-
Notifications
You must be signed in to change notification settings - Fork 496
Expand file tree
/
Copy pathGPUO2Interface.cxx
More file actions
162 lines (144 loc) · 5.18 KB
/
GPUO2Interface.cxx
File metadata and controls
162 lines (144 loc) · 5.18 KB
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
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
// Copyright CERN and copyright holders of ALICE O2. This software is
// distributed under the terms of the GNU General Public License v3 (GPL
// Version 3), copied verbatim in the file "COPYING".
//
// See http://alice-o2.web.cern.ch/license for full licensing information.
//
// In applying this license CERN does not waive the privileges and immunities
// granted to it by virtue of its status as an Intergovernmental Organization
// or submit itself to any jurisdiction.
/// \file GPUO2Interface.cxx
/// \author David Rohr
#include "GPUO2Interface.h"
#include "GPUReconstruction.h"
#include "GPUChainTracking.h"
#include "GPUMemorySizeScalers.h"
#include "GPUOutputControl.h"
#include "GPUO2InterfaceConfiguration.h"
#include "GPUParam.inc"
#include "GPUQA.h"
#include "GPUOutputControl.h"
#include <iostream>
#include <fstream>
using namespace o2::gpu;
#include "DataFormatsTPC/ClusterNative.h"
GPUO2Interface::GPUO2Interface() = default;
GPUO2Interface::~GPUO2Interface() { Deinitialize(); }
int GPUO2Interface::Initialize(const GPUO2InterfaceConfiguration& config)
{
if (mInitialized) {
return (1);
}
mConfig.reset(new GPUO2InterfaceConfiguration(config));
mContinuous = mConfig->configGRP.continuousMaxTimeBin != 0;
mRec.reset(GPUReconstruction::CreateInstance(mConfig->configDeviceBackend));
if (mRec == nullptr) {
GPUError("Error obtaining instance of GPUReconstruction");
return 1;
}
mChain = mRec->AddChain<GPUChainTracking>(mConfig->configInterface.maxTPCHits, mConfig->configInterface.maxTRDTracklets);
mChain->mConfigDisplay = &mConfig->configDisplay;
mChain->mConfigQA = &mConfig->configQA;
if (mConfig->configWorkflow.inputs.isSet(GPUDataTypes::InOutType::TPCRaw)) {
mConfig->configGRP.needsClusterer = 1;
}
mRec->SetSettings(&mConfig->configGRP, &mConfig->configReconstruction, &mConfig->configProcessing, &mConfig->configWorkflow);
mChain->SetCalibObjects(mConfig->configCalib);
mOutputRegions.reset(new GPUTrackingOutputs);
if (mConfig->configInterface.outputToExternalBuffers) {
for (unsigned int i = 0; i < mOutputRegions->count(); i++) {
mChain->SetSubOutputControl(i, &mOutputRegions->asArray()[i]);
}
GPUOutputControl dummy;
dummy.set([](size_t size) -> void* {throw std::runtime_error("invalid output memory request, no common output buffer set"); return nullptr; });
mRec->SetOutputControl(dummy);
}
if (mRec->Init()) {
return (1);
}
if (!mRec->IsGPU() && mRec->GetProcessingSettings().memoryAllocationStrategy == GPUMemoryResource::ALLOCATION_INDIVIDUAL) {
mRec->MemoryScalers()->factor *= 2;
}
mRec->MemoryScalers()->factor *= mConfig->configInterface.memoryBufferScaleFactor;
mInitialized = true;
return (0);
}
void GPUO2Interface::Deinitialize()
{
if (mInitialized) {
mRec->Finalize();
mRec.reset();
}
mInitialized = false;
}
int GPUO2Interface::RunTracking(GPUTrackingInOutPointers* data, GPUInterfaceOutputs* outputs)
{
if (!mInitialized) {
return (1);
}
if (mConfig->configInterface.dumpEvents) {
static int nEvent = 0;
mChain->ClearIOPointers();
mChain->mIOPtrs.clustersNative = data->clustersNative;
mChain->mIOPtrs.tpcPackedDigits = data->tpcPackedDigits;
mChain->mIOPtrs.tpcZS = data->tpcZS;
char fname[1024];
sprintf(fname, "event.%d.dump", nEvent);
mChain->DumpData(fname);
if (nEvent == 0) {
mRec->DumpSettings();
#ifdef GPUCA_BUILD_QA
if (mConfig->configProcessing.runMC) {
mChain->ForceInitQA();
sprintf(fname, "mc.%d.dump", nEvent);
mChain->GetQA()->DumpO2MCData(fname);
}
#endif
}
nEvent++;
if (mConfig->configInterface.dumpEvents >= 2) {
return 0;
}
}
mChain->mIOPtrs = *data;
if (mConfig->configInterface.outputToExternalBuffers) {
for (unsigned int i = 0; i < mOutputRegions->count(); i++) {
if (outputs->asArray()[i].allocator) {
mOutputRegions->asArray()[i].set(outputs->asArray()[i].allocator);
} else if (outputs->asArray()[i].ptrBase) {
mOutputRegions->asArray()[i].set(outputs->asArray()[i].ptrBase, outputs->asArray()[i].size);
} else {
mOutputRegions->asArray()[i].reset();
}
}
}
int retVal = mRec->RunChains();
if (retVal == 2) {
retVal = 0; // 2 signals end of event display, ignore
}
if (retVal) {
mRec->ClearAllocatedMemory();
return retVal;
}
if (mConfig->configQA.shipToQC) {
outputs->qa.hist1 = &mChain->GetQA()->getHistograms1D();
outputs->qa.hist2 = &mChain->GetQA()->getHistograms2D();
outputs->qa.hist3 = &mChain->GetQA()->getHistograms1Dd();
}
*data = mChain->mIOPtrs;
return 0;
}
void GPUO2Interface::Clear(bool clearOutputs) { mRec->ClearAllocatedMemory(clearOutputs); }
void GPUO2Interface::GetClusterErrors2(int row, float z, float sinPhi, float DzDs, short clusterState, float& ErrY2, float& ErrZ2) const
{
mRec->GetParam().GetClusterErrors2(row, z, sinPhi, DzDs, ErrY2, ErrZ2);
mRec->GetParam().UpdateClusterError2ByState(clusterState, ErrY2, ErrZ2);
}
int GPUO2Interface::registerMemoryForGPU(const void* ptr, size_t size)
{
return mRec->registerMemoryForGPU(ptr, size);
}
int GPUO2Interface::unregisterMemoryForGPU(const void* ptr)
{
return mRec->unregisterMemoryForGPU(ptr);
}