intel/compute-runtime
1
0
Fork 0
mirror of https://github.com/intel/compute-runtime.git synced 2026-01-03 23:03:02 +08:00
Code Issues Packages Projects Releases Wiki Activity

performance: Reuse GPU timestamps by default

Browse Source 
Related-To: NEO-10615

Signed-off-by: Szymon Morek <szymon.morek@intel.com>
This commit is contained in:
Szymon Morek
2024-05-15 15:22:40 +00:00
committed by Compute-Runtime-Automation
parent 9989829487
commit 7aceed58ca
6 changed files with 94 additions and 86 deletions
Download Patch File Download Diff File Expand all files Collapse all files

38
opencl/source/event/event.cpp
View File

@@ -395,26 +395,30 @@ void Event::calculateProfilingDataInternal(uint64_t contextStartTS, uint64_t con
auto &gfxCoreHelper = device.getGfxCoreHelper();
auto resolution = device.getDeviceInfo().profilingTimerResolution;
startTimeStamp.gpuTimeStamp = globalStartTS;
addOverflowToTimestamp(startTimeStamp.gpuTimeStamp, submitTimeStamp.gpuTimeStamp);
if (startTimeStamp.gpuTimeStamp < submitTimeStamp.gpuTimeStamp) {
auto diff = submitTimeStamp.gpuTimeStamp - startTimeStamp.gpuTimeStamp;
auto diffInNS = gfxCoreHelper.getGpuTimeStampInNS(diff, resolution);
auto osTime = device.getOSTime();
if (diffInNS < osTime->getTimestampRefreshTimeout()) {
auto alignedSubmitTimestamp = startTimeStamp.gpuTimeStamp - 1;
auto alignedQueueTimestamp = startTimeStamp.gpuTimeStamp - 2;
if (startTimeStamp.gpuTimeStamp <= 2) {
alignedSubmitTimestamp = 0;
alignedQueueTimestamp = 0;
// Calculate startTimestamp only if it was not already set on CPU
if (startTimeStamp.cpuTimeInNs == 0) {
startTimeStamp.gpuTimeStamp = globalStartTS;
addOverflowToTimestamp(startTimeStamp.gpuTimeStamp, submitTimeStamp.gpuTimeStamp);
if (startTimeStamp.gpuTimeStamp < submitTimeStamp.gpuTimeStamp) {
auto diff = submitTimeStamp.gpuTimeStamp - startTimeStamp.gpuTimeStamp;
auto diffInNS = gfxCoreHelper.getGpuTimeStampInNS(diff, resolution);
auto osTime = device.getOSTime();
if (diffInNS < osTime->getTimestampRefreshTimeout()) {
auto alignedSubmitTimestamp = startTimeStamp.gpuTimeStamp - 1;
auto alignedQueueTimestamp = startTimeStamp.gpuTimeStamp - 2;
if (startTimeStamp.gpuTimeStamp <= 2) {
alignedSubmitTimestamp = 0;
alignedQueueTimestamp = 0;
}
updateTimestamp(submitTimeStamp, alignedSubmitTimestamp);
updateTimestamp(queueTimeStamp, alignedQueueTimestamp);
osTime->setRefreshTimestampsFlag();
} else {
startTimeStamp.gpuTimeStamp += static_cast<uint64_t>(1ULL << gfxCoreHelper.getGlobalTimeStampBits());
}
updateTimestamp(submitTimeStamp, alignedSubmitTimestamp);
updateTimestamp(queueTimeStamp, alignedQueueTimestamp);
osTime->setRefreshTimestampsFlag();
} else {
startTimeStamp.gpuTimeStamp += static_cast<uint64_t>(1ULL << gfxCoreHelper.getGlobalTimeStampBits());
}
}
UNRECOVERABLE_IF(startTimeStamp.gpuTimeStamp < submitTimeStamp.gpuTimeStamp);
auto gpuTicksDiff = startTimeStamp.gpuTimeStamp - submitTimeStamp.gpuTimeStamp;
auto timeDiff = static_cast<uint64_t>(gpuTicksDiff * resolution);