74 lines
3.2 KiB
C++
74 lines
3.2 KiB
C++
/*
|
|
* Copyright (C) 2021 Intel Corporation
|
|
*
|
|
* SPDX-License-Identifier: MIT
|
|
*
|
|
*/
|
|
|
|
#include "shared/source/helpers/cache_flush_xehp_plus.inl"
|
|
|
|
#include "opencl/extensions/public/cl_ext_private.h"
|
|
#include "opencl/source/command_queue/command_queue_hw_base.inl"
|
|
#include "opencl/source/memory_manager/resource_surface.h"
|
|
|
|
namespace NEO {
|
|
|
|
template <typename GfxFamily>
|
|
void CommandQueueHw<GfxFamily>::runSchedulerSimulation(DeviceQueueHw<GfxFamily> &devQueueHw, Kernel &parentKernel) {
|
|
}
|
|
|
|
template <>
|
|
void CommandQueueHw<Family>::submitCacheFlush(Surface **surfaces,
|
|
size_t numSurfaces,
|
|
LinearStream *commandStream,
|
|
uint64_t postSyncAddress) {
|
|
if constexpr (Family::isUsingL3Control) {
|
|
StackVec<L3Range, 128> subranges;
|
|
for (auto surface : CreateRange(surfaces, numSurfaces)) {
|
|
auto resource = reinterpret_cast<ResourceSurface *>(surface);
|
|
auto alloc = resource->getGraphicsAllocation();
|
|
coverRangeExact(alloc->getGpuAddress(), alloc->getUnderlyingBufferSize(), subranges, resource->resourceType);
|
|
}
|
|
|
|
for (size_t subrangeNumber = 0; subrangeNumber < subranges.size(); subrangeNumber += maxFlushSubrangeCount) {
|
|
size_t rangeCount = subranges.size() <= subrangeNumber + maxFlushSubrangeCount ? subranges.size() - subrangeNumber : maxFlushSubrangeCount;
|
|
Range<L3Range> range = CreateRange(subranges.begin() + subrangeNumber, rangeCount);
|
|
uint64_t postSyncAddressToFlush = 0;
|
|
if (rangeCount < maxFlushSubrangeCount || subranges.size() - subrangeNumber - maxFlushSubrangeCount == 0) {
|
|
postSyncAddressToFlush = postSyncAddress;
|
|
}
|
|
|
|
flushGpuCache<Family>(commandStream, range, postSyncAddressToFlush, device->getHardwareInfo());
|
|
}
|
|
}
|
|
}
|
|
|
|
template <>
|
|
bool CommandQueueHw<Family>::isCacheFlushCommand(uint32_t commandType) const {
|
|
return commandType == CL_COMMAND_RESOURCE_BARRIER;
|
|
}
|
|
|
|
template <>
|
|
LinearStream &getCommandStream<Family, CL_COMMAND_RESOURCE_BARRIER>(CommandQueue &commandQueue, const CsrDependencies &csrDeps, bool reserveProfilingCmdsSpace, bool reservePerfCounterCmdsSpace, bool blitEnqueue, const MultiDispatchInfo &multiDispatchInfo, Surface **surfaces, size_t numSurfaces, bool isMarkerWithProfiling) {
|
|
size_t expectedSizeCS = 0;
|
|
bool usePostSync = false;
|
|
if (commandQueue.getGpgpuCommandStreamReceiver().peekTimestampPacketWriteEnabled()) {
|
|
expectedSizeCS += TimestampPacketHelper::getRequiredCmdStreamSize<Family>(csrDeps);
|
|
usePostSync = true;
|
|
}
|
|
|
|
if constexpr (Family::isUsingL3Control) {
|
|
StackVec<L3Range, 128> subranges;
|
|
for (auto surface : CreateRange(surfaces, numSurfaces)) {
|
|
ResourceSurface *resource = reinterpret_cast<ResourceSurface *>(surface);
|
|
auto alloc = resource->getGraphicsAllocation();
|
|
coverRangeExact(alloc->getGpuAddress(), alloc->getUnderlyingBufferSize(), subranges, resource->resourceType);
|
|
}
|
|
expectedSizeCS += getSizeNeededToFlushGpuCache<Family>(subranges, usePostSync);
|
|
}
|
|
|
|
return commandQueue.getCS(expectedSizeCS);
|
|
}
|
|
|
|
} // namespace NEO
|