mirror of
https://github.com/intel/compute-runtime.git
synced 2025-12-20 00:24:58 +08:00
1c52017ceb
Globals surface allocation via USM manager will have correct allocation type set (instead of just BUFFER) and will use cpu copy when possible. Related-To: NEO-7796 Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
75 lines
3.6 KiB
C++
75 lines
3.6 KiB
C++
/*
|
|
* Copyright (C) 2020-2023 Intel Corporation
|
|
*
|
|
* SPDX-License-Identifier: MIT
|
|
*
|
|
*/
|
|
|
|
#include "program_initialization.h"
|
|
|
|
#include "shared/source/compiler_interface/linker.h"
|
|
#include "shared/source/device/device.h"
|
|
#include "shared/source/helpers/blit_commands_helper.h"
|
|
#include "shared/source/helpers/gfx_core_helper.h"
|
|
#include "shared/source/helpers/string.h"
|
|
#include "shared/source/memory_manager/allocation_properties.h"
|
|
#include "shared/source/memory_manager/memory_manager.h"
|
|
#include "shared/source/memory_manager/unified_memory_manager.h"
|
|
#include "shared/source/program/program_info.h"
|
|
|
|
namespace NEO {
|
|
|
|
GraphicsAllocation *allocateGlobalsSurface(NEO::SVMAllocsManager *const svmAllocManager, NEO::Device &device, size_t totalSize, size_t zeroInitSize, bool constant,
|
|
LinkerInput *const linkerInput, const void *initData) {
|
|
bool globalsAreExported = false;
|
|
GraphicsAllocation *gpuAllocation = nullptr;
|
|
const auto rootDeviceIndex = device.getRootDeviceIndex();
|
|
const auto deviceBitfield = device.getDeviceBitfield();
|
|
|
|
if (linkerInput != nullptr) {
|
|
globalsAreExported = constant ? linkerInput->getTraits().exportsGlobalConstants : linkerInput->getTraits().exportsGlobalVariables;
|
|
}
|
|
const auto allocationType = constant ? AllocationType::CONSTANT_SURFACE : AllocationType::GLOBAL_SURFACE;
|
|
if (globalsAreExported && (svmAllocManager != nullptr)) {
|
|
RootDeviceIndicesContainer rootDeviceIndices;
|
|
rootDeviceIndices.push_back(rootDeviceIndex);
|
|
std::map<uint32_t, DeviceBitfield> subDeviceBitfields;
|
|
subDeviceBitfields.insert({rootDeviceIndex, deviceBitfield});
|
|
NEO::SVMAllocsManager::UnifiedMemoryProperties unifiedMemoryProperties(InternalMemoryType::DEVICE_UNIFIED_MEMORY, rootDeviceIndices, subDeviceBitfields);
|
|
unifiedMemoryProperties.device = &device;
|
|
unifiedMemoryProperties.requestedAllocationType = allocationType;
|
|
auto ptr = svmAllocManager->createUnifiedMemoryAllocation(totalSize, unifiedMemoryProperties);
|
|
DEBUG_BREAK_IF(ptr == nullptr);
|
|
if (ptr == nullptr) {
|
|
return nullptr;
|
|
}
|
|
auto usmAlloc = svmAllocManager->getSVMAlloc(ptr);
|
|
UNRECOVERABLE_IF(usmAlloc == nullptr);
|
|
gpuAllocation = usmAlloc->gpuAllocations.getGraphicsAllocation(rootDeviceIndex);
|
|
} else {
|
|
gpuAllocation = device.getMemoryManager()->allocateGraphicsMemoryWithProperties({rootDeviceIndex,
|
|
true, // allocateMemory
|
|
totalSize, allocationType,
|
|
false, // isMultiStorageAllocation
|
|
deviceBitfield});
|
|
}
|
|
|
|
if (!gpuAllocation) {
|
|
return nullptr;
|
|
}
|
|
|
|
auto &rootDeviceEnvironment = device.getRootDeviceEnvironment();
|
|
auto &productHelper = device.getProductHelper();
|
|
|
|
bool isOnlyBssData = (totalSize == zeroInitSize);
|
|
if (false == isOnlyBssData) {
|
|
auto initSize = totalSize - zeroInitSize;
|
|
auto success = MemoryTransferHelper::transferMemoryToAllocation(productHelper.isBlitCopyRequiredForLocalMemory(rootDeviceEnvironment, *gpuAllocation),
|
|
device, gpuAllocation, 0, initData, initSize);
|
|
UNRECOVERABLE_IF(!success);
|
|
}
|
|
return gpuAllocation;
|
|
}
|
|
|
|
} // namespace NEO
|