1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
|
#ifndef THREADEDVBO_H
#define THREADEDVBO_H
#include "Runtime/GfxDevice/GfxDevice.h"
#include "Configuration/UnityConfigure.h"
#include "Runtime/Shaders/BufferedVBO.h"
class Mutex;
class GfxDeviceClient;
struct ClientDeviceVBO;
class GfxDeviceWorker;
class ThreadedStreamBuffer;
class ThreadedVBO : public BufferedVBO {
public:
ThreadedVBO(GfxDeviceClient& device);
virtual ~ThreadedVBO();
virtual void UpdateVertexData( const VertexBufferData& buffer );
virtual void UpdateIndexData (const IndexBufferData& buffer);
virtual void DrawVBO (const ChannelAssigns& channels, UInt32 firstIndexByte, UInt32 indexCount,
GfxPrimitiveType topology, UInt32 firstVertex, UInt32 vertexCount );
#if GFX_ENABLE_DRAW_CALL_BATCHING
virtual void DrawCustomIndexed( const ChannelAssigns& channels, void* indices, UInt32 indexCount,
GfxPrimitiveType topology, UInt32 vertexRangeBegin, UInt32 vertexRangeEnd, UInt32 drawVertexCount );
#endif
virtual void Recreate();
// For writing directly to VBO. VBO must be filled (UpdateData)
// at least once; and vertex layout + topology from the last fill
// is used. For example, for skinned meshes you have to call
// UpdateData on start and each time layout/topology changes;
// then map,write,unmap for each skinning.
//
// In some situations a vertex buffer might become lost; then you need to do UpdateData
// again before using Map.
virtual bool MapVertexStream( VertexStreamData& outData, unsigned stream );
virtual void UnmapVertexStream( unsigned stream );
virtual bool IsVertexBufferLost() const;
virtual bool IsIndexBufferLost() const;
virtual void SetMappedFromRenderThread( bool renderThread );
virtual void SetVertexStreamMode( unsigned stream, StreamMode mode );
virtual void SetIndicesDynamic(bool dynamic);
virtual void ResetDynamicVB();
virtual void MarkBuffersLost();
virtual int GetRuntimeMemorySize() const;
virtual void UseAsStreamOutput();
#if UNITY_XENON
virtual void AddExtraUvChannels( const UInt8* data, UInt32 size, int extraUvCount );
virtual void CopyExtraUvChannels( VBO* source );
#endif
ClientDeviceVBO* GetClientDeviceVBO() { return m_ClientVBO; } //Todo:any nicer way?
VBO* GetNonThreadedVBO() { return m_NonThreadedVBO; }
protected:
ThreadedStreamBuffer& GetCommandQueue();
GfxDeviceWorker* GetGfxDeviceWorker();
void SubmitCommands();
void DoLockstep();
GfxDeviceClient& m_ClientDevice;
ClientDeviceVBO* m_ClientVBO;
VBO* m_NonThreadedVBO;
int m_VertexBufferSize;
int m_IndexBufferSize;
bool m_MappedFromRenderThread;
bool m_VertexBufferLost;
bool m_IndexBufferLost;
};
class ThreadedDynamicVBO : public DynamicVBO {
public:
ThreadedDynamicVBO(GfxDeviceClient& device);
virtual ~ThreadedDynamicVBO() { }
virtual bool GetChunk( UInt32 shaderChannelMask, UInt32 maxVertices, UInt32 maxIndices, RenderMode renderMode, void** outVB, void** outIB );
virtual void ReleaseChunk( UInt32 actualVertices, UInt32 actualIndices );
virtual void DrawChunk (const ChannelAssigns& channels);
private:
ThreadedStreamBuffer& GetCommandQueue();
GfxDeviceWorker* GetGfxDeviceWorker();
void SubmitCommands();
void DoLockstep();
GfxDeviceClient& m_ClientDevice;
ThreadedStreamBuffer* m_CommandQueue;
dynamic_array<UInt8> m_ChunkVertices;
dynamic_array<UInt16> m_ChunkIndices;
bool m_ValidChunk;
};
#endif
|