffer_offset");
s_palette_multiplier_uniform[GX_TL_RGB5A3] =
glGetUniformLocation(s_palette_pixel_shader[GX_TL_RGB5A3].glprogid, "multiplier");
s_palette_copy_position_uniform[GX_TL_RGB5A3] =
glGetUniformLocation(s_palette_pixel_shader[GX_TL_RGB5A3].glprogid, "copy_position");
}
}
void TextureCache::DeleteShaders()
{
s_ColorMatrixProgram.Destroy();
s_DepthMatrixProgram.Destroy();
s_ColorMatrixProgram.Destroy();
s_DepthMatrixProgram.Destroy();
if (g_ActiveConfig.backend_info.bSupportsPaletteConversion)
for (auto& shader : s_palette_pixel_shader)
shader.Destroy();
if (g_ActiveConfig.backend_info.bSupportsPaletteConversion)
for (auto& shader : s_palette_pixel_shader)
shader.Destroy();
}
void TextureCache::ConvertTexture(TCacheEntryBase* _entry, TCacheEntryBase* _unconverted, void* palette, TlutFormat format)
void TextureCache::ConvertTexture(TCacheEntryBase* _entry, TCacheEntryBase* _unconverted,
void* palette, TlutFormat format)
{
if (!g_ActiveConfig.backend_info.bSupportsPaletteConversion)
return;
if (!g_ActiveConfig.backend_info.bSupportsPaletteConversion)
return;
g_renderer->ResetAPIState();
g_renderer->ResetAPIState();
TCacheEntry* entry = (TCacheEntry*) _entry;
TCacheEntry* unconverted = (TCacheEntry*) _unconverted;
TCacheEntry* entry = (TCacheEntry*)_entry;
TCacheEntry* unconverted = (TCacheEntry*)_unconverted;
glActiveTexture(GL_TEXTURE9);
glBindTexture(GL_TEXTURE_2D_ARRAY, unconverted->texture);
g_sampler_cache->BindNearestSampler(9);
glActiveTexture(GL_TEXTURE9);
glBindTexture(GL_TEXTURE_2D_ARRAY, unconverted->texture);
g_sampler_cache->BindNearestSampler(9);
FramebufferManager::SetFramebuffer(entry->framebuffer);
glViewport(0, 0, entry->config.width, entry->config.height);
s_palette_pixel_shader[format].Bind();
FramebufferManager::SetFramebuffer(entry->framebuffer);
glViewport(0, 0, entry->config.width, entry->config.height);
s_palette_pixel_shader[format].Bind();
// C14 textures are currently unsupported
int size = (unconverted->format & 0xf) == GX_TF_I4 ? 32 : 512;
auto buffer = s_palette_stream_buffer->Map(size);
memcpy(buffer.first, palette, size);
s_palette_stream_buffer->Unmap(size);
glUniform1i(s_palette_buffer_offset_uniform[format], buffer.second / 2);
glUniform1f(s_palette_multiplier_uniform[format], (unconverted->format & 0xf) == 0 ? 15.0f : 255.0f);
glUniform4f(s_palette_copy_position_uniform[format], 0.0f, 0.0f, (float)unconverted->config.width, (float)unconverted->config.height);
// C14 textures are currently unsupported
int size = (unconverted->format & 0xf) == GX_TF_I4 ? 32 : 512;
auto buffer = s_palette_stream_buffer->Map(size);
memcpy(buffer.first, palette, size);
s_palette_stream_buffer->Unmap(size);
glUniform1i(s_palette_buffer_offset_uniform[format], buffer.second / 2);
glUniform1f(s_palette_multiplier_uniform[format],
(unconverted->format & 0xf) == 0 ? 15.0f : 255.0f);
glUniform4f(s_palette_copy_position_uniform[format], 0.0f, 0.0f, (float)unconverted->config.width,
(float)unconverted->config.height);
glActiveTexture(GL_TEXTURE10);
glBindTexture(GL_TEXTURE_BUFFER, s_palette_resolv_texture);
g_sampler_cache->BindNearestSampler(10);
glActiveTexture(GL_TEXTURE10);
glBindTexture(GL_TEXTURE_BUFFER, s_palette_resolv_texture);
g_sampler_cache->BindNearestSampler(10);
OpenGL_BindAttributelessVAO();
glDrawArrays(GL_TRIANGLE_STRIP, 0, 4);
OpenGL_BindAttributelessVAO();
glDrawArrays(GL_TRIANGLE_STRIP, 0, 4);
FramebufferManager::SetFramebuffer(0);
g_renderer->RestoreAPIState();
FramebufferManager::SetFramebuffer(0);
g_renderer->RestoreAPIState();
}
}

65
Source/Core/VideoBackends/OGL/TextureCache.h
View File

@ -14,54 +14,53 @@
namespace OGL
{
class TextureCache : public TextureCacheBase
{
public:
TextureCache();
~TextureCache();
TextureCache();
~TextureCache();
static void DisableStage(unsigned int stage);
static void SetStage();
static void DisableStage(unsigned int stage);
static void SetStage();
private:
struct TCacheEntry : TCacheEntryBase
{
GLuint texture;
GLuint framebuffer;
struct TCacheEntry : TCacheEntryBase
{
GLuint texture;
GLuint framebuffer;
//TexMode0 mode; // current filter and clamp modes that texture is set to
//TexMode1 mode1; // current filter and clamp modes that texture is set to
// TexMode0 mode; // current filter and clamp modes that texture is set to
// TexMode1 mode1; // current filter and clamp modes that texture is set to
TCacheEntry(const TCacheEntryConfig& config);
~TCacheEntry();
TCacheEntry(const TCacheEntryConfig& config);
~TCacheEntry();
void CopyRectangleFromTexture(
const TCacheEntryBase* source,
const MathUtil::Rectangle<int> &srcrect,
const MathUtil::Rectangle<int> &dstrect) override;
void CopyRectangleFromTexture(const TCacheEntryBase* source,
const MathUtil::Rectangle<int>& srcrect,
const MathUtil::Rectangle<int>& dstrect) override;
void Load(unsigned int width, unsigned int height,
unsigned int expanded_width, unsigned int level) override;
void Load(unsigned int width, unsigned int height, unsigned int expanded_width,
unsigned int level) override;
void FromRenderTarget(u8 *dst, PEControl::PixelFormat srcFormat, const EFBRectangle& srcRect,
bool scaleByHalf, unsigned int cbufid, const float *colmat) override;
void FromRenderTarget(u8* dst, PEControl::PixelFormat srcFormat, const EFBRectangle& srcRect,
bool scaleByHalf, unsigned int cbufid, const float* colmat) override;
void Bind(unsigned int stage) override;
bool Save(const std::string& filename, unsigned int level) override;
};
void Bind(unsigned int stage) override;
bool Save(const std::string& filename, unsigned int level) override;
};
TCacheEntryBase* CreateTexture(const TCacheEntryConfig& config) override;
void ConvertTexture(TCacheEntryBase* entry, TCacheEntryBase* unconverted, void* palette, TlutFormat format) override;
TCacheEntryBase* CreateTexture(const TCacheEntryConfig& config) override;
void ConvertTexture(TCacheEntryBase* entry, TCacheEntryBase* unconverted, void* palette,
TlutFormat format) override;
void CopyEFB(u8* dst, u32 format, u32 native_width, u32 bytes_per_row, u32 num_blocks_y, u32 memory_stride,
PEControl::PixelFormat srcFormat, const EFBRectangle& srcRect,
bool isIntensity, bool scaleByHalf) override;
void CopyEFB(u8* dst, u32 format, u32 native_width, u32 bytes_per_row, u32 num_blocks_y,
u32 memory_stride, PEControl::PixelFormat srcFormat, const EFBRectangle& srcRect,
bool isIntensity, bool scaleByHalf) override;
void CompileShaders() override;
void DeleteShaders() override;
void CompileShaders() override;
void DeleteShaders() override;
};
bool SaveTexture(const std::string& filename, u32 textarget, u32 tex, int virtual_width, int virtual_height, unsigned int level);
bool SaveTexture(const std::string& filename, u32 textarget, u32 tex, int virtual_width,
int virtual_height, unsigned int level);
}

459
Source/Core/VideoBackends/OGL/TextureConverter.cpp
View File

@ -24,18 +24,15 @@
#include "VideoCommon/TextureConversionShader.h"
#include "VideoCommon/VideoConfig.h"
namespace OGL
{
namespace TextureConverter
{
using OGL::TextureCache;
static GLuint s_texConvFrameBuffer[2] = {0,0};
static GLuint s_srcTexture = 0; // for decoding from RAM
static GLuint s_dstTexture = 0; // for encoding to RAM
static GLuint s_texConvFrameBuffer[2] = {0, 0};
static GLuint s_srcTexture = 0; // for decoding from RAM
static GLuint s_dstTexture = 0; // for encoding to RAM
const int renderBufferWidth = EFB_WIDTH * 4;
const int renderBufferHeight = 1024;
@ -50,301 +47,305 @@ const u32 NUM_ENCODING_PROGRAMS = 64;
static SHADER s_encodingPrograms[NUM_ENCODING_PROGRAMS];
static int s_encodingUniforms[NUM_ENCODING_PROGRAMS];
static GLuint s_PBO = 0; // for readback with different strides
static GLuint s_PBO = 0; // for readback with different strides
static void CreatePrograms()
{
/* TODO: Accuracy Improvements
*
* This shader doesn't really match what the GameCube does internally in the
* copy pipeline.
* 1. It uses OpenGL's built in filtering when yscaling, someone could work
* out how the copypipeline does it's filtering and implement it correctly
* in this shader.
* 2. Deflickering isn't implemented, a futher filtering over 3 lines.
* Isn't really needed on non-interlaced monitors (and would lower quality;
* But hey, accuracy!)
* 3. Flipper's YUYV conversion implements a 3 pixel horizontal blur on the
* UV channels, centering the U channel on the Left pixel and the V channel
* on the Right pixel.
* The current implementation Centers both UV channels at the same place
* inbetween the two Pixels, and only blurs over these two pixels.
*/
// Output is BGRA because that is slightly faster than RGBA.
const char *VProgramRgbToYuyv =
"out vec2 uv0;\n"
"uniform vec4 copy_position;\n" // left, top, right, bottom
"SAMPLER_BINDING(9) uniform sampler2DArray samp9;\n"
"void main()\n"
"{\n"
" vec2 rawpos = vec2(gl_VertexID&1, gl_VertexID&2);\n"
" gl_Position = vec4(rawpos*2.0-1.0, 0.0, 1.0);\n"
" uv0 = mix(copy_position.xy, copy_position.zw, rawpos) / vec2(textureSize(samp9, 0).xy);\n"
"}\n";
const char *FProgramRgbToYuyv =
"SAMPLER_BINDING(9) uniform sampler2DArray samp9;\n"
"in vec2 uv0;\n"
"out vec4 ocol0;\n"
"void main()\n"
"{\n"
" vec3 c0 = texture(samp9, vec3(uv0 - dFdx(uv0) * 0.25, 0.0)).rgb;\n"
" vec3 c1 = texture(samp9, vec3(uv0 + dFdx(uv0) * 0.25, 0.0)).rgb;\n"
" vec3 c01 = (c0 + c1) * 0.5;\n"
" vec3 y_const = vec3(0.257,0.504,0.098);\n"
" vec3 u_const = vec3(-0.148,-0.291,0.439);\n"
" vec3 v_const = vec3(0.439,-0.368,-0.071);\n"
" vec4 const3 = vec4(0.0625,0.5,0.0625,0.5);\n"
" ocol0 = vec4(dot(c1,y_const),dot(c01,u_const),dot(c0,y_const),dot(c01, v_const)) + const3;\n"
"}\n";
ProgramShaderCache::CompileShader(s_rgbToYuyvProgram, VProgramRgbToYuyv, FProgramRgbToYuyv);
s_rgbToYuyvUniform_loc = glGetUniformLocation(s_rgbToYuyvProgram.glprogid, "copy_position");
/* TODO: Accuracy Improvements
*
* This shader doesn't really match what the GameCube does internally in the
* copy pipeline.
* 1. It uses OpenGL's built in filtering when yscaling, someone could work
* out how the copypipeline does it's filtering and implement it correctly
* in this shader.
* 2. Deflickering isn't implemented, a futher filtering over 3 lines.
* Isn't really needed on non-interlaced monitors (and would lower quality;
* But hey, accuracy!)
* 3. Flipper's YUYV conversion implements a 3 pixel horizontal blur on the
* UV channels, centering the U channel on the Left pixel and the V channel
* on the Right pixel.
* The current implementation Centers both UV channels at the same place
* inbetween the two Pixels, and only blurs over these two pixels.
*/
// Output is BGRA because that is slightly faster than RGBA.
const char* VProgramRgbToYuyv =
"out vec2 uv0;\n"
"uniform vec4 copy_position;\n" // left, top, right, bottom
"SAMPLER_BINDING(9) uniform sampler2DArray samp9;\n"
"void main()\n"
"{\n"
" vec2 rawpos = vec2(gl_VertexID&1, gl_VertexID&2);\n"
" gl_Position = vec4(rawpos*2.0-1.0, 0.0, 1.0);\n"
" uv0 = mix(copy_position.xy, copy_position.zw, rawpos) / vec2(textureSize(samp9, 0).xy);\n"
"}\n";
const char* FProgramRgbToYuyv =
"SAMPLER_BINDING(9) uniform sampler2DArray samp9;\n"
"in vec2 uv0;\n"
"out vec4 ocol0;\n"
"void main()\n"
"{\n"
" vec3 c0 = texture(samp9, vec3(uv0 - dFdx(uv0) * 0.25, 0.0)).rgb;\n"
" vec3 c1 = texture(samp9, vec3(uv0 + dFdx(uv0) * 0.25, 0.0)).rgb;\n"
" vec3 c01 = (c0 + c1) * 0.5;\n"
" vec3 y_const = vec3(0.257,0.504,0.098);\n"
" vec3 u_const = vec3(-0.148,-0.291,0.439);\n"
" vec3 v_const = vec3(0.439,-0.368,-0.071);\n"
" vec4 const3 = vec4(0.0625,0.5,0.0625,0.5);\n"
" ocol0 = vec4(dot(c1,y_const),dot(c01,u_const),dot(c0,y_const),dot(c01, v_const)) + "
"const3;\n"
"}\n";
ProgramShaderCache::CompileShader(s_rgbToYuyvProgram, VProgramRgbToYuyv, FProgramRgbToYuyv);
s_rgbToYuyvUniform_loc = glGetUniformLocation(s_rgbToYuyvProgram.glprogid, "copy_position");
/* TODO: Accuracy Improvements
*
* The YVYU to RGB conversion here matches the RGB to YUYV done above, but
* if a game modifies or adds images to the XFB then it should be using the
* same algorithm as the flipper, and could result in slight color inaccuracies
* when run back through this shader.
*/
const char *VProgramYuyvToRgb =
"void main()\n"
"{\n"
" vec2 rawpos = vec2(gl_VertexID&1, gl_VertexID&2);\n"
" gl_Position = vec4(rawpos*2.0-1.0, 0.0, 1.0);\n"
"}\n";
const char *FProgramYuyvToRgb =
"SAMPLER_BINDING(9) uniform sampler2D samp9;\n"
"in vec2 uv0;\n"
"out vec4 ocol0;\n"
"void main()\n"
"{\n"
" ivec2 uv = ivec2(gl_FragCoord.xy);\n"
// We switch top/bottom here. TODO: move this to screen blit.
" ivec2 ts = textureSize(samp9, 0);\n"
" vec4 c0 = texelFetch(samp9, ivec2(uv.x>>1, ts.y-uv.y-1), 0);\n"
" float y = mix(c0.r, c0.b, (uv.x & 1) == 1);\n"
" float yComp = 1.164 * (y - 0.0625);\n"
" float uComp = c0.g - 0.5;\n"
" float vComp = c0.a - 0.5;\n"
" ocol0 = vec4(yComp + (1.596 * vComp),\n"
" yComp - (0.813 * vComp) - (0.391 * uComp),\n"
" yComp + (2.018 * uComp),\n"
" 1.0);\n"
"}\n";
ProgramShaderCache::CompileShader(s_yuyvToRgbProgram, VProgramYuyvToRgb, FProgramYuyvToRgb);
/* TODO: Accuracy Improvements
*
* The YVYU to RGB conversion here matches the RGB to YUYV done above, but
* if a game modifies or adds images to the XFB then it should be using the
* same algorithm as the flipper, and could result in slight color inaccuracies
* when run back through this shader.
*/
const char* VProgramYuyvToRgb = "void main()\n"
"{\n"
" vec2 rawpos = vec2(gl_VertexID&1, gl_VertexID&2);\n"
" gl_Position = vec4(rawpos*2.0-1.0, 0.0, 1.0);\n"
"}\n";
const char* FProgramYuyvToRgb = "SAMPLER_BINDING(9) uniform sampler2D samp9;\n"
"in vec2 uv0;\n"
"out vec4 ocol0;\n"
"void main()\n"
"{\n"
" ivec2 uv = ivec2(gl_FragCoord.xy);\n"
// We switch top/bottom here. TODO: move this to screen blit.
" ivec2 ts = textureSize(samp9, 0);\n"
" vec4 c0 = texelFetch(samp9, ivec2(uv.x>>1, ts.y-uv.y-1), 0);\n"
" float y = mix(c0.r, c0.b, (uv.x & 1) == 1);\n"
" float yComp = 1.164 * (y - 0.0625);\n"
" float uComp = c0.g - 0.5;\n"
" float vComp = c0.a - 0.5;\n"
" ocol0 = vec4(yComp + (1.596 * vComp),\n"
" yComp - (0.813 * vComp) - (0.391 * uComp),\n"
" yComp + (2.018 * uComp),\n"
" 1.0);\n"
"}\n";
ProgramShaderCache::CompileShader(s_yuyvToRgbProgram, VProgramYuyvToRgb, FProgramYuyvToRgb);
}
static SHADER &GetOrCreateEncodingShader(u32 format)
static SHADER& GetOrCreateEncodingShader(u32 format)
{
if (format >= NUM_ENCODING_PROGRAMS)
{
PanicAlert("Unknown texture copy format: 0x%x\n", format);
return s_encodingPrograms[0];
}
if (format >= NUM_ENCODING_PROGRAMS)
{
PanicAlert("Unknown texture copy format: 0x%x\n", format);
return s_encodingPrograms[0];
}
if (s_encodingPrograms[format].glprogid == 0)
{
const char* shader = TextureConversionShader::GenerateEncodingShader(format, API_OPENGL);
if (s_encodingPrograms[format].glprogid == 0)
{
const char* shader = TextureConversionShader::GenerateEncodingShader(format, API_OPENGL);
#if defined(_DEBUG) || defined(DEBUGFAST)
if (g_ActiveConfig.iLog & CONF_SAVESHADERS && shader)
{
static int counter = 0;
std::string filename = StringFromFormat("%senc_%04i.txt", File::GetUserPath(D_DUMP_IDX).c_str(), counter++);
if (g_ActiveConfig.iLog & CONF_SAVESHADERS && shader)
{
static int counter = 0;
std::string filename =
StringFromFormat("%senc_%04i.txt", File::GetUserPath(D_DUMP_IDX).c_str(), counter++);
SaveData(filename, shader);
}
SaveData(filename, shader);
}
#endif
const char *VProgram =
"void main()\n"
"{\n"
" vec2 rawpos = vec2(gl_VertexID&1, gl_VertexID&2);\n"
" gl_Position = vec4(rawpos*2.0-1.0, 0.0, 1.0);\n"
"}\n";
const char* VProgram = "void main()\n"
"{\n"
" vec2 rawpos = vec2(gl_VertexID&1, gl_VertexID&2);\n"
" gl_Position = vec4(rawpos*2.0-1.0, 0.0, 1.0);\n"
"}\n";
ProgramShaderCache::CompileShader(s_encodingPrograms[format], VProgram, shader);
ProgramShaderCache::CompileShader(s_encodingPrograms[format], VProgram, shader);
s_encodingUniforms[format] = glGetUniformLocation(s_encodingPrograms[format].glprogid, "position");
}
return s_encodingPrograms[format];
s_encodingUniforms[format] =
glGetUniformLocation(s_encodingPrograms[format].glprogid, "position");
}
return s_encodingPrograms[format];
}
void Init()
{
glGenFramebuffers(2, s_texConvFrameBuffer);
glGenFramebuffers(2, s_texConvFrameBuffer);
glActiveTexture(GL_TEXTURE9);
glGenTextures(1, &s_srcTexture);
glBindTexture(GL_TEXTURE_2D, s_srcTexture);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, 0);
glActiveTexture(GL_TEXTURE9);
glGenTextures(1, &s_srcTexture);
glBindTexture(GL_TEXTURE_2D, s_srcTexture);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, 0);
glGenTextures(1, &s_dstTexture);
glBindTexture(GL_TEXTURE_2D, s_dstTexture);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, 0);
glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, renderBufferWidth, renderBufferHeight, 0, GL_RGBA, GL_UNSIGNED_BYTE, nullptr);
glGenTextures(1, &s_dstTexture);
glBindTexture(GL_TEXTURE_2D, s_dstTexture);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, 0);
glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, renderBufferWidth, renderBufferHeight, 0, GL_RGBA,
GL_UNSIGNED_BYTE, nullptr);
FramebufferManager::SetFramebuffer(s_texConvFrameBuffer[0]);
glFramebufferTexture2D(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, s_dstTexture, 0);
FramebufferManager::SetFramebuffer(0);
FramebufferManager::SetFramebuffer(s_texConvFrameBuffer[0]);
glFramebufferTexture2D(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, s_dstTexture, 0);
FramebufferManager::SetFramebuffer(0);
glGenBuffers(1, &s_PBO);
glGenBuffers(1, &s_PBO);
CreatePrograms();
CreatePrograms();
}
void Shutdown()
{
glDeleteTextures(1, &s_srcTexture);
glDeleteTextures(1, &s_dstTexture);
glDeleteBuffers(1, &s_PBO);
glDeleteFramebuffers(2, s_texConvFrameBuffer);
glDeleteTextures(1, &s_srcTexture);
glDeleteTextures(1, &s_dstTexture);
glDeleteBuffers(1, &s_PBO);
glDeleteFramebuffers(2, s_texConvFrameBuffer);
s_rgbToYuyvProgram.Destroy();
s_yuyvToRgbProgram.Destroy();
s_rgbToYuyvProgram.Destroy();
s_yuyvToRgbProgram.Destroy();
for (auto& program : s_encodingPrograms)
program.Destroy();
for (auto& program : s_encodingPrograms)
program.Destroy();
s_srcTexture = 0;
s_dstTexture = 0;
s_PBO = 0;
s_texConvFrameBuffer[0] = 0;
s_texConvFrameBuffer[1] = 0;
s_srcTexture = 0;
s_dstTexture = 0;
s_PBO = 0;
s_texConvFrameBuffer[0] = 0;
s_texConvFrameBuffer[1] = 0;
}
// dst_line_size, writeStride in bytes
static void EncodeToRamUsingShader(GLuint srcTexture,
u8* destAddr, u32 dst_line_size, u32 dstHeight,
u32 writeStride, bool linearFilter)
static void EncodeToRamUsingShader(GLuint srcTexture, u8* destAddr, u32 dst_line_size,
u32 dstHeight, u32 writeStride, bool linearFilter)
{
// switch to texture converter frame buffer
// attach render buffer as color destination
FramebufferManager::SetFramebuffer(s_texConvFrameBuffer[0]);
// switch to texture converter frame buffer
// attach render buffer as color destination
FramebufferManager::SetFramebuffer(s_texConvFrameBuffer[0]);
OpenGL_BindAttributelessVAO();
OpenGL_BindAttributelessVAO();
// set source texture
glActiveTexture(GL_TEXTURE9);
glBindTexture(GL_TEXTURE_2D_ARRAY, srcTexture);
// set source texture
glActiveTexture(GL_TEXTURE9);
glBindTexture(GL_TEXTURE_2D_ARRAY, srcTexture);
if (linearFilter)
g_sampler_cache->BindLinearSampler(9);
else
g_sampler_cache->BindNearestSampler(9);
if (linearFilter)
g_sampler_cache->BindLinearSampler(9);
else
g_sampler_cache->BindNearestSampler(9);
glViewport(0, 0, (GLsizei)(dst_line_size / 4), (GLsizei)dstHeight);
glViewport(0, 0, (GLsizei)(dst_line_size / 4), (GLsizei)dstHeight);
glDrawArrays(GL_TRIANGLE_STRIP, 0, 4);
glDrawArrays(GL_TRIANGLE_STRIP, 0, 4);
int dstSize = dst_line_size * dstHeight;
int dstSize = dst_line_size * dstHeight;
if ((writeStride != dst_line_size) && (dstHeight > 1))
{
// writing to a texture of a different size
// also copy more then one block line, so the different strides matters
// copy into one pbo first, map this buffer, and then memcpy into GC memory
// in this way, we only have one vram->ram transfer, but maybe a bigger
// CPU overhead because of the pbo
glBindBuffer(GL_PIXEL_PACK_BUFFER, s_PBO);
glBufferData(GL_PIXEL_PACK_BUFFER, dstSize, nullptr, GL_STREAM_READ);
glReadPixels(0, 0, (GLsizei)(dst_line_size / 4), (GLsizei)dstHeight, GL_BGRA, GL_UNSIGNED_BYTE, nullptr);
u8* pbo = (u8*)glMapBufferRange(GL_PIXEL_PACK_BUFFER, 0, dstSize, GL_MAP_READ_BIT);
if ((writeStride != dst_line_size) && (dstHeight > 1))
{
// writing to a texture of a different size
// also copy more then one block line, so the different strides matters
// copy into one pbo first, map this buffer, and then memcpy into GC memory
// in this way, we only have one vram->ram transfer, but maybe a bigger
// CPU overhead because of the pbo
glBindBuffer(GL_PIXEL_PACK_BUFFER, s_PBO);
glBufferData(GL_PIXEL_PACK_BUFFER, dstSize, nullptr, GL_STREAM_READ);
glReadPixels(0, 0, (GLsizei)(dst_line_size / 4), (GLsizei)dstHeight, GL_BGRA, GL_UNSIGNED_BYTE,
nullptr);
u8* pbo = (u8*)glMapBufferRange(GL_PIXEL_PACK_BUFFER, 0, dstSize, GL_MAP_READ_BIT);
for (size_t i = 0; i < dstHeight; ++i)
{
memcpy(destAddr, pbo, dst_line_size);
pbo += dst_line_size;
destAddr += writeStride;
}
for (size_t i = 0; i < dstHeight; ++i)
{
memcpy(destAddr, pbo, dst_line_size);
pbo += dst_line_size;
destAddr += writeStride;
}
glUnmapBuffer(GL_PIXEL_PACK_BUFFER);
glBindBuffer(GL_PIXEL_PACK_BUFFER, 0);
}
else
{
glReadPixels(0, 0, (GLsizei)(dst_line_size / 4), (GLsizei)dstHeight, GL_BGRA, GL_UNSIGNED_BYTE, destAddr);
}
glUnmapBuffer(GL_PIXEL_PACK_BUFFER);
glBindBuffer(GL_PIXEL_PACK_BUFFER, 0);
}
else
{
glReadPixels(0, 0, (GLsizei)(dst_line_size / 4), (GLsizei)dstHeight, GL_BGRA, GL_UNSIGNED_BYTE,
destAddr);
}
}
void EncodeToRamFromTexture(u8* dest_ptr, u32 format, u32 native_width, u32 bytes_per_row, u32 num_blocks_y, u32 memory_stride,
PEControl::PixelFormat srcFormat, bool bIsIntensityFmt, int bScaleByHalf, const EFBRectangle& source)
void EncodeToRamFromTexture(u8* dest_ptr, u32 format, u32 native_width, u32 bytes_per_row,
u32 num_blocks_y, u32 memory_stride, PEControl::PixelFormat srcFormat,
bool bIsIntensityFmt, int bScaleByHalf, const EFBRectangle& source)
{
g_renderer->ResetAPIState();
g_renderer->ResetAPIState();
SHADER& texconv_shader = GetOrCreateEncodingShader(format);
SHADER& texconv_shader = GetOrCreateEncodingShader(format);
texconv_shader.Bind();
glUniform4i(s_encodingUniforms[format],
source.left, source.top, native_width, bScaleByHalf ? 2 : 1);
texconv_shader.Bind();
glUniform4i(s_encodingUniforms[format], source.left, source.top, native_width,
bScaleByHalf ? 2 : 1);
const GLuint read_texture = (srcFormat == PEControl::Z24) ?
FramebufferManager::ResolveAndGetDepthTarget(source) :
FramebufferManager::ResolveAndGetRenderTarget(source);
const GLuint read_texture = (srcFormat == PEControl::Z24) ?
FramebufferManager::ResolveAndGetDepthTarget(source) :
FramebufferManager::ResolveAndGetRenderTarget(source);
EncodeToRamUsingShader(read_texture,
dest_ptr, bytes_per_row, num_blocks_y,
memory_stride, bScaleByHalf > 0 && srcFormat != PEControl::Z24);
EncodeToRamUsingShader(read_texture, dest_ptr, bytes_per_row, num_blocks_y, memory_stride,
bScaleByHalf > 0 && srcFormat != PEControl::Z24);
FramebufferManager::SetFramebuffer(0);
g_renderer->RestoreAPIState();
FramebufferManager::SetFramebuffer(0);
g_renderer->RestoreAPIState();
}
void EncodeToRamYUYV(GLuint srcTexture, const TargetRectangle& sourceRc, u8* destAddr, u32 dstWidth, u32 dstStride, u32 dstHeight)
void EncodeToRamYUYV(GLuint srcTexture, const TargetRectangle& sourceRc, u8* destAddr, u32 dstWidth,
u32 dstStride, u32 dstHeight)
{
g_renderer->ResetAPIState();
g_renderer->ResetAPIState();
s_rgbToYuyvProgram.Bind();
s_rgbToYuyvProgram.Bind();
glUniform4f(s_rgbToYuyvUniform_loc, static_cast<float>(sourceRc.left), static_cast<float>(sourceRc.top),
static_cast<float>(sourceRc.right), static_cast<float>(sourceRc.bottom));
glUniform4f(s_rgbToYuyvUniform_loc, static_cast<float>(sourceRc.left),
static_cast<float>(sourceRc.top), static_cast<float>(sourceRc.right),
static_cast<float>(sourceRc.bottom));
// We enable linear filtering, because the GameCube does filtering in the vertical direction when
// yscale is enabled.
// Otherwise we get jaggies when a game uses yscaling (most PAL games)
EncodeToRamUsingShader(srcTexture, destAddr, dstWidth * 2, dstHeight, dstStride, true);
FramebufferManager::SetFramebuffer(0);
TextureCache::DisableStage(0);
g_renderer->RestoreAPIState();
// We enable linear filtering, because the GameCube does filtering in the vertical direction when
// yscale is enabled.
// Otherwise we get jaggies when a game uses yscaling (most PAL games)
EncodeToRamUsingShader(srcTexture, destAddr, dstWidth * 2, dstHeight, dstStride, true);
FramebufferManager::SetFramebuffer(0);
TextureCache::DisableStage(0);
g_renderer->RestoreAPIState();
}
// Should be scale free.
void DecodeToTexture(u32 xfbAddr, int srcWidth, int srcHeight, GLuint destTexture)
{
u8* srcAddr = Memory::GetPointer(xfbAddr);
if (!srcAddr)
{
WARN_LOG(VIDEO, "Tried to decode from invalid memory address");
return;
}
u8* srcAddr = Memory::GetPointer(xfbAddr);
if (!srcAddr)
{
WARN_LOG(VIDEO, "Tried to decode from invalid memory address");
return;
}
g_renderer->ResetAPIState(); // reset any game specific settings
g_renderer->ResetAPIState(); // reset any game specific settings
OpenGL_BindAttributelessVAO();
OpenGL_BindAttributelessVAO();
// switch to texture converter frame buffer
// attach destTexture as color destination
FramebufferManager::SetFramebuffer(s_texConvFrameBuffer[1]);
FramebufferManager::FramebufferTexture(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D_ARRAY, destTexture, 0);
// switch to texture converter frame buffer
// attach destTexture as color destination
FramebufferManager::SetFramebuffer(s_texConvFrameBuffer[1]);
FramebufferManager::FramebufferTexture(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D_ARRAY,
destTexture, 0);
// activate source texture
// set srcAddr as data for source texture
glActiveTexture(GL_TEXTURE9);
glBindTexture(GL_TEXTURE_2D, s_srcTexture);
glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, srcWidth / 2, srcHeight, 0, GL_RGBA, GL_UNSIGNED_BYTE, srcAddr);
g_sampler_cache->BindNearestSampler(9);
// activate source texture
// set srcAddr as data for source texture
glActiveTexture(GL_TEXTURE9);
glBindTexture(GL_TEXTURE_2D, s_srcTexture);
glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, srcWidth / 2, srcHeight, 0, GL_RGBA, GL_UNSIGNED_BYTE,
srcAddr);
g_sampler_cache->BindNearestSampler(9);
glViewport(0, 0, srcWidth, srcHeight);
s_yuyvToRgbProgram.Bind();
glViewport(0, 0, srcWidth, srcHeight);
s_yuyvToRgbProgram.Bind();
glDrawArrays(GL_TRIANGLE_STRIP, 0, 4);
glDrawArrays(GL_TRIANGLE_STRIP, 0, 4);
FramebufferManager::SetFramebuffer(0);
FramebufferManager::SetFramebuffer(0);
g_renderer->RestoreAPIState();
g_renderer->RestoreAPIState();
}
} // namespace

12
Source/Core/VideoBackends/OGL/TextureConverter.h
View File

@ -11,24 +11,22 @@
namespace OGL
{
// Converts textures between formats using shaders
// TODO: support multiple texture formats
namespace TextureConverter
{
void Init();
void Shutdown();
void EncodeToRamYUYV(GLuint srcTexture, const TargetRectangle& sourceRc,
u8* destAddr, u32 dstWidth, u32 dstStride, u32 dstHeight);
void EncodeToRamYUYV(GLuint srcTexture, const TargetRectangle& sourceRc, u8* destAddr, u32 dstWidth,
u32 dstStride, u32 dstHeight);
void DecodeToTexture(u32 xfbAddr, int srcWidth, int srcHeight, GLuint destTexture);
// returns size of the encoded data (in bytes)
void EncodeToRamFromTexture(u8* dest_ptr, u32 format, u32 native_width, u32 bytes_per_row, u32 num_blocks_y, u32 memory_stride,
PEControl::PixelFormat srcFormat, bool bIsIntensityFmt, int bScaleByHalf, const EFBRectangle& source);
void EncodeToRamFromTexture(u8* dest_ptr, u32 format, u32 native_width, u32 bytes_per_row,
u32 num_blocks_y, u32 memory_stride, PEControl::PixelFormat srcFormat,
bool bIsIntensityFmt, int bScaleByHalf, const EFBRectangle& source);
}
} // namespace OGL

265
Source/Core/VideoBackends/OGL/VertexManager.cpp
View File

@ -9,8 +9,8 @@
#include "Common/CommonTypes.h"
#include "Common/FileUtil.h"
#include "Common/StringUtil.h"
#include "Common/GL/GLExtensions/GLExtensions.h"
#include "Common/StringUtil.h"
#include "VideoBackends/OGL/ProgramShaderCache.h"
#include "VideoBackends/OGL/Render.h"
@ -25,199 +25,204 @@
namespace OGL
{
//This are the initially requested size for the buffers expressed in bytes
const u32 MAX_IBUFFER_SIZE = 2*1024*1024;
const u32 MAX_VBUFFER_SIZE = 32*1024*1024;
// This are the initially requested size for the buffers expressed in bytes
const u32 MAX_IBUFFER_SIZE = 2 * 1024 * 1024;
const u32 MAX_VBUFFER_SIZE = 32 * 1024 * 1024;
static std::unique_ptr<StreamBuffer> s_vertexBuffer;
static std::unique_ptr<StreamBuffer> s_indexBuffer;
static size_t s_baseVertex;
static size_t s_index_offset;
VertexManager::VertexManager()
: m_cpu_v_buffer(MAX_VBUFFER_SIZE), m_cpu_i_buffer(MAX_IBUFFER_SIZE)
VertexManager::VertexManager() : m_cpu_v_buffer(MAX_VBUFFER_SIZE), m_cpu_i_buffer(MAX_IBUFFER_SIZE)
{
CreateDeviceObjects();
CreateDeviceObjects();
}
VertexManager::~VertexManager()
{
DestroyDeviceObjects();
DestroyDeviceObjects();
}
void VertexManager::CreateDeviceObjects()
{
s_vertexBuffer = StreamBuffer::Create(GL_ARRAY_BUFFER, MAX_VBUFFER_SIZE);
m_vertex_buffers = s_vertexBuffer->m_buffer;
s_vertexBuffer = StreamBuffer::Create(GL_ARRAY_BUFFER, MAX_VBUFFER_SIZE);
m_vertex_buffers = s_vertexBuffer->m_buffer;
s_indexBuffer = StreamBuffer::Create(GL_ELEMENT_ARRAY_BUFFER, MAX_IBUFFER_SIZE);
m_index_buffers = s_indexBuffer->m_buffer;
s_indexBuffer = StreamBuffer::Create(GL_ELEMENT_ARRAY_BUFFER, MAX_IBUFFER_SIZE);
m_index_buffers = s_indexBuffer->m_buffer;
m_last_vao = 0;
m_last_vao = 0;
}
void VertexManager::DestroyDeviceObjects()
{
s_vertexBuffer.reset();
s_indexBuffer.reset();
s_vertexBuffer.reset();
s_indexBuffer.reset();
}
void VertexManager::PrepareDrawBuffers(u32 stride)
{
u32 vertex_data_size = IndexGenerator::GetNumVerts() * stride;
u32 index_data_size = IndexGenerator::GetIndexLen() * sizeof(u16);
u32 vertex_data_size = IndexGenerator::GetNumVerts() * stride;
u32 index_data_size = IndexGenerator::GetIndexLen() * sizeof(u16);
s_vertexBuffer->Unmap(vertex_data_size);
s_indexBuffer->Unmap(index_data_size);
s_vertexBuffer->Unmap(vertex_data_size);
s_indexBuffer->Unmap(index_data_size);
ADDSTAT(stats.thisFrame.bytesVertexStreamed, vertex_data_size);
ADDSTAT(stats.thisFrame.bytesIndexStreamed, index_data_size);
ADDSTAT(stats.thisFrame.bytesVertexStreamed, vertex_data_size);
ADDSTAT(stats.thisFrame.bytesIndexStreamed, index_data_size);
}
void VertexManager::ResetBuffer(u32 stride)
{
if (s_cull_all)
{
// This buffer isn't getting sent to the GPU. Just allocate it on the cpu.
s_pCurBufferPointer = s_pBaseBufferPointer = m_cpu_v_buffer.data();
s_pEndBufferPointer = s_pBaseBufferPointer + m_cpu_v_buffer.size();
if (s_cull_all)
{
// This buffer isn't getting sent to the GPU. Just allocate it on the cpu.
s_pCurBufferPointer = s_pBaseBufferPointer = m_cpu_v_buffer.data();
s_pEndBufferPointer = s_pBaseBufferPointer + m_cpu_v_buffer.size();
IndexGenerator::Start((u16*)m_cpu_i_buffer.data());
}
else
{
auto buffer = s_vertexBuffer->Map(MAXVBUFFERSIZE, stride);
s_pCurBufferPointer = s_pBaseBufferPointer = buffer.first;
s_pEndBufferPointer = buffer.first + MAXVBUFFERSIZE;
s_baseVertex = buffer.second / stride;
IndexGenerator::Start((u16*)m_cpu_i_buffer.data());
}
else
{
auto buffer = s_vertexBuffer->Map(MAXVBUFFERSIZE, stride);
s_pCurBufferPointer = s_pBaseBufferPointer = buffer.first;
s_pEndBufferPointer = buffer.first + MAXVBUFFERSIZE;
s_baseVertex = buffer.second / stride;
buffer = s_indexBuffer->Map(MAXIBUFFERSIZE * sizeof(u16));
IndexGenerator::Start((u16*)buffer.first);
s_index_offset = buffer.second;
}
buffer = s_indexBuffer->Map(MAXIBUFFERSIZE * sizeof(u16));
IndexGenerator::Start((u16*)buffer.first);
s_index_offset = buffer.second;
}
}
void VertexManager::Draw(u32 stride)
{
u32 index_size = IndexGenerator::GetIndexLen();
u32 max_index = IndexGenerator::GetNumVerts();
GLenum primitive_mode = 0;
u32 index_size = IndexGenerator::GetIndexLen();
u32 max_index = IndexGenerator::GetNumVerts();
GLenum primitive_mode = 0;
switch (current_primitive_type)
{
case PRIMITIVE_POINTS:
primitive_mode = GL_POINTS;
glDisable(GL_CULL_FACE);
break;
case PRIMITIVE_LINES:
primitive_mode = GL_LINES;
glDisable(GL_CULL_FACE);
break;
case PRIMITIVE_TRIANGLES:
primitive_mode = g_ActiveConfig.backend_info.bSupportsPrimitiveRestart ? GL_TRIANGLE_STRIP : GL_TRIANGLES;
break;
}
switch (current_primitive_type)
{
case PRIMITIVE_POINTS:
primitive_mode = GL_POINTS;
glDisable(GL_CULL_FACE);
break;
case PRIMITIVE_LINES:
primitive_mode = GL_LINES;
glDisable(GL_CULL_FACE);
break;
case PRIMITIVE_TRIANGLES:
primitive_mode =
g_ActiveConfig.backend_info.bSupportsPrimitiveRestart ? GL_TRIANGLE_STRIP : GL_TRIANGLES;
break;
}
if (g_ogl_config.bSupportsGLBaseVertex)
{
glDrawRangeElementsBaseVertex(primitive_mode, 0, max_index, index_size, GL_UNSIGNED_SHORT, (u8*)nullptr+s_index_offset, (GLint)s_baseVertex);
}
else
{
glDrawRangeElements(primitive_mode, 0, max_index, index_size, GL_UNSIGNED_SHORT, (u8*)nullptr+s_index_offset);
}
if (g_ogl_config.bSupportsGLBaseVertex)
{
glDrawRangeElementsBaseVertex(primitive_mode, 0, max_index, index_size, GL_UNSIGNED_SHORT,
(u8*)nullptr + s_index_offset, (GLint)s_baseVertex);
}
else
{
glDrawRangeElements(primitive_mode, 0, max_index, index_size, GL_UNSIGNED_SHORT,
(u8*)nullptr + s_index_offset);
}
INCSTAT(stats.thisFrame.numDrawCalls);
INCSTAT(stats.thisFrame.numDrawCalls);
if (current_primitive_type != PRIMITIVE_TRIANGLES)
static_cast<Renderer*>(g_renderer.get())->SetGenerationMode();
if (current_primitive_type != PRIMITIVE_TRIANGLES)
static_cast<Renderer*>(g_renderer.get())->SetGenerationMode();
}
void VertexManager::vFlush(bool useDstAlpha)
{
GLVertexFormat *nativeVertexFmt = (GLVertexFormat*)VertexLoaderManager::GetCurrentVertexFormat();
u32 stride = nativeVertexFmt->GetVertexStride();
GLVertexFormat* nativeVertexFmt = (GLVertexFormat*)VertexLoaderManager::GetCurrentVertexFormat();
u32 stride = nativeVertexFmt->GetVertexStride();
if (m_last_vao != nativeVertexFmt->VAO)
{
glBindVertexArray(nativeVertexFmt->VAO);
m_last_vao = nativeVertexFmt->VAO;
}
if (m_last_vao != nativeVertexFmt->VAO)
{
glBindVertexArray(nativeVertexFmt->VAO);
m_last_vao = nativeVertexFmt->VAO;
}
PrepareDrawBuffers(stride);
PrepareDrawBuffers(stride);
// Makes sure we can actually do Dual source blending
bool dualSourcePossible = g_ActiveConfig.backend_info.bSupportsDualSourceBlend;
// Makes sure we can actually do Dual source blending
bool dualSourcePossible = g_ActiveConfig.backend_info.bSupportsDualSourceBlend;
// If host supports GL_ARB_blend_func_extended, we can do dst alpha in
// the same pass as regular rendering.
if (useDstAlpha && dualSourcePossible)
{
ProgramShaderCache::SetShader(DSTALPHA_DUAL_SOURCE_BLEND, current_primitive_type);
}
else
{
ProgramShaderCache::SetShader(DSTALPHA_NONE, current_primitive_type);
}
// If host supports GL_ARB_blend_func_extended, we can do dst alpha in
// the same pass as regular rendering.
if (useDstAlpha && dualSourcePossible)
{
ProgramShaderCache::SetShader(DSTALPHA_DUAL_SOURCE_BLEND, current_primitive_type);
}
else
{
ProgramShaderCache::SetShader(DSTALPHA_NONE, current_primitive_type);
}
// upload global constants
ProgramShaderCache::UploadConstants();
// upload global constants
ProgramShaderCache::UploadConstants();
// setup the pointers
nativeVertexFmt->SetupVertexPointers();
// setup the pointers
nativeVertexFmt->SetupVertexPointers();
Draw(stride);
Draw(stride);
// run through vertex groups again to set alpha
if (useDstAlpha && !dualSourcePossible)
{
ProgramShaderCache::SetShader(DSTALPHA_ALPHA_PASS, current_primitive_type);
// run through vertex groups again to set alpha
if (useDstAlpha && !dualSourcePossible)
{
ProgramShaderCache::SetShader(DSTALPHA_ALPHA_PASS, current_primitive_type);
// only update alpha
glColorMask(GL_FALSE, GL_FALSE, GL_FALSE, GL_TRUE);
// only update alpha
glColorMask(GL_FALSE, GL_FALSE, GL_FALSE, GL_TRUE);
glDisable(GL_BLEND);
glDisable(GL_BLEND);
Draw(stride);
Draw(stride);
// restore color mask
g_renderer->SetColorMask();
// restore color mask
g_renderer->SetColorMask();
if (bpmem.blendmode.blendenable || bpmem.blendmode.subtract)
glEnable(GL_BLEND);
}
if (bpmem.blendmode.blendenable || bpmem.blendmode.subtract)
glEnable(GL_BLEND);
}
#if defined(_DEBUG) || defined(DEBUGFAST)
if (g_ActiveConfig.iLog & CONF_SAVESHADERS)
{
// save the shaders
ProgramShaderCache::PCacheEntry prog = ProgramShaderCache::GetShaderProgram();
std::string filename = StringFromFormat("%sps%.3d.txt", File::GetUserPath(D_DUMPFRAMES_IDX).c_str(), g_ActiveConfig.iSaveTargetId);
std::ofstream fps;
OpenFStream(fps, filename, std::ios_base::out);
fps << prog.shader.strpprog.c_str();
if (g_ActiveConfig.iLog & CONF_SAVESHADERS)
{
// save the shaders
ProgramShaderCache::PCacheEntry prog = ProgramShaderCache::GetShaderProgram();
std::string filename = StringFromFormat(
"%sps%.3d.txt", File::GetUserPath(D_DUMPFRAMES_IDX).c_str(), g_ActiveConfig.iSaveTargetId);
std::ofstream fps;
OpenFStream(fps, filename, std::ios_base::out);
fps << prog.shader.strpprog.c_str();
filename = StringFromFormat("%svs%.3d.txt", File::GetUserPath(D_DUMPFRAMES_IDX).c_str(), g_ActiveConfig.iSaveTargetId);
std::ofstream fvs;
OpenFStream(fvs, filename, std::ios_base::out);
fvs << prog.shader.strvprog.c_str();
}
filename = StringFromFormat("%svs%.3d.txt", File::GetUserPath(D_DUMPFRAMES_IDX).c_str(),
g_ActiveConfig.iSaveTargetId);
std::ofstream fvs;
OpenFStream(fvs, filename, std::ios_base::out);
fvs << prog.shader.strvprog.c_str();
}
if (g_ActiveConfig.iLog & CONF_SAVETARGETS)
{
std::string filename = StringFromFormat("%starg%.3d.png", File::GetUserPath(D_DUMPFRAMES_IDX).c_str(), g_ActiveConfig.iSaveTargetId);
TargetRectangle tr;
tr.left = 0;
tr.right = Renderer::GetTargetWidth();
tr.top = 0;
tr.bottom = Renderer::GetTargetHeight();
g_renderer->SaveScreenshot(filename, tr);
}
if (g_ActiveConfig.iLog & CONF_SAVETARGETS)
{
std::string filename =
StringFromFormat("%starg%.3d.png", File::GetUserPath(D_DUMPFRAMES_IDX).c_str(),
g_ActiveConfig.iSaveTargetId);
TargetRectangle tr;
tr.left = 0;
tr.right = Renderer::GetTargetWidth();
tr.top = 0;
tr.bottom = Renderer::GetTargetHeight();
g_renderer->SaveScreenshot(filename, tr);
}
#endif
g_Config.iSaveTargetId++;
g_Config.iSaveTargetId++;
ClearEFBCache();
ClearEFBCache();
}
} // namespace

50
Source/Core/VideoBackends/OGL/VertexManager.h
View File

@ -9,43 +9,43 @@
namespace OGL
{
class GLVertexFormat : public NativeVertexFormat
{
public:
GLVertexFormat(const PortableVertexDeclaration& vtx_decl);
~GLVertexFormat();
class GLVertexFormat : public NativeVertexFormat
{
public:
GLVertexFormat(const PortableVertexDeclaration& vtx_decl);
~GLVertexFormat();
void SetupVertexPointers() override;
void SetupVertexPointers() override;
GLuint VAO;
};
GLuint VAO;
};
// Handles the OpenGL details of drawing lots of vertices quickly.
// Other functionality is moving out.
class VertexManager : public VertexManagerBase
{
public:
VertexManager();
~VertexManager();
NativeVertexFormat* CreateNativeVertexFormat(const PortableVertexDeclaration& vtx_decl) override;
void CreateDeviceObjects() override;
void DestroyDeviceObjects() override;
VertexManager();
~VertexManager();
NativeVertexFormat* CreateNativeVertexFormat(const PortableVertexDeclaration& vtx_decl) override;
void CreateDeviceObjects() override;
void DestroyDeviceObjects() override;
// NativeVertexFormat use this
GLuint m_vertex_buffers;
GLuint m_index_buffers;
GLuint m_last_vao;
// NativeVertexFormat use this
GLuint m_vertex_buffers;
GLuint m_index_buffers;
GLuint m_last_vao;
protected:
void ResetBuffer(u32 stride) override;
void ResetBuffer(u32 stride) override;
private:
void Draw(u32 stride);
void vFlush(bool useDstAlpha) override;
void PrepareDrawBuffers(u32 stride);
void Draw(u32 stride);
void vFlush(bool useDstAlpha) override;
void PrepareDrawBuffers(u32 stride);
// Alternative buffers in CPU memory for primatives we are going to discard.
std::vector<u8> m_cpu_v_buffer;
std::vector<u16> m_cpu_i_buffer;
// Alternative buffers in CPU memory for primatives we are going to discard.
std::vector<u8> m_cpu_v_buffer;
std::vector<u16> m_cpu_i_buffer;
};
}

18
Source/Core/VideoBackends/OGL/VideoBackend.h
View File

@ -9,21 +9,19 @@
namespace OGL
{
class VideoBackend : public VideoBackendBase
{
bool Initialize(void *) override;
void Shutdown() override;
bool Initialize(void*) override;
void Shutdown() override;
std::string GetName() const override;
std::string GetDisplayName() const override;
std::string GetName() const override;
std::string GetDisplayName() const override;
void Video_Prepare() override;
void Video_Cleanup() override;
void Video_Prepare() override;
void Video_Cleanup() override;
void ShowConfig(void* parent) override;
void ShowConfig(void* parent) override;
unsigned int PeekMessages() override;
unsigned int PeekMessages() override;
};
}

205
Source/Core/VideoBackends/OGL/main.cpp
View File

@ -2,8 +2,6 @@
// Licensed under GPLv2+
// Refer to the license.txt file included.
// OpenGL Backend Documentation
/*
@ -74,168 +72,165 @@ Make AA apply instantly during gameplay if possible
namespace OGL
{
// Draw messages on top of the screen
unsigned int VideoBackend::PeekMessages()
{
return GLInterface->PeekMessages();
return GLInterface->PeekMessages();
}
std::string VideoBackend::GetName() const
{
return "OGL";
return "OGL";
}
std::string VideoBackend::GetDisplayName() const
{
if (GLInterface != nullptr && GLInterface->GetMode() == GLInterfaceMode::MODE_OPENGLES3)
return "OpenGLES";
else
return "OpenGL";
if (GLInterface != nullptr && GLInterface->GetMode() == GLInterfaceMode::MODE_OPENGLES3)
return "OpenGLES";
else
return "OpenGL";
}
static std::vector<std::string> GetShaders(const std::string &sub_dir = "")
static std::vector<std::string> GetShaders(const std::string& sub_dir = "")
{
std::vector<std::string> paths = DoFileSearch({".glsl"}, {
File::GetUserPath(D_SHADERS_IDX) + sub_dir,
File::GetSysDirectory() + SHADERS_DIR DIR_SEP + sub_dir
});
std::vector<std::string> result;
for (std::string path : paths)
{
std::string name;
SplitPath(path, nullptr, &name, nullptr);
result.push_back(name);
}
return result;
std::vector<std::string> paths =
DoFileSearch({".glsl"}, {File::GetUserPath(D_SHADERS_IDX) + sub_dir,
File::GetSysDirectory() + SHADERS_DIR DIR_SEP + sub_dir});
std::vector<std::string> result;
for (std::string path : paths)
{
std::string name;
SplitPath(path, nullptr, &name, nullptr);
result.push_back(name);
}
return result;
}
static void InitBackendInfo()
{
g_Config.backend_info.APIType = API_OPENGL;
g_Config.backend_info.bSupportsExclusiveFullscreen = false;
g_Config.backend_info.bSupportsOversizedViewports = true;
g_Config.backend_info.bSupportsGeometryShaders = true;
g_Config.backend_info.bSupports3DVision = false;
g_Config.backend_info.bSupportsPostProcessing = true;
g_Config.backend_info.bSupportsSSAA = true;
g_Config.backend_info.APIType = API_OPENGL;
g_Config.backend_info.bSupportsExclusiveFullscreen = false;
g_Config.backend_info.bSupportsOversizedViewports = true;
g_Config.backend_info.bSupportsGeometryShaders = true;
g_Config.backend_info.bSupports3DVision = false;
g_Config.backend_info.bSupportsPostProcessing = true;
g_Config.backend_info.bSupportsSSAA = true;
g_Config.backend_info.Adapters.clear();
g_Config.backend_info.Adapters.clear();
// aamodes - 1 is to stay consistent with D3D (means no AA)
g_Config.backend_info.AAModes = { 1, 2, 4, 8 };
// aamodes - 1 is to stay consistent with D3D (means no AA)
g_Config.backend_info.AAModes = {1, 2, 4, 8};
// pp shaders
g_Config.backend_info.PPShaders = GetShaders("");
g_Config.backend_info.AnaglyphShaders = GetShaders(ANAGLYPH_DIR DIR_SEP);
// pp shaders
g_Config.backend_info.PPShaders = GetShaders("");
g_Config.backend_info.AnaglyphShaders = GetShaders(ANAGLYPH_DIR DIR_SEP);
}
void VideoBackend::ShowConfig(void* parent_handle)
{
if (!m_initialized)
InitBackendInfo();
if (!m_initialized)
InitBackendInfo();
Host_ShowVideoConfig(parent_handle, GetDisplayName(), "gfx_opengl");
Host_ShowVideoConfig(parent_handle, GetDisplayName(), "gfx_opengl");
}
bool VideoBackend::Initialize(void* window_handle)
{
InitializeShared();
InitBackendInfo();
InitializeShared();
InitBackendInfo();
frameCount = 0;
frameCount = 0;
if (File::Exists(File::GetUserPath(D_CONFIG_IDX) + "GFX.ini"))
g_Config.Load(File::GetUserPath(D_CONFIG_IDX) + "GFX.ini");
else
g_Config.Load(File::GetUserPath(D_CONFIG_IDX) + "gfx_opengl.ini");
g_Config.GameIniLoad();
g_Config.UpdateProjectionHack();
g_Config.VerifyValidity();
UpdateActiveConfig();
if (File::Exists(File::GetUserPath(D_CONFIG_IDX) + "GFX.ini"))
g_Config.Load(File::GetUserPath(D_CONFIG_IDX) + "GFX.ini");
else
g_Config.Load(File::GetUserPath(D_CONFIG_IDX) + "gfx_opengl.ini");
g_Config.GameIniLoad();
g_Config.UpdateProjectionHack();
g_Config.VerifyValidity();
UpdateActiveConfig();
InitInterface();
GLInterface->SetMode(GLInterfaceMode::MODE_DETECT);
if (!GLInterface->Create(window_handle))
return false;
InitInterface();
GLInterface->SetMode(GLInterfaceMode::MODE_DETECT);
if (!GLInterface->Create(window_handle))
return false;
// Do our OSD callbacks
OSD::DoCallbacks(OSD::CallbackType::Initialization);
// Do our OSD callbacks
OSD::DoCallbacks(OSD::CallbackType::Initialization);
m_initialized = true;
m_initialized = true;
return true;
return true;
}
// This is called after Initialize() from the Core
// Run from the graphics thread
void VideoBackend::Video_Prepare()
{
GLInterface->MakeCurrent();
GLInterface->MakeCurrent();
g_renderer = std::make_unique<Renderer>();
g_renderer = std::make_unique<Renderer>();
CommandProcessor::Init();
PixelEngine::Init();
CommandProcessor::Init();
PixelEngine::Init();
BPInit();
g_vertex_manager = std::make_unique<VertexManager>();
g_perf_query = GetPerfQuery();
Fifo::Init(); // must be done before OpcodeDecoder::Init()
OpcodeDecoder::Init();
IndexGenerator::Init();
VertexShaderManager::Init();
PixelShaderManager::Init();
GeometryShaderManager::Init();
ProgramShaderCache::Init();
g_texture_cache = std::make_unique<TextureCache>();
g_sampler_cache = std::make_unique<SamplerCache>();
Renderer::Init();
VertexLoaderManager::Init();
TextureConverter::Init();
BoundingBox::Init();
BPInit();
g_vertex_manager = std::make_unique<VertexManager>();
g_perf_query = GetPerfQuery();
Fifo::Init(); // must be done before OpcodeDecoder::Init()
OpcodeDecoder::Init();
IndexGenerator::Init();
VertexShaderManager::Init();
PixelShaderManager::Init();
GeometryShaderManager::Init();
ProgramShaderCache::Init();
g_texture_cache = std::make_unique<TextureCache>();
g_sampler_cache = std::make_unique<SamplerCache>();
Renderer::Init();
VertexLoaderManager::Init();
TextureConverter::Init();
BoundingBox::Init();
// Notify the core that the video backend is ready
Host_Message(WM_USER_CREATE);
// Notify the core that the video backend is ready
Host_Message(WM_USER_CREATE);
}
void VideoBackend::Shutdown()
{
m_initialized = false;
m_initialized = false;
// Do our OSD callbacks
OSD::DoCallbacks(OSD::CallbackType::Shutdown);
// Do our OSD callbacks
OSD::DoCallbacks(OSD::CallbackType::Shutdown);
GLInterface->Shutdown();
GLInterface.reset();
GLInterface->Shutdown();
GLInterface.reset();
}
void VideoBackend::Video_Cleanup()
{
if (!g_renderer)
return;
if (!g_renderer)
return;
Fifo::Shutdown();
Fifo::Shutdown();
// The following calls are NOT Thread Safe
// And need to be called from the video thread
Renderer::Shutdown();
BoundingBox::Shutdown();
TextureConverter::Shutdown();
VertexLoaderManager::Shutdown();
g_sampler_cache.reset();
g_texture_cache.reset();
ProgramShaderCache::Shutdown();
VertexShaderManager::Shutdown();
PixelShaderManager::Shutdown();
GeometryShaderManager::Shutdown();
// The following calls are NOT Thread Safe
// And need to be called from the video thread
Renderer::Shutdown();
BoundingBox::Shutdown();
TextureConverter::Shutdown();
VertexLoaderManager::Shutdown();
g_sampler_cache.reset();
g_texture_cache.reset();
ProgramShaderCache::Shutdown();
VertexShaderManager::Shutdown();
PixelShaderManager::Shutdown();
GeometryShaderManager::Shutdown();
g_perf_query.reset();
g_vertex_manager.reset();
g_perf_query.reset();
g_vertex_manager.reset();
OpcodeDecoder::Shutdown();
g_renderer.reset();
GLInterface->ClearCurrent();
OpcodeDecoder::Shutdown();
g_renderer.reset();
GLInterface->ClearCurrent();
}
}