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
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
|
#include "optick.config.h"
#if USE_OPTICK
#include "optick_gpu.h"
#include "optick_core.h"
#include "optick_memory.h"
#include <thread>
namespace Optick
{
static_assert((1ULL << 32) % GPUProfiler::MAX_QUERIES_COUNT == 0, "(1 << 32) should be a multiple of MAX_QUERIES_COUNT to handle query index overflow!");
GPUProfiler::GPUProfiler() : currentState(STATE_OFF), currentNode(0), frameNumber(0)
{
}
void GPUProfiler::InitNode(const char *nodeName, uint32_t nodeIndex)
{
Node* node = Memory::New<Node>();
for (int i = 0; i < GPU_QUEUE_COUNT; ++i)
{
char name[128] = { 0 };
sprintf_s(name, "%s [%s]", nodeName, GetGPUQueueName((GPUQueueType)i));
node->gpuEventStorage[i] = RegisterStorage(name, uint64_t(-1), ThreadMask::GPU);
node->name = nodeName;
}
nodes[nodeIndex] = node;
}
void GPUProfiler::Start(uint32 /*mode*/)
{
std::lock_guard<std::recursive_mutex> lock(updateLock);
Reset();
currentState = STATE_STARTING;
}
void GPUProfiler::Stop(uint32 /*mode*/)
{
std::lock_guard<std::recursive_mutex> lock(updateLock);
currentState = STATE_OFF;
}
void GPUProfiler::Dump(uint32 /*mode*/)
{
for (size_t nodeIndex = 0; nodeIndex < nodes.size(); ++nodeIndex)
{
Node* node = nodes[nodeIndex];
for (int queueIndex = 0; queueIndex < GPU_QUEUE_COUNT; ++queueIndex)
{
EventBuffer& gpuBuffer = node->gpuEventStorage[queueIndex]->eventBuffer;
const vector<ThreadEntry*>& threads = Core::Get().GetThreads();
for (size_t threadIndex = 0; threadIndex < threads.size(); ++threadIndex)
{
ThreadEntry* thread = threads[threadIndex];
thread->storage.gpuStorage.gpuBuffer[nodeIndex][queueIndex].ForEachChunk([&gpuBuffer](const EventData* events, int count)
{
gpuBuffer.AddRange(events, count);
});
}
}
}
}
string GPUProfiler::GetName() const
{
return !nodes.empty() ? nodes[0]->name : string();
}
GPUProfiler::~GPUProfiler()
{
for (Node* node : nodes)
Memory::Delete(node);
nodes.clear();
}
void GPUProfiler::Reset()
{
for (uint32_t nodeIndex = 0; nodeIndex < nodes.size(); ++nodeIndex)
{
Node& node = *nodes[nodeIndex];
node.Reset();
node.clock = GetClockSynchronization(nodeIndex);
}
}
EventData& GPUProfiler::AddFrameEvent()
{
static const EventDescription* GPUFrameDescription = EventDescription::Create("GPU Frame", __FILE__, __LINE__);
EventData& event = nodes[currentNode]->gpuEventStorage[GPU_QUEUE_GRAPHICS]->eventBuffer.Add();
event.description = GPUFrameDescription;
event.start = EventTime::INVALID_TIMESTAMP;
event.finish = EventTime::INVALID_TIMESTAMP;
return event;
}
EventData& GPUProfiler::AddVSyncEvent()
{
static const EventDescription* VSyncDescription = EventDescription::Create("VSync", __FILE__, __LINE__);
EventData& event = nodes[currentNode]->gpuEventStorage[GPU_QUEUE_VSYNC]->eventBuffer.Add();
event.description = VSyncDescription;
event.start = EventTime::INVALID_TIMESTAMP;
event.finish = EventTime::INVALID_TIMESTAMP;
return event;
}
TagData<uint32>& GPUProfiler::AddFrameTag()
{
static const EventDescription* FrameTagDescription = EventDescription::CreateShared("Frame");
TagData<uint32>& tag = nodes[currentNode]->gpuEventStorage[GPU_QUEUE_GRAPHICS]->tagU32Buffer.Add();
tag.description = FrameTagDescription;
tag.timestamp = EventTime::INVALID_TIMESTAMP;
tag.data = Core::Get().GetCurrentFrame();
return tag;
}
const char * GetGPUQueueName(GPUQueueType queue)
{
const char* GPUQueueToName[GPU_QUEUE_COUNT] = { "Graphics", "Compute", "Transfer", "VSync" };
return GPUQueueToName[queue];
}
void GPUProfiler::Node::Reset()
{
queryIndex = 0;
for (size_t frameIndex = 0; frameIndex < queryGpuframes.size(); ++frameIndex)
queryGpuframes[frameIndex].Reset();
}
}
#endif //USE_OPTICK
|
