Media/compute-runtime
1
0
Fork 0
mirror of https://github.com/intel/compute-runtime.git synced 2025-09-15 13:01:45 +08:00
Code Issues Packages Projects Releases Wiki Activity

Add profiling calculation from timestamp packets

Browse Source 
Change-Id: Ie7f8c703ca5ea5eb1f5207871ef94cbc7ece18b7
Signed-off-by: Jobczyk, Lukasz <lukasz.jobczyk@intel.com>
This commit is contained in:
Jobczyk, Lukasz
2019-01-21 11:44:56 +01:00
committed by sys_ocldev
parent 3fe78d263b
commit c1cb1f9be6
7 changed files with 175 additions and 47 deletions
Download Patch File Download Diff File Expand all files Collapse all files

99
runtime/event/event.cpp
View File

@ -1,5 +1,5 @@
/*
* Copyright (C) 2017-2018 Intel Corporation
* Copyright (C) 2017-2019 Intel Corporation
*
* SPDX-License-Identifier: MIT
*
@ -251,49 +251,76 @@ cl_ulong Event::getDelta(cl_ulong startTime,
}
bool Event::calcProfilingData() {
if (!dataCalculated && !profilingCpuPath) {
if (timestampPacketContainer && timestampPacketContainer->peekNodes().size() > 0) {
const auto timestamps = timestampPacketContainer->peekNodes();
uint64_t contextStartTS = timestamps[0]->tag->getData(TimestampPacket::DataIndex::ContextStart);
uint64_t contextEndTS = timestamps[0]->tag->getData(TimestampPacket::DataIndex::ContextEnd);
uint64_t globalStartTS = timestamps[0]->tag->getData(TimestampPacket::DataIndex::GlobalStart);
for (const auto &timestamp : timestamps) {
if (timestamp->tag->getData(TimestampPacket::DataIndex::ContextStart) < contextStartTS) {
contextStartTS = timestamp->tag->getData(TimestampPacket::DataIndex::ContextStart);
}
if (timestamp->tag->getData(TimestampPacket::DataIndex::ContextEnd) > contextEndTS) {
contextEndTS = timestamp->tag->getData(TimestampPacket::DataIndex::ContextEnd);
}
if (timestamp->tag->getData(TimestampPacket::DataIndex::GlobalStart) < globalStartTS) {
globalStartTS = timestamp->tag->getData(TimestampPacket::DataIndex::GlobalStart);
}
}
calcProfilingData(contextStartTS, contextEndTS, &contextEndTS, globalStartTS);
} else if (timeStampNode) {
calcProfilingData(
(reinterpret_cast<HwTimeStamps *>(timeStampNode->tag))->ContextStartTS,
(reinterpret_cast<HwTimeStamps *>(timeStampNode->tag))->ContextEndTS,
&(reinterpret_cast<HwTimeStamps *>(timeStampNode->tag))->ContextCompleteTS,
(reinterpret_cast<HwTimeStamps *>(timeStampNode->tag))->GlobalStartTS);
}
}
return dataCalculated;
}
void Event::calcProfilingData(uint64_t contextStartTS, uint64_t contextEndTS, uint64_t *contextCompleteTS, uint64_t globalStartTS) {
uint64_t gpuDuration = 0;
uint64_t cpuDuration = 0;
uint64_t gpuCompleteDuration = 0;
uint64_t cpuCompleteDuration = 0;
int64_t c0 = 0;
if (!dataCalculated && timeStampNode && !profilingCpuPath) {
double frequency = cmdQueue->getDevice().getDeviceInfo().profilingTimerResolution;
/* calculation based on equation
CpuTime = GpuTime * scalar + const( == c0)
scalar = DeltaCpu( == dCpu) / DeltaGpu( == dGpu)
to determine the value of the const we can use one pair of values
const = CpuTimeQueue - GpuTimeQueue * scalar
*/
double frequency = cmdQueue->getDevice().getDeviceInfo().profilingTimerResolution;
int64_t c0 = queueTimeStamp.CPUTimeinNS - static_cast<uint64_t>(queueTimeStamp.GPUTimeStamp * frequency);
/* calculation based on equation
CpuTime = GpuTime * scalar + const( == c0)
scalar = DeltaCpu( == dCpu) / DeltaGpu( == dGpu)
to determine the value of the const we can use one pair of values
const = CpuTimeQueue - GpuTimeQueue * scalar
*/
//If device enqueue has not updated complete timestamp, assign end timestamp
if (((HwTimeStamps *)timeStampNode->tag)->ContextCompleteTS == 0)
((HwTimeStamps *)timeStampNode->tag)->ContextCompleteTS = ((HwTimeStamps *)timeStampNode->tag)->ContextEndTS;
c0 = queueTimeStamp.CPUTimeinNS - static_cast<uint64_t>(queueTimeStamp.GPUTimeStamp * frequency);
gpuDuration = getDelta(
((HwTimeStamps *)timeStampNode->tag)->ContextStartTS,
((HwTimeStamps *)timeStampNode->tag)->ContextEndTS);
gpuCompleteDuration = getDelta(
((HwTimeStamps *)timeStampNode->tag)->ContextStartTS,
((HwTimeStamps *)timeStampNode->tag)->ContextCompleteTS);
cpuDuration = static_cast<uint64_t>(gpuDuration * frequency);
cpuCompleteDuration = static_cast<uint64_t>(gpuCompleteDuration * frequency);
startTimeStamp = static_cast<uint64_t>(((HwTimeStamps *)timeStampNode->tag)->GlobalStartTS * frequency) + c0;
endTimeStamp = startTimeStamp + cpuDuration;
completeTimeStamp = startTimeStamp + cpuCompleteDuration;
if (DebugManager.flags.ReturnRawGpuTimestamps.get()) {
startTimeStamp = ((HwTimeStamps *)timeStampNode->tag)->ContextStartTS;
endTimeStamp = ((HwTimeStamps *)timeStampNode->tag)->ContextEndTS;
completeTimeStamp = ((HwTimeStamps *)timeStampNode->tag)->ContextCompleteTS;
}
dataCalculated = true;
//If device enqueue has not updated complete timestamp, assign end timestamp
gpuDuration = getDelta(contextStartTS, contextEndTS);
if (*contextCompleteTS == 0) {
*contextCompleteTS = contextEndTS;
gpuCompleteDuration = gpuDuration;
} else {
gpuCompleteDuration = getDelta(contextStartTS, *contextCompleteTS);
}
return dataCalculated;
cpuDuration = static_cast<uint64_t>(gpuDuration * frequency);
cpuCompleteDuration = static_cast<uint64_t>(gpuCompleteDuration * frequency);
startTimeStamp = static_cast<uint64_t>(globalStartTS * frequency) + c0;
endTimeStamp = startTimeStamp + cpuDuration;
completeTimeStamp = startTimeStamp + cpuCompleteDuration;
if (DebugManager.flags.ReturnRawGpuTimestamps.get()) {
startTimeStamp = contextStartTS;
endTimeStamp = contextEndTS;
completeTimeStamp = *contextCompleteTS;
}
dataCalculated = true;
}
inline bool Event::wait(bool blocking, bool useQuickKmdSleep) {