ShuriZma
/
dolphin
mirror of https://github.com/dolphin-emu/dolphin.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Stop doing nastly shit to OpenGL stream buffers. 

Instead we keep the loaded vertices in CPU memory.
This commit is contained in:
Scott Mansell 2015-01-24 14:37:20 +13:00
parent 5510c86b81
commit 14baf038e7
6 changed files with 79 additions and 54 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

27
Source/Core/VideoBackends/OGL/VertexManager.cpp
View File

@ -43,6 +43,8 @@ static size_t s_index_offset;
VertexManager::VertexManager()
{
CreateDeviceObjects();
CpuVBuffer.resize(MAX_VBUFFER_SIZE);
CpuIBuffer.resize(MAX_IBUFFER_SIZE);
}
VertexManager::~VertexManager()
@ -81,14 +83,25 @@ void VertexManager::PrepareDrawBuffers(u32 stride)
void VertexManager::ResetBuffer(u32 stride)
{
auto buffer = s_vertexBuffer->Map(MAXVBUFFERSIZE, stride);
s_pCurBufferPointer = s_pBaseBufferPointer = buffer.first;
s_pEndBufferPointer = buffer.first + MAXVBUFFERSIZE;
s_baseVertex = buffer.second / stride;
if (CullAll)
{
// This buffer isn't getting sent to the GPU. Just allocate it on the cpu.
s_pCurBufferPointer = s_pBaseBufferPointer = CpuVBuffer.data();
s_pEndBufferPointer = s_pBaseBufferPointer + CpuVBuffer.size();
buffer = s_indexBuffer->Map(MAXIBUFFERSIZE * sizeof(u16));
IndexGenerator::Start((u16*)buffer.first);
s_index_offset = buffer.second;
IndexGenerator::Start((u16*)CpuIBuffer.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;
}
}
void VertexManager::Draw(u32 stride)

4
Source/Core/VideoBackends/OGL/VertexManager.h
View File

@ -45,6 +45,10 @@ private:
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> CpuVBuffer;
std::vector<u16> CpuIBuffer;
};
}

2
Source/Core/VideoCommon/NativeVertexFormat.h
View File

@ -110,7 +110,7 @@ public:
virtual void SetupVertexPointers() = 0;
u32 GetVertexStride() const { return vtx_decl.stride; }
PortableVertexDeclaration GetVertexDeclaration() const { return vtx_decl; }
const PortableVertexDeclaration& GetVertexDeclaration() const { return vtx_decl; }
// TODO: move this under private:
u32 m_components; // VB_HAS_X. Bitmask telling what vertex components are present.

6
Source/Core/VideoCommon/VertexLoaderManager.cpp
View File

@ -157,8 +157,12 @@ int RunVertices(int vtx_attr_group, int primitive, int count, DataReader src, bo
VertexManager::Flush();
s_current_vtx_fmt = loader->m_native_vertex_format;
// if cull mode is CULL_ALL, tell VertexManager to skip triangles and quads.
// They still need to go through vertex loading, because we need to calculate a zfreeze refrence slope.
bool cullall = (bpmem.genMode.cullmode == GenMode::CULL_ALL && primitive < 5);
DataReader dst = VertexManager::PrepareForAdditionalData(primitive, count,
loader->m_native_vtx_decl.stride);
loader->m_native_vtx_decl.stride, cullall);
count = loader->RunVertices(primitive, count, src, dst);

90
Source/Core/VideoCommon/VertexManagerBase.cpp
View File

@ -29,6 +29,7 @@ PrimitiveType VertexManager::current_primitive_type;
Slope VertexManager::ZSlope;
bool VertexManager::IsFlushed;
bool VertexManager::CullAll;
static const PrimitiveType primitive_from_gx[8] = {
PRIMITIVE_TRIANGLES, // GX_DRAW_QUADS
@ -44,6 +45,7 @@ static const PrimitiveType primitive_from_gx[8] = {
VertexManager::VertexManager()
{
IsFlushed = true;
CullAll = false;
}
VertexManager::~VertexManager()
@ -55,7 +57,7 @@ u32 VertexManager::GetRemainingSize()
return (u32)(s_pEndBufferPointer - s_pCurBufferPointer);
}
DataReader VertexManager::PrepareForAdditionalData(int primitive, u32 count, u32 stride)
DataReader VertexManager::PrepareForAdditionalData(int primitive, u32 count, u32 stride, bool cullall)
{
// The SSE vertex loader can write up to 4 bytes past the end
u32 const needed_vertex_bytes = count * stride + 4;
@ -81,6 +83,8 @@ DataReader VertexManager::PrepareForAdditionalData(int primitive, u32 count, u32
"Increase MAXVBUFFERSIZE or we need primitive breaking after all.");
}
CullAll = cullall;
// need to alloc new buffer
if (IsFlushed)
{
@ -192,34 +196,36 @@ void VertexManager::Flush()
(int)bpmem.genMode.numtexgens, (u32)bpmem.dstalpha.enable, (bpmem.alpha_test.hex>>16)&0xff);
#endif
BitSet32 usedtextures;
for (u32 i = 0; i < bpmem.genMode.numtevstages + 1u; ++i)
if (bpmem.tevorders[i / 2].getEnable(i & 1))
usedtextures[bpmem.tevorders[i/2].getTexMap(i & 1)] = true;
if (bpmem.genMode.numindstages > 0)
for (unsigned int i = 0; i < bpmem.genMode.numtevstages + 1u; ++i)
if (bpmem.tevind[i].IsActive() && bpmem.tevind[i].bt < bpmem.genMode.numindstages)
usedtextures[bpmem.tevindref.getTexMap(bpmem.tevind[i].bt)] = true;
for (unsigned int i : usedtextures)
// If the primitave is marked CullAll. All we need to do is update the vertex constants and calculate the zfreeze refrence slope
if (!CullAll)
{
g_renderer->SetSamplerState(i & 3, i >> 2);
const TextureCache::TCacheEntryBase* tentry = TextureCache::Load(i);
BitSet32 usedtextures;
for (u32 i = 0; i < bpmem.genMode.numtevstages + 1u; ++i)
if (bpmem.tevorders[i / 2].getEnable(i & 1))
usedtextures[bpmem.tevorders[i/2].getTexMap(i & 1)] = true;
if (tentry)
if (bpmem.genMode.numindstages > 0)
for (unsigned int i = 0; i < bpmem.genMode.numtevstages + 1u; ++i)
if (bpmem.tevind[i].IsActive() && bpmem.tevind[i].bt < bpmem.genMode.numindstages)
usedtextures[bpmem.tevindref.getTexMap(bpmem.tevind[i].bt)] = true;
for (unsigned int i : usedtextures)
{
// 0s are probably for no manual wrapping needed.
PixelShaderManager::SetTexDims(i, tentry->native_width, tentry->native_height, 0, 0);
g_renderer->SetSamplerState(i & 3, i >> 2);
const TextureCache::TCacheEntryBase* tentry = TextureCache::Load(i);
if (tentry)
{
// 0s are probably for no manual wrapping needed.
PixelShaderManager::SetTexDims(i, tentry->native_width, tentry->native_height, 0, 0);
}
else
ERROR_LOG(VIDEO, "error loading texture");
}
else
ERROR_LOG(VIDEO, "error loading texture");
}
// set global constants
// set global vertex constants
VertexShaderManager::SetConstants();
GeometryShaderManager::SetConstants();
PixelShaderManager::SetConstants();
// Calculate ZSlope for zfreeze
if (!bpmem.genMode.zfreeze)
@ -227,41 +233,37 @@ void VertexManager::Flush()
// Must be done after VertexShaderManager::SetConstants()
CalculateZSlope(VertexLoaderManager::GetCurrentVertexFormat());
}
else if (ZSlope.dirty) // or apply any dirty ZSlopes
else if (ZSlope.dirty && !CullAll) // or apply any dirty ZSlopes
{
PixelShaderManager::SetZSlope(ZSlope.dfdx, ZSlope.dfdy, ZSlope.f0);
ZSlope.dirty = false;
}
// If cull mode is CULL_ALL, we shouldn't render any triangles/quads (points and lines don't get culled)
// vertex loader has already converted any quads into triangles, so we just check for triangles.
// TODO: These culled primites need to get this far through the pipeline to be used as zfreeze refrence
// planes. But currently we apply excessive processing and store the vertices in buffers on the
// video card, which is a waste of bandwidth.
if (bpmem.genMode.cullmode == GenMode::CULL_ALL && current_primitive_type == PRIMITIVE_TRIANGLES)
if (!CullAll)
{
GFX_DEBUGGER_PAUSE_AT(NEXT_FLUSH, true);
IsFlushed = true;
return;
// set the rest of the global constants
GeometryShaderManager::SetConstants();
PixelShaderManager::SetConstants();
bool useDstAlpha = !g_ActiveConfig.bDstAlphaPass &&
bpmem.dstalpha.enable &&
bpmem.blendmode.alphaupdate &&
bpmem.zcontrol.pixel_format == PEControl::RGBA6_Z24;
if (PerfQueryBase::ShouldEmulate())
g_perf_query->EnableQuery(bpmem.zcontrol.early_ztest ? PQG_ZCOMP_ZCOMPLOC : PQG_ZCOMP);
g_vertex_manager->vFlush(useDstAlpha);
if (PerfQueryBase::ShouldEmulate())
g_perf_query->DisableQuery(bpmem.zcontrol.early_ztest ? PQG_ZCOMP_ZCOMPLOC : PQG_ZCOMP);
}
bool useDstAlpha = !g_ActiveConfig.bDstAlphaPass &&
bpmem.dstalpha.enable &&
bpmem.blendmode.alphaupdate &&
bpmem.zcontrol.pixel_format == PEControl::RGBA6_Z24;
if (PerfQueryBase::ShouldEmulate())
g_perf_query->EnableQuery(bpmem.zcontrol.early_ztest ? PQG_ZCOMP_ZCOMPLOC : PQG_ZCOMP);
g_vertex_manager->vFlush(useDstAlpha);
if (PerfQueryBase::ShouldEmulate())
g_perf_query->DisableQuery(bpmem.zcontrol.early_ztest ? PQG_ZCOMP_ZCOMPLOC : PQG_ZCOMP);
GFX_DEBUGGER_PAUSE_AT(NEXT_FLUSH, true);
if (xfmem.numTexGen.numTexGens != bpmem.genMode.numtexgens)
ERROR_LOG(VIDEO, "xf.numtexgens (%d) does not match bp.numtexgens (%d). Error in command stream.", xfmem.numTexGen.numTexGens, bpmem.genMode.numtexgens.Value());
IsFlushed = true;
CullAll = false;
}
void VertexManager::DoState(PointerWrap& p)
@ -279,7 +281,7 @@ void VertexManager::CalculateZSlope(NativeVertexFormat *format)
// Global matrix ID.
u32 mtxIdx = g_main_cp_state.matrix_index_a.PosNormalMtxIdx;
PortableVertexDeclaration vert_decl = format->GetVertexDeclaration();
const PortableVertexDeclaration vert_decl = format->GetVertexDeclaration();
size_t posOff = vert_decl.position.offset;
size_t mtxOff = vert_decl.posmtx.offset;

4
Source/Core/VideoCommon/VertexManagerBase.h
View File

@ -41,7 +41,7 @@ public:
// needs to be virtual for DX11's dtor
virtual ~VertexManager();
static DataReader PrepareForAdditionalData(int primitive, u32 count, u32 stride);
static DataReader PrepareForAdditionalData(int primitive, u32 count, u32 stride, bool cullall);
static void FlushData(u32 count, u32 stride);
static void Flush();
@ -67,6 +67,8 @@ protected:
static Slope ZSlope;
static void CalculateZSlope(NativeVertexFormat *format);
static bool CullAll;
private:
static bool IsFlushed;