intel/compute-runtime
1
0
Fork 0
mirror of https://github.com/intel/compute-runtime.git synced 2026-01-06 10:26:29 +08:00
Code Issues Packages Projects Releases Wiki Activity

performance(ocl): flag to not dcFlush on no event

Browse Source 
If waitForBarrier is not passed outEvent then do
dcFlush on the next synchronize call.

Related-To: NEO-8147

Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
This commit is contained in:
Dominik Dabek
2023-07-13 09:26:41 +00:00
committed by Compute-Runtime-Automation
parent faa8907344
commit 622a3ed89c
17 changed files with 262 additions and 50 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
opencl/source/helpers/task_information.cpp
View File

@@ -84,7 +84,9 @@ CompletionStamp &CommandMapUnmap::submit(TaskCountType taskLevel, bool terminate
false, // hasStallingCmds
false, // hasRelaxedOrderingDependencies
false, // stateCacheInvalidation
commandQueue.isStallingCommandsOnNextFlushRequired()); // isStallingCommandsOnNextFlushRequired
commandQueue.isStallingCommandsOnNextFlushRequired(), // isStallingCommandsOnNextFlushRequired
commandQueue.isDcFlushRequiredOnStallingCommandsOnNextFlush() // isDcFlushRequiredOnStallingCommandsOnNextFlush
);
DEBUG_BREAK_IF(taskLevel >= CompletionStamp::notReady);
@@ -179,6 +181,7 @@ CompletionStamp &CommandComputeKernel::submit(TaskCountType taskLevel, bool term
*currentTimestampPacketNodes, csrDeps,
commandQueue.getGpgpuCommandStreamReceiver(), *bcsCsrForAuxTranslation);
commandQueue.setStallingCommandsOnNextFlush(true);
commandQueue.setDcFlushRequiredOnStallingCommandsOnNextFlush(true);
}
if (timestampPacketDependencies && commandQueue.isOOQEnabled()) {
@@ -221,8 +224,9 @@ CompletionStamp &CommandComputeKernel::submit(TaskCountType taskLevel, bool term
false, // hasStallingCmds
false, // hasRelaxedOrderingDependencies
false, // stateCacheInvalidation
commandQueue.isStallingCommandsOnNextFlushRequired()); // isStallingCommandsOnNextFlushRequired
commandQueue.isStallingCommandsOnNextFlushRequired(), // isStallingCommandsOnNextFlushRequired
commandQueue.isDcFlushRequiredOnStallingCommandsOnNextFlush() // isDcFlushRequiredOnStallingCommandsOnNextFlush
);
if (commandQueue.getContext().getRootDeviceIndices().size() > 1) {
eventsRequest.fillCsrDependenciesForRootDevices(dispatchFlags.csrDependencies, commandStreamReceiver);
}
@@ -269,6 +273,7 @@ CompletionStamp &CommandComputeKernel::submit(TaskCountType taskLevel, bool term
if (isHandlingBarrier) {
commandQueue.clearLastBcsPackets();
commandQueue.setStallingCommandsOnNextFlush(false);
commandQueue.setDcFlushRequiredOnStallingCommandsOnNextFlush(false);
}
if (kernelOperation->blitPropertiesContainer.size() > 0) {
@@ -398,7 +403,9 @@ CompletionStamp &CommandWithoutKernel::submit(TaskCountType taskLevel, bool term
false, // hasStallingCmds
false, // hasRelaxedOrderingDependencies
false, // stateCacheInvalidation
commandQueue.isStallingCommandsOnNextFlushRequired()); // isStallingCommandsOnNextFlushRequired
commandQueue.isStallingCommandsOnNextFlushRequired(), // isStallingCommandsOnNextFlushRequired
commandQueue.isDcFlushRequiredOnStallingCommandsOnNextFlush() // isDcFlushRequiredOnStallingCommandsOnNextFlush
);
if (commandQueue.getContext().getRootDeviceIndices().size() > 1) {
eventsRequest.fillCsrDependenciesForRootDevices(dispatchFlags.csrDependencies, commandStreamReceiver);
@@ -427,6 +434,7 @@ CompletionStamp &CommandWithoutKernel::submit(TaskCountType taskLevel, bool term
if (isHandlingBarrier) {
commandQueue.clearLastBcsPackets();
commandQueue.setStallingCommandsOnNextFlush(false);
commandQueue.setDcFlushRequiredOnStallingCommandsOnNextFlush(false);
}
if (kernelOperation->blitEnqueue) {