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 new debug key to control dependency resolution.

Browse Source 
Currently only supported scenario is one in order queue.
Instead of resolving dependencies via semaphores, do this with pipe controls.

Signed-off-by: Michal Mrozek <michal.mrozek@intel.com>
This commit is contained in:
Michal Mrozek
2021-12-09 16:18:18 +00:00
committed by Compute-Runtime-Automation
parent 34d9d9b0d3
commit ebb16c8b74
7 changed files with 145 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

100
opencl/test/unit_test/helpers/timestamp_packet_1_tests.cpp
View File

@ -1318,6 +1318,106 @@ HWTEST_F(TimestampPacketTests, givenTimestampPacketWriteEnabledOnDifferentCSRsFr
EXPECT_EQ(3u, semaphoresFound); // total number of semaphores found in cmdList
}
HWTEST_F(TimestampPacketTests, givenTimestampPacketWriteEnabledAndDependenciesResolvedViaPipeControlsIfPreviousOperationIsBlitThenStillProgramSemaphores) {
DebugManagerStateRestore restorer;
DebugManager.flags.ResolveDependenciesViaPipeControls.set(1);
using MI_SEMAPHORE_WAIT = typename FamilyType::MI_SEMAPHORE_WAIT;
using WALKER = typename FamilyType::WALKER_TYPE;
device->getUltCommandStreamReceiver<FamilyType>().timestampPacketWriteEnabled = true;
MockMultiDispatchInfo multiDispatchInfo(device.get(), std::vector<Kernel *>({kernel->mockKernel}));
auto &cmdStream = mockCmdQ->getCS(0);
mockCmdQ->updateLatestSentEnqueueType(NEO::EnqueueProperties::Operation::Blit);
const cl_uint eventsOnWaitlist = 1;
MockTimestampPacketContainer timestamp(*device->getGpgpuCommandStreamReceiver().getTimestampPacketAllocator(), 1);
Event event(mockCmdQ, 0, 0, 0);
event.addTimestampPacketNodes(timestamp);
cl_event waitlist[] = {&event};
EventsRequest eventsRequest(eventsOnWaitlist, waitlist, nullptr);
CsrDependencies csrDeps;
eventsRequest.fillCsrDependenciesForTimestampPacketContainer(csrDeps, mockCmdQ->getGpgpuCommandStreamReceiver(), CsrDependencies::DependenciesType::OnCsr);
HardwareInterface<FamilyType>::dispatchWalker(
*mockCmdQ,
multiDispatchInfo,
csrDeps,
nullptr,
nullptr,
nullptr,
nullptr,
nullptr,
CL_COMMAND_NDRANGE_KERNEL);
HardwareParse hwParser;
hwParser.parseCommands<FamilyType>(cmdStream, 0);
uint32_t semaphoresFound = 0;
for (auto it = hwParser.cmdList.begin(); it != hwParser.cmdList.end(); it++) {
auto semaphoreCmd = genCmdCast<MI_SEMAPHORE_WAIT *>(*it);
if (semaphoreCmd) {
semaphoresFound++;
}
}
EXPECT_EQ(1u, semaphoresFound); // total number of semaphores found in cmdList
}
HWTEST_F(TimestampPacketTests, givenTimestampPacketWriteEnabledAndDependenciesResolvedViaPipeControlsIfPreviousOperationIsGPUKernelThenDoNotProgramSemaphores) {
DebugManagerStateRestore restorer;
DebugManager.flags.ResolveDependenciesViaPipeControls.set(1);
using MI_SEMAPHORE_WAIT = typename FamilyType::MI_SEMAPHORE_WAIT;
using WALKER = typename FamilyType::WALKER_TYPE;
device->getUltCommandStreamReceiver<FamilyType>().timestampPacketWriteEnabled = true;
MockMultiDispatchInfo multiDispatchInfo(device.get(), std::vector<Kernel *>({kernel->mockKernel}));
auto &cmdStream = mockCmdQ->getCS(0);
mockCmdQ->updateLatestSentEnqueueType(NEO::EnqueueProperties::Operation::GpuKernel);
const cl_uint eventsOnWaitlist = 1;
MockTimestampPacketContainer timestamp(*device->getGpgpuCommandStreamReceiver().getTimestampPacketAllocator(), 1);
Event event(mockCmdQ, 0, 0, 0);
event.addTimestampPacketNodes(timestamp);
cl_event waitlist[] = {&event};
EventsRequest eventsRequest(eventsOnWaitlist, waitlist, nullptr);
CsrDependencies csrDeps;
eventsRequest.fillCsrDependenciesForTimestampPacketContainer(csrDeps, mockCmdQ->getGpgpuCommandStreamReceiver(), CsrDependencies::DependenciesType::OnCsr);
HardwareInterface<FamilyType>::dispatchWalker(
*mockCmdQ,
multiDispatchInfo,
csrDeps,
nullptr,
nullptr,
nullptr,
nullptr,
nullptr,
CL_COMMAND_NDRANGE_KERNEL);
HardwareParse hwParser;
hwParser.parseCommands<FamilyType>(cmdStream, 0);
uint32_t semaphoresFound = 0;
for (auto it = hwParser.cmdList.begin(); it != hwParser.cmdList.end(); it++) {
auto semaphoreCmd = genCmdCast<MI_SEMAPHORE_WAIT *>(*it);
if (semaphoreCmd) {
semaphoresFound++;
}
}
EXPECT_EQ(0u, semaphoresFound);
}
HWTEST_F(TimestampPacketTests, givenAlreadyAssignedNodeWhenEnqueueingNonBlockedThenMakeItResident) {
auto mockTagAllocator = new MockTagAllocator<>(device->getRootDeviceIndex(), executionEnvironment->memoryManager.get(), 1);